Vaccinia viruses: vaccines against smallpox and vectors against infectious diseases and tumors

Abstract

Less than 200 years after its introduction, widespread use of vaccinia virus (VACV) as a smallpox vaccine has eradicated variola virus. Along with the remarkable success of the vaccination program, frequent and sometimes severe adverse reactions to VACV were encountered. After eradication, VACV has been reserved for select populations who might be at significant risk for orthopoxvirus infections. Events over the past decade have renewed concerns over the potential use of variola virus as a biological weapon. Accordingly, interest in VACV and attenuated derivatives has increased, both as vaccines against smallpox and as vectors for other vaccines. This article will focus on new developments in the field of orthopoxvirus immunization and will highlight recent advances in the use of vaccinia viruses as vectors for infectious diseases and malignancies.

Smallpox is a contagious and highly morbid disease with approximately 30% mortality [1]. It is notable for a striking and characteristic exanthem that scars survivors [2–5]. The causative agent, variola virus (VARV), is a member of the orthopoxvirus family [6], of which the best characterized is the smallpox vaccine virus, vaccinia virus (VACV). Although VARV was eradicated in 1978 following an intensive worldwide vaccination effort [1,2,5] and is now maintained in only two secure laboratories [7], recent events have renewed concerns that VARV from an undeclared stockpile might be used as a biological weapon [3,8,9] or could be inadvertently released from the two reference laboratories. Over 100 million Americans have been born since universal immunization ceased [8], and immunity has waned in an estimated 150 million Americans who were vaccinated at earlier times. This means that a smallpox outbreak in the USA could lead to catastrophic consequences [10–14], although rapid diagnosis of cases and aggressive public health interventions could help contain an epidemic [15,16]. The smallpox vaccine used in the USA until 2008 was live calf lymphderived Dryvax^® (Wyeth; derived from the New York City Board of Health [NYCBH] strain of VACV), and while highly effective in preventing smallpox [4,5,8], it has a significant incidence of adverse effects, which have prompted efforts to develop safer smallpox vaccines [17,18].

Historically, VACV vaccination was associated with a large number of side effects including encephalitis, encephalopathy, vaccinia conjunctivitis, fetal vaccinia, inadvertent spread of VACV and even death [19–25]. The recent limited civilian vaccination program for ‘first-responders,’ as well as the larger military vaccination program, provide some insights into the frequency of adverse events (AEs) in highly selected populations. While the overall frequency of significant side effects appeared to be quite low [26–28], there were rare AEs [29], most notably myopericarditis [30,31], which had been underappreciated in historical studies [32]. Dermatologic complications following smallpox vaccination via scarification have long been recognized and include potentially life-threatening reactions such as progressive vaccinia or vaccinia necrosum [33,34], eczema vaccinatum [35,36] and disseminated vaccinia [37,38]. More commonly seen, however, are less serious reactions such as satellite lesions, viral cellulitis and a variety of exanthems, including vaccinia folliculitis [39,40], which have been estimated to occur at an overall frequency of approximately one per 3700 vaccinees [201]. These numerous and significant AEs have led to an extensive list of contraindications to vaccination, and therefore a substantial fraction of the public cannot receive VACV. Indeed, some have estimated that up to 25% of the US population have contraindications to the current smallpox vaccine [41]. These contraindications include immunosuppression (i.e., organ transplantation or immunosuppressive medications), pregnancy, breast-feeding, HIV infection, atopic dermatitis or eczema, household contact with any person who cannot receive VACV and all persons under 18 years of age [42,43].

As smallpox is both a cutaneous and systemic disease [44], an important correlate of protection that has been examined in both preclinical studies and clinical trials of novel smallpox vaccines has been the effect of vaccination on skin lesions induced by challenge viruses. Traditional vaccination via scarification results in the development of a lesion at the inoculation site that progresses from a papule to a large deep-seated pustule that ulcerates and heals over in approximately 3 weeks [1,3,45]. This ‘Jennerian pustule’ is commonly called a ‘take reaction’ and has been considered to confer immunity from clinical smallpox for at least 10 years following scarification [1,3,8]. Upon revaccination, individuals who had been recently vaccinated exhibit no take at all or a modified or blunted take that is smaller and heals faster than a primary vaccination, suggesting that recall immune responses control VACV replication in the skin [1,3,8], although the relative contributions of humoral and cellular immune responses are not known. Second-generation smallpox vaccines containing tissue culture-derived VACV (including the recently approved ACAM2000 vaccine, see later and Table 1) are able to induce take responses identical to Dryvax and other calf lymph-derived vaccines [46–48], which has been the historical correlate with effectiveness against VARV [1]. However, attenuated vaccines, such as modified vaccinia Ankara (MVA) and NYVAC (see later and Table 2), do not induce skin lesions, and hence cannot be used to provide evidence of either vaccine efficacy [49] or clinical effectiveness. However, assessment of take following VACV challenge can provide important information as to the nature of immune responses induced by MVA or NYVAC.

Table 1.

Second-generation vaccinia-based smallpox vaccines.

Vaccine	Subjects Enrolled 1	Findings	AEs/SAEs	Reference
ACAM1000 (Acambis)	60 VACV-naïve healthy volunteers	100% take and seroconversion rate	No SAEs; no difference in local reactogenicity between ACAM1000 & Dryvax	[49,57]
	70 VACV-naïve healthy volunteers	100% take rate, 94% seroconversion rate	No SAEs; local reactogenicity common; other AEs mild
ACAM2000	100 VACV-naïve healthy volunteers	99% take rate, 96% seroconversion rate	Local reactogenicity common; 1 SAE (new-onset seizure)	[47,57,59,60]
	90 VACV-naïve healthy volunteers	100% take rate, 96% seroconversion in ACAM2000 group	Local reactogenicity common; 2 SAEs (transient EKG changes in an ACAM2000 vaccinee, transient neutropenia in ACAM1000 vaccinee)
	353 VACV-naïve healthy volunteers	100% take rate at full dose, <90% take rate when diluted; dose-response trend for antibodies	Local reactogenicity slightly less with ACAM2000 than Dryvax; 1 case of myocarditis in ACAM2000 group
	357 VACV-experienced healthy volunteers	88% take rate at full dose, greatly decreased take rate when diluted; dose-response trend for antibodies	Not reported
	1162 VACV-naïve healthy volunteers	96% take rate, not inferior to Dryvax	No differences in AEs between ACAM2000 and Dryvax; 5 cases of myopericarditis with ACAM2000 and 3 cases with Dryvax
	1819 VACV-experienced healthy volunteers	84% take rate, inferior to Dryvax; GMT of NAb not inferior to Dryvax	Local reactogenicity common; no differences in SAEs between groups; 1 case of generalized VACV in Dryvax group
CCSV (DynPort)	91 VACV-naïve healthy volunteers	IM route least immunogenic; ID route more immunogenic if pock lesion develops	Reactogenicity symptoms lower with ID or IM route and no serious AEs	[48,67]
	150 VACV-naïve healthy volunteers 100 VACV-experienced healthy volunteers 100 VACV-naïve healthy volunteers (dilution subgroup)	99.7% developed take responses; immunogenic even at 1:50 dilution	No SAEs; no difference in local or systemic reactogenicity between CCSV & Dryvax; more systemic symptoms in naïve subjects
CJ-50300	24 VACV-naïve healthy volunteers	100% take rate and seroconversion, 100% response rate by ELISPOT	Not reported	[69,70]
	123 VACV-naïve healthy volunteers	99% take rate and seroconversion with one dose; 91% response rate by ELISPOT	Local and systemic reactogenicity similar between high dose and low dose groups; no cases of myopericarditis; one possible case of generalized VACV

Abbreviations: VACV – vaccinia virus; AEs – adverse events; SAEs – serious adverse events; ID – intradermal; SC – subcutaneous; IM – intramuscular; EKG – electrocardiogram; GMT – geometric mean titer

¹Subjects enrolled are enumerated in the tables, but not all subjects completed the respective studies or could be included in the final analyses.

Table 2.

Third-generation smallpox vaccines.

Vaccine	Subjects Enrolled	Findings	AEs/SAEs	Reference
IMVAMUNE (MVA-BN)	68 VACV-naïve and 18 VACV-experienced healthy volunteers	Dose-dependent seroconversion rates and NAb titers in VACV-naïve subjects, anamnestic responses in VACV-experienced	Mild-moderate local reactogenicity common; rare systemic reactogenicity	[87,89,91]
	90 healthy, VACV-naïve volunteers	MVA attenuates take following Dryvax and primes for higher antibody titers after Dryvax challenge	Mild-moderate local reactogenicity common; no SAEs related to vaccination
	165 VACV-naive volunteers	All subjects seroconverted after second dose; higher NAb titers with higher dose	Mild-moderate local reactogenicity common; no SAEs, no vaccine-related AEs
ACAM3000 MVA (Acambis/Sanofi)	72 healthy, VACV-naïve volunteers	Dose-dependent antibody titers; ID route immunogenic at 1/10th dose of IM or SC; MVA attenuates take and reduces viral shedding following Dryvax challenge	Local reactogenicity more pronounced with ID or SC injection than IM; no SAEs	[74,90]
TBC-MVA (Therion)	76 VACV-naïve and 75 VACV-experienced healthy volunteers	MVA attenuates take and reduces viral shedding following Dryvax challenge; MVA primes for increased CD8+ T cell responses after Dryvax	Local reactogenicity common; 3 exanthems following Dryvax challenge in placebo recipients	[88]
LC16m8	1529 VACV-naïve and 1692 VACV-experienced military personnel	94.4% vs 86.6% take responses and 90.2% vs 60% seroconversion respectively	No SAEs; one case of allergic dermatitis, one case of erythema multiforme	[72]
CVI-78	594 children with eczema 1009 patients (mostly 1– 5 years old) with eczema 95 healthy children	100% seroconversion, 87% take rate when given via scarification	Less reactogenicity than traditionally seen with Dryvax; no cases of eczema vaccinatum	[77,78,80,81]

Abbreviations: VACV – vaccinia virus; AEs – adverse events; SAEs – serious adverse events; ID – intradermal; SC – subcutaneous; IM - intramuscular

In the modern era, immunologic assessments of novel VACV-derived smallpox vaccines have included binding antibody titers by ELISA, neutralizing antibody (NAb) assays, and T-cell induction via IFN-γ enzyme-linked immunosorbent spot (ELISpot) or intracellular cytokine staining. NAb induction following vaccination with a successful take response has long been described in the literature [50,51] and the correlation between a successful take reaction and NAb induction is best seen in naive vaccinees. In a clinical trial using serial dilutions of Dryvax, Belshe et al. found that all previously naive subjects scarificated with either undiluted, 1:5, or 1:10 dilutions of Dryvax who developed a take also seroconverted with geometric mean titers (GMT) exceeding 1000 in all groups [52]. Furthermore, Belshe et al. noted a highly significant correlation between lesion size and NAb titers [52]. In a study comparing serial dilutions of Dryvax to another NYCBH-based calf lymph vaccine (Sanofi-Pasteur smallpox vaccine; SPSV) in vaccinia-naive subjects, Couch et al. noted that 97.5% of subjects with a take also had increased NAb titers and they reported that the correlation between take rate and NAb response was highly significant [53]. These studies affirm that take response correlates well with NAb induction in naive vaccinees and provide an important background for interpreting data from clinical trials with novel VACV-derived vaccines.

ACAM2000: a second-generation smallpox vaccine

Even prior to the biocriminal anthrax attacks in the USA in the fall of 2001, it was recognized that existing stockpiles of smallpox vaccines were inadequate for many reasons. Only 15.4 million doses of Dryvax were estimated to exist in the USA in 1999, and only enough vaccinia immunoglobulin was available to treat 675 cases of severe adverse reactions [54]. These doses of Dryvax had been prepared in the late 1970s by the traditional calf-lymph method, and thus the lyophilized product contained bovine impurities. Also, the diluent contained several antibiotics, since sterility of the vaccine was in question. In addition to questions concerning the safety of calf-lymph vaccines, means were not available to quickly upscale production in case of a smallpox release. The CDC therefore contracted with OraVax (later Acambis, now Sanofi Pasteur) to produce a live, VACV smallpox vaccine grown in tissue culture and produced with modern good manufacturing practices [55].

To develop the seed strain for the cell-culture-derived VACV vaccine, 30 vials of Dryvax from three different lots were pooled, and serial plaque-purification on MRC-5 cells was used to isolate clonal derivatives [48]. Six clones were assayed for pock formation on rabbit skin and neurovirulence in mice; clone 2 appeared to yield similar skin lesions but less neurovirulence than Dryvax and was therefore selected as the parental strain for ACAM1000 [48]. Based on encouraging preclinical safety and efficacy results [48], two clinical trials (Table 1) were conducted. In the first trial, 60 healthy VACV-naive volunteers received either ACAM1000 or Dryvax via traditional scarification in a double-blind randomized protocol [48]. All 30 subjects scarified with ACAM1000 had a major take reaction, while 29 out of 30 scarified with Dryvax had a major take, and identical proportions had a greater than four-fold increase in antibody titer compared with baseline [48]. Local reactogenicity was very common, but there was no difference in AEs between the groups. Although GMT of VACV-specific NAb were slightly lower in the ACAM1000 group, the difference was not statistically significant [48]. Cytotoxic T lymphocyte responses measured by lytic activity and cellular immune responses measured by IFN-γ ELISpot did not differ between the groups following vaccination, but ACAM1000 appeared to induce more frequent and quantitatively higher T-cell proliferative responses [48]. In the second trial, 70 healthy VACV-naive adults received ACAM1000 via scarification in an open-label trial [56]. Again, 100% of subjects had a major take response to vaccination and 94% seroconverted [56]. While no serious adverse events (SAEs) were noted, local reactogenicity was again quite common, and other AEs were mild [56].

As the contract for the Strategic National Stockpile was for over 200 million doses, Acambis collaborated with Baxter Biosciences and used an extant validated large-scale bioreactor technology for further development [56]. Since this technology uses Vero cells rather than MRC-5 cells, ACAM1000 was used as the seed virus and subsequent passages were named ACAM2000 [56]. ACAM2000 underwent extensive testing to exclude adventitious agents, and preclinical testing demonstrated that ACAM2000 was similar to ACAM1000 with respect to pock development in rabbits, and that both strains were less neurovirulent than the parental Dryvax vaccine [56]. ACAM2000 was immunogenic in mice and protected against virulent intranasal challenge with efficacy equal to ACAM1000 and Dryvax [56]. ACAM2000 was also immunogenic in cynomolgous macaques and had comparable efficacy to Dryvax in protection against intravenous monkeypox-virus challenge [57].

ACAM2000 was first tested in an open-label study in 100 healthy VACV-naive volunteers screened for VACV vaccination contraindications [56]. Nearly all (99%) had a successful take response, and 96% seroconverted [56]. Local reactogenicity was nearly universal (as had been seen with ACAM1000 and Dryvax), and a single SAE (new onset seizure 8 days after vaccination) was noted [56]. Other AEs were mild to moderate, and one fever higher than 38.9°C was noted in a subject with concurrent streptococcal pharyngitis [56].

A double-blind study then compared ACAM2000 with ACAM1000 and Dryvax in 30 healthy VACV-naive volunteers randomized into each group [58]. All subjects experienced a major take reaction at the scarification site, but Dryvax recipients had larger areas of erythema than either ACAM1000 or ACAM2000 vaccinees [58]. Local reactogenicity was again very common and did not differ among groups, nor did rates of AEs [58]. Two SAEs were noted and graded as ‘probably related’ to study product: transient asymptomatic electrocardiography (EKG) changes in an ACAM2000 vaccinee and transient neutropenia in an ACAM1000 recipient [58]. Viral shedding from the scarification site did not differ among groups [58]. GMT of VACV NAb were slightly lower in the ACAM2000 group, but this trend was not statistically significant [58]. Seroconversion rates were identical in the Dryvax and ACAM2000 groups (96.7%) and not significantly lower in the ACAM1000 group (90%) [58]. No differences were seen among the groups by in vitro assays of cell-mediated immune responses [58].

A larger study was then performed to compare Dryvax with serial dilutions of ACAM2000 in healthy VACV-naive volunteers [59]. A total of 353 subjects were randomized into five groups: Dryvax at 1.6 × 10⁸ PFU/ml, ACAM2000 at 6.8 × 10⁷ PFU/ml, ACAM2000 at 1.4 × 10⁷ PFU/ml (1:5 dilution), ACAM2000 at 6.8 × 10⁶ PFU/ml (1:10 dilution) and ACAM2000 at 3.4 × 10⁶ PFU/ml (1:20 dilution) [59]. The target dose of ACAM2000 had been 1 × 10⁸ to allow direct comparison with Dryvax but back-titration revealed that, following reconstitution, the lot of ACAM2000 used had less than half the titer of the Dryvax lot [59]. All subjects who received either Dryvax or full-dose ACAM2000 had take responses. However, lower frequencies of major takes were seen in the dilution cohorts with rates of 86, 80 and 59%, respectively [59]. As the authors had established a 90% response rate for an efficacy threshold [59], diluted ACAM2000 did not meet this end point. Antibody responses showed a similar dose-response trend, and the full-dose ACAM2000 and Dryvax cohorts had no statistically significant differences in seroconversion rates or NAb titers [59]. Based on the dose-response data, the authors calculated that an ACAM2000 dose of 2.71 × 10⁷ PFU/ml would be required for a 90% seroconversion rate with 95% confidence [59]. Although no differences in take rates were reported among the four sites in this clinical trial [59], the decades-old technique of scarification is somewhat unfamiliar in the modern era and if there had been minor technical issues, this may have contributed to the lower take rates seen with diluted ACAM2000. Local reactogenicity was somewhat reduced in the dilution groups of ACAM2000, but no statistically significant differences in AEs were seen between the full-dose ACAM2000 and Dryvax groups [59]. One subject in the highest dose ACAM2000 group was diagnosed with myopericarditis 10 days following vaccination, which resolved within 36 h [59].

A second study was performed contemporaneously that compared the effect of serial dilutions of ACAM2000 at the same doses in volunteers who had previously been vaccinated against smallpox [46]. All subjects who were scarified with Dryvax had a successful take response, but the rates of successful takes were less in the ACAM2000 cohorts with rates of 88, 51, 40, and 27% in the dilution titers [46]. Similar dose-response trends were seen with NAb titers with a GMT of 447 for Dryvax and GMTs of 256, 115, 84 and 59 in the respective ACAM2000 dilution groups [46]. Since the investigators had established a 90% response rate for an efficacy threshold [46], even undiluted ACAM2000 did not meet this end point in previously vaccinated subjects. In theory, the adventitial agents present in calf lymph could act as xenogenic adjuvants, which might explain the enhanced take rates seen with Dryvax. However, another tissue culture grown vaccine (CCSV, see later) has been tested in humans and a clinical trial suggested it could be diluted even further than ACAM2000 without a decrement in its capacity to elicit a take [47]. This, as well as the differences between potential seed viruses described in the development of ACAM1000 [48], suggests that even limited serial passage in tissue culture selects for clones (in contrast to the nonclonal ‘swarm’ of viruses in Dryvax and other traditional smallpox vaccines) that display differences in biological behavior.

Two Phase III studies to meet US FDA requirements for approval of ACAM2000 were conducted. Both were randomized, double-blind comparisons between ACAM2000 and Dryvax, with a randomization ratio of 3:1 for ACAM2000:Dryvax. One trial (H400-009) was conducted in VACV-naive subjects while the other (H400-012) was done in subjects who received VACV previously [46]. In H400–009, 1162 VACV-naive subjects were enrolled, and 96% of those vaccinated with ACAM2000 had a successful take, compared with 99% in the Dryvax group [46]. ACAM2000 was therefore found to be noninferior to Dryvax with respect to induction of a major take following scarification. However, GMT of VACV NAb were lower in the ACAM2000 group and did not meet the noninferiority criterion compared with Dryvax [202]. Of the 20 evaluable subjects who did not have a successful take following ACAM2000, all failed to seroconvert with NAb titers ≤20 [202]. Overall, there were no differences in AEs between the two groups, but there were eight cases of myopericarditis, five in the ACAM2000 group and three in the Dryvax group [46]. Due to the incidence of myopericarditis, this trial was stopped before the planned enrollment of 2720 subjects [46].

The H400–012 Phase III trial in 1819 VACV-experienced subjects had the same randomization ratio of 3:1 for ACAM2000:Dryvax recipients as did H400–009 [46]. As it recruited only subjects who had been previously vaccinated, the mean age was accordingly older (49 vs 23 years of age) in H400–009 [46]. Subjects scarified with ACAM2000 were less likely to have a successful take than those vaccinated with Dryvax (84 vs 98%) [46] and since the investigators had established a 90% response rate for an efficacy threshold, ACAM2000 did not meet the noninferiority end point in previously vaccinated subjects. In both groups, subjects had an anamnestic NAb response after scarification (8.6-fold increase in titer in ACAM2000 recipients, 15.9-fold increase in Dryvax recipients), but despite the lower GMT in ACAM2000 recipients, the results were found to be noninferior compared with Dryvax [46]. Take rates in ACAM2000 recipients were inversely correlated with baseline VACV NAb titers, as were the NAb responses in both ACAM2000 and Dryvax recipients, but take rates following Dryvax were not affected by baseline NAb titers [202]. Slightly fewer subjects in the ACAM2000 group experienced AEs than with Dryvax [46], but most AEs in both groups consisted of local reactogenicity symptoms. No differences in the incidence of SAEs were seen, but one subject in the Dryvax group had generalized vaccinia [46]. No cases of myopericarditis were noted, but chest pain was reported in three subjects (one Dryvax subject, two ACAM2000 subjects), and one case of atrial fibrillation (in an ACAM2000 subject) was noted [46]. Due to the high incidence of myopericarditis in the companion trial, this trial was stopped before the planned enrollment of 2720 subjects [46].

Based on the results of these trials, on 31 August 2007, the FDA approved ACAM2000 by scarification as a vaccine for people at high risk of exposure to smallpox [46]. Given the high frequency of myopericarditis (one in 175 for vaccinia-naive persons), ACAM2000 is restricted to the same population as Dryvax had been. The risk of myopericarditis in the ACAM2000 versus Dryvax trials did not differ between the two vaccine products, but the incidence was considerably higher than that reported from the military or civilian vaccination programs [28], suggesting that passive surveillance may underestimate the true incidence [46]. Due to the restricted inclusion/exclusion criteria, the risk of other vaccinia complications, such as eczema vaccinatum, vaccinia necrosum and vaccinia encephalitis, from ACAM2000 is not known, but may be similar to Dryvax as both contain live VACV [56,60]. A recent report from the military vaccination program described a case of generalized vaccinia and seven cases of benign exanthems in ACAM2000 vaccinees [61]. Inadvertent secondary transmission of ACAM2000 has not yet been reported, but surveillance for such cases should be maintained. To better understand the risks of ACAM2000 vaccination, a Phase IV study is being conducted under the auspices of the military smallpox vaccination program and aims to enroll 15,000 ACAM2000 recipients [46]. In addition to a pregnancy registry to track inadvertent primary and secondary exposures to ACAM2000, Sanofi Pasteur has established a myopericarditis registry and will perform enhanced surveillance for myopericarditis [46].

There are a number of unanswered questions regarding ACAM2000 (reviewed in [46]). Safety and efficacy in the pediatric age group has not been studied. Efficacy against variola infection has clearly not been demonstrated, but small studies from the pre-eradication era suggest that a VACV NAb titer >1:32 [62] or a VARV titer >1:20 [63] may correlate with protection. Thus, protection is assumed in ACAM2000 recipients with a successful take because of the correlation with seroconversion. Another issue is the decreased take rate with ACAM2000 scarification in previously vaccinated persons [46], which raises the concern that ACAM2000 may be less effective as a booster vaccine in case of a threat of smallpox exposure. However, a successful anamnestic rise in NAb titers following vaccination with ACAM2000 was seen in revaccinees and the GMT elicited by booster vaccination exceeded the GMT in primary vaccinees [202], which should temper this concern. Finally, in contrast with Dryvax [53,64], ACAM2000 appears to have a steep drop in efficacy after dilution, which is particularly notable in VACV-experienced persons [46]. This suggests that the Strategic National Stockpile will not be able to recommend dilution of ACAM2000 to deliver additional doses, and also suggests that careful monitoring will be necessary to ensure the potency of the doses amassed.

Cell-cultured smallpox vaccine

A second cell culture-derived VACV vaccine has been used in a Phase I clinical trial in both vaccinia-naive and remotely vaccinated individuals [47]. This vaccine, called cell-cultured smallpox vaccine (CCSV), was derived from a master viral seed lot used in the past to generate the Connaught vaccine which was withdrawn in the late 1970s [65]. Like Dryvax, the Connaught product traces its origin to the NYCBH strain of VACV, and similar to the other second-generation smallpox vaccines, CCSV was plaque-purified three times in MRC-5 cells [47,65]. Early studies with CCSV tested its safety and immunogenicity via alternative routes of vaccination in comparison with Dryvax. It was noted that with increasing doses of CCSV via the subcutaneous (sc.) route, more subjects developed cutaneous pox lesions and these subjects tended to have earlier and stronger immune responses [65]. A subsequent study compared intramuscular (im.) and intradermal (id.) CCSV with Dryvax administered by traditional scarification, at 1 × 10⁵ PFU per dose in 91 VACV-naive subjects randomized evenly among the three groups. Eight subjects were later excluded from the analysis since serologic evidence of remote vaccination was found. The study revealed that scarified subjects, as well as those who had received id. CCSV and developed a take, had higher cell-mediated and humoral immune responses than those vaccinated via the im. or id. routes without development of a take [65]. Local reactogenicity symptoms were highest in the scarification group and lowest in the im. group, and no serious AEs were associated with the vaccine [65].

As part of the efforts to replace Dryvax, CCSV was then studied in a direct, double-blind comparison with Dryvax in both VACV-naive and VACV-experienced subjects via scarification [47]. Additional vaccinia-naive subjects were recruited to receive CCSV via serial dilution (undiluted at 2.5 × 10⁵ PFU per dose and 1:5, 1:10, 1:25 and 1:50 dilutions) in a single-blind (subject only) fashion. No SAEs were noted, and mild to moderate local reactogenicity was similar between the groups, but systemic reactogenicity was somewhat more common in VACV-naive subjects [47]. All subjects who received undiluted CCSV or Dryvax had takes regardless of whether or not they were VACV-naive. Satellite lesions were reported in a small fraction of VACV-naive individuals and were seen more often in CCSV recipients [47]. NAb titers peaked earlier and were higher in VACV-experienced versus VACV-naive subjects, which was consistent with an anamnestic response, but titers did not differ between CCSV and Dryvax in each cohort [47]. Cell-mediated immune responses were similar between the two vaccines [47], but only a subset of vaccinees were assayed. In the dilution groups, local and systemic reactogenicity were similar to the undiluted cohorts, and only one subject (in the 1:25 group) did not develop a take reaction and failed to seroconvert [47]. NAb titers did not show a clear dose-response trend, since titers were similar in all groups at all time points assayed [47]. CCSV was not selected for further development [66].

CJ-50300

A third cell-culture derived VACV vaccine based on NYCBH has been studied in humans, called CJ-50300 [67,68]. The vaccine was derived by passaging a strain of NYCBH (ATCC VR-118) in MRC-5 cells without plaque purification [68]. Preclinical studies found that CJ-50300 was similar to the European Lancy-Vaxina VACV vaccine with respect to pock formation, immunogenicity and neurovirulence in small animal models [68]. Healthy VACV-naive volunteers were recruited and screened for contraindications to live VACV vaccination for two clinical trials with CJ-50300. The first trial was a double-blind, placebo-controlled, randomized trial of 24 subjects with a 3:1 randomization ratio of vaccine:placebo [68] at a dose of 1 × 10⁸ PFU/ml. Although the route of vaccination was not specified, it appears to have been epidermal scarification. All 18 vaccine recipients had a take response, all shed virus from the site, and five vaccinees had moderate to severe reactogenicity after vaccination [68]. All 18 seroconverted by day 28 (defined as a fourfold increase in NAb titers from baseline) and all had positive IFN-γ ELISpot responses by day 14 following vaccination [68].

The second trial was a randomized, double-blind trial of two dilutions of CJ-50300 (1 × 10⁸ and 1 × 10⁷ PFU/ml) at a 2:1 ratio [67]. Almost all volunteers had a successful take with one vaccination: 81 of 82 (98.7%) in the 1 × 10⁸ group and 100% in the 1 × 10⁷ group [67]. Seroconversion rates were 99 and 100%, respectively, and cell-mediated immune responses as measured by ELISpot were not statistically different between the two groups [67]. There were no differences in AEs between the groups, and no SAEs were reported. However, a woman in the 1 × 10⁸ group had an exanthem that raised a concern for possible generalized vaccinia, but the lesions were culture negative [67]. A Phase III open-label, single-arm trial of CJ-50300 in VACV-naive volunteers is ongoing [203].

Third-generation smallpox vaccines

As the global smallpox eradication campaign continued to reduce both the number of worldwide cases and the number of countries endemic for VARV, developed countries became increasingly concerned about the safety of first-generation smallpox vaccines. The USA and Canada opted to phase out routine smallpox vaccination starting in 1972, since neither country had had recent imported cases of smallpox. Countries in Asia and Europe faced the more likely possibility of imported smallpox, thus several countries established programs to develop attenuated vaccines that could replace traditional vaccination. With the renewed interest in smallpox vaccination has come a renewed interested in attenuated smallpox vaccines, and several clinical trials have recently been published (Table 2).

LC16m8

In Japan, concern over the safety of the available first-generation smallpox vaccines (Ikeda and Lister), led to the development of an attenuated alternative [69]. The first-generation Lister vaccine was serially passaged in primary rabbit kidney cells at 30°C and repeatedly selected for small pock formation on chicken egg chorioallantoic membranes (CAMs) [69]. The resultant strain, called LC16m8, was still replication competent in vivo in mammals, but was less neurotoxic than the parental Lister strain [69]. LC16m8 was used in an open-label, nonrandomized fashion in approximately 50,000 children in the 1970s, but only a subset consisting of 10,578 LC16m8 vaccinees was closely followed for clinical events. In this cohort, LC16m8 induced take responses in 95.1%, but the lesions were smaller than those elicited by the Lister vaccine. Local and systemic reactogenicity was less common than with Lister [69]. Hemagglutination inhibitory (HI) titers and NAb titers were comparable between LC16m8 and Lister [69]. Furthermore, LC16m8 vaccination modified the take response upon subsequent challenge with Lister [69]. While LC16m8 was reported to be generally safe, a single case of eczema vaccinatum was noted, and nine cases of auto-inoculation following scarification with LC16m8 were seen [69]. Once smallpox was confirmed to have been eradicated in South Asia, Japan ceased routine vaccination in 1976, but retained a stockpile of LC16m8 [69].

A recently published large study assessed LC16m8 in Japanese military personnel prior to deployment with United Nations Peacekeeping missions [70]. This was an open-label study with no comparator vaccine that assessed clinical and humoral immune responses in 1529 VACV-naive and 1692 previously vaccinated individuals [70]. Exclusion criteria were similar to other studies of live VACV vaccines, but individuals with atopic dermatitis could be included if their skin lesions were clinically stable [70]. Successful take rates were significantly higher in VACV-naive subjects than in previously vaccinated subjects (94.4 vs 86.6%) and appeared comparable to those historically seen with Lister [70]. GMT of NAb did not differ between the two cohorts following vaccination, but seroconversion rates were significantly higher among primary vaccinees [70]. No SAEs were reported, but two primary vaccinees had generalized exanthems thought to be possibly related to vaccination: one was diagnosed with allergic dermatitis and the other with erythema multiforme [70]. Local and systemic reactogenicity were generally lower in revaccinees, and no cardiac AEs were noted [70]. Results from a Phase I/II double-blind clinical trial comparing LC16m8 to Dryvax have not yet been published [204]. Detailed immunologic assessments of humoral or cellular immune responses from subjects vaccinated with LC16m8 are not yet available, but a murine study found that while LC16m8-vaccinated mice raised IgG against B5R, the titer was lower than from Dryvax-vaccinated mice [71]. The relative importance of the specific orthopoxvirus antigens targeted by vaccination remain incompletely understood; however, we found that MVA-inoculated vaccinees did have high titers of anti-B5R antibodies [72], suggesting that there may be differences between attenuated derivatives of VACV.

CVI-78

Attempts to develop safer smallpox vaccines date back to at least the 1930s [73,74]. One such candidate, CVI-78, was derived from the NYCBH strain, passaged repeatedly in rabbit testes and chick embryonic tissue, and then passaged in CAM [75,76]. CVI-78 was proposed as a primary vaccine for children in whom traditional smallpox vaccination was contraindicated. In its first study, CVI-78 was administered via scarification at a dose of 1 × 10^8.4 50% tissue culture infectious dose (TCID₅₀)/ml (and produced a take reaction in 87% of recipients) or subcutaneously at dilutions ranging from 1 × 10³ to 1 × 10^4.5 TCID₅₀ per dose to 594 children with eczema [76] (The TCID₅₀ is calculated by either the Reed-Muench or Karber statistical methods and is related to PFU by the Poisson distribution such that one TCID₅₀ is roughly 0.7 PFU [77]). As randomization was not mentioned in the report, the method of subject allocation is not clear. There was no difference in GMT NAb titers between scarification and sc. administration [76]. However, NAb titers decreased rapidly with serial dilution of CVI-78 [76]. None of the children suffered eczema vaccinatum [76].

A follow-up study recruited 1009 patients (mostly children ranging from 1 to 5 years of age), most of whom had a history of or active eczema and 20 children had ichthyosis while 20 had severe atopy without eczema [75]. Of the 1009 subjects, 879 were VACV-naive and 130 had been previously vaccinated [75]. A total of 326 subjects were scarified and the remainder received CVI-78 sc. at increasing dilutions [75], but subject assignment was not described. A total of 77% of scarified subjects had a take reaction while only a minority of sc.-inoculated subjects had any local reaction such as erythema or edema. Larger sc. injection volumes seemed to correlate with local reactogenicity more than the inoculum titer [75]. Fevers occurred in less than 20% of the children, and mild malaise and anorexia were reported in 15–22% [75]. Eczema vaccinatum was not seen, but two cases of erythema multiforme were noted, which is approximately the same incidence as historically seen with wild-type VACV (i.e., 1:500 primary vaccinees) [75]. All subjects seroconverted and NAb titers were generally comparable between the groups. The titers were similar to those induced by traditional scarification, except in the sc. group at the highest dilution where lower titers were seen [75]. In total, 104 subjects were challenged with standard calf-lymph derived VACV 3–6 months later, and all had attenuated take responses [75], but the original group assignments were not mentioned.

The final study with CVI-78 was performed in 1970 and was a single-center, randomized, double-blind comparison between CVI-78 (1 × 10^7.7 PFU/ml) and Dryvax (1 × 10^7.8 PFU/ml) via scarification in 95 healthy children between 1 and 5 years of age [78,79]. Major take responses were seen in 30 out of 49 (61%) children inoculated with CVI-78 compared with 44 out of 46 (96%) of those who received Dryvax [79]. Fever was noted in 8% of CVI-78 recipients compared with 26% of Dryvax recipients and generalized eruptions occurred in 2% of children scarified with CVI-78 versus 6.5% in the Dryvax group, although one case in the Dryvax group appeared to be an allergic reaction to a concomitant medication [79]. HI and NAb titers were lower in the CVI-78 group, and only 67 and 16% of the children had titers above the respective thresholds compared with 98 and 89% in the Dryvax cohort [79]. Antibody titers correlated with the development of a take in both groups, and 27% of the children in the CVI-78 group had neither a take nor a serologic response compared with only one child in the Dryvax group [79]. A secondary study was then undertaken with 48 of the children (26 from the CVI-78 group, 22 from the Dryvax group) who were challenged with open-label Dryvax (1 × 10^8.4 PFU/ml) via scarification at 5–9 months after the initial vaccination [78]. Major take responses were seen in 96% of the children originally vaccinated with CVI-78 versus 73% of the Dryvax group [78]. Whereas the primary take in the CVI-78 vaccinees had been smaller than that induced by Dryvax [79], the take after challenge was larger in the CVI-78 vaccinees [78]. Serum samples at day 28 following challenge were available from 20 children in the CVI-78 group and 19 in the Dryvax group. Of these 20 children, only two had seroconverted following CVI-78 compared with 17 of 19 in the Dryvax group. Following challenge, all of the children in the Dryvax group were seropositive for NAb compared with only 65% in the CVI-78 group [78]. A similar trend was seen for HI antibodies: 80% seroconverted in the CVI-78 group and 100% seroconverted in the Dryvax group [78]. The authors interpreted these results to suggest that CVI-78 alone would be insufficient to protect against smallpox, but that it might be useful as a priming vaccine to be boosted with conventional calf-lymph vaccine with the caveat that even boosting did not lead to seroconversion in some CVI-78 recipients [78]. However, 5 years after this study was published, smallpox was declared eradicated [1] and interest in vaccination against VARV waned.

Modified vaccinia Ankara: a new-generation smallpox vaccine

In an another effort to develop an attenuated smallpox vaccine for persons with contraindications to traditional smallpox vaccines, the wild-type chorioallantois vaccinia Ankara strain was serially passaged on chick embryo fibroblasts over 570 times until it was unable to grow in nearly all mammalian cells [80–83]. The resulting strain, MVA, was used in West Germany and Turkey at the end of the global smallpox eradication efforts and administered to approximately 120,000 persons without apparent AEs [81]. However, MVA was never used in a variola endemic area, and its efficacy at preventing clinical smallpox is unknown. With the renewed interest in biodefense vaccines, MVA has undergone several preclinical [84] and clinical trials [72,85–89] in an effort to obtain approval for use as a smallpox vaccine. Approximately 15% of the MVA genome was deleted during in vitro passage compared with the parental chorioallantois vaccinia Ankara strain [80,90], but the block in virus replication in nonpermissive mammalian cells occurs late in the viral life cycle after immature virions are formed, and hence MVA-infected cells express very high levels of virally encoded proteins [91–93]. The block is probably multifactorial and is not simply due to disruptions of orthopoxvirus immunomodulatory genes [90,91]. Despite the block in viral replication, MVA remains immunogenic and, in addition to studies of its applicability as an alternative smallpox vaccine [84], it is under study as a vaccine vector for HIV, malaria and other pathogens [83,94] as well as tumor-associated antigens (Table 3) [94–97].

Table 3.

Selected clinical trials of VACV-derived smallpox vaccines used as vectors for other infectious diseases or cancer.

Vaccine	Subjects Enrolled	Findings	AEs/SAEs	Reference
MVA and FPV expressing HIV-1 env, gag, tat, rev, nef, RT (PACTG P1059)	20 HIV-infected young adults	Safe and well-tolerated; increased HIV-1 specific CD4+ and CD8+ T cell responses vs baseline	Injection site reactions common; no SAEs or AEs related to product	[145]
MVA expressing HIV-1 nef	14 HIV-infected, remotedly VACV-vaccinated men	Increased total CD4+ and CD8+ T cell counts; 9 of 14 had ELISPOT responses to nef insert; MVA-specific antibodies induced	Local reactogenicity common, systemic reactogenicity seen; no SAEs	[146]
VACV and FPV expressing PSA and co-stimulatory molecules (PROSTAVAC-VF)	125 men with castration-resistant metastatic prostate cancer	44% reduction in deaths over 3 years and 8.5 months longer median survival	Injection site reactions common; one case of TTP and MI graded as “possibly related”	[110]
NYVAC-C (vP2010) expressing HIV-1 clade C gag, pol, env, nef	24 healthy volunteers (EV01)	50% ELISPOT responses to HIV-1 peptides (EV01)	Mild to moderate reactogenicity; no SAEs; one subject with high ALT after first DNA vaccination (EV02	[134,136–138]
	40 healthy volunteers (EV02)	83% ELISPOT responses to HIV-1 peptides in DNA-NYVAC group vs 35% in NYVAC alone; polyfunctional HIV-1 specific CD4+ and CD8+ T cells induced by DNA-NYVAC(EV02)
	147 healthy volunteers (EV03/ANRS Vac20)	91% ELISPOT response with 3 doses of DNA prime vs 80% with 2 doses; increased breadth of responses as well	Study pending	clinicaltrials.gov NCT00490074
NYVAC-B (vP2009) expressing HIV-1 clade B gp120 gag-pol-nef	80 healthy volunteers	HVTN 078 study ongoing	Not yet available	clinicaltrials.gov NCT00961883
MVA expressing HIV-1 clade B gag, protease, RT, env (MVA/HIV62)	120 healthy volunteers (HVTN 065)	T cell responses higher in DNA prime/MVA boost; binding antibodies higher in MVA alone group	Mild to moderate local and systemic reactogenicity	clinicaltrials.gov NCT00301184 [117]
	300 healthy volunteers (HVTN 205)	HVTN 205 study ongoing	Not yet available	clinicaltrials.gov NCT00820846
MVA expressing CD8+ T cell epitopes and truncated gag derived from HIV-1 clade A (MVA.HIVA)	35 healthy volunteers	Immunogenic for HIV-derived insert in majority of MVA recipients	Local reactogenicity common; one SAE (possible concomitant viral gastroenteritis)	[111–115]
	119 healthy volunteers	Only 10% of vaccines responsed to HIVA insert	Local reactogenicity common, no related SAEs
	24 healthy volunteers	4 of 8 MVA vaccinees and 8 of 8 DNA/MVA vaccinees developed T cell responses (mostly CD4+)	Not reported
MVA expressing HIV-1 clade C gag-RT-tat-nef and truncated gp160 (SAAVI-MVA-C)	48 healthy volunteers	HVTN 073 study ongoing	Study ongoing	clinicaltrials.gov NCT00574600
MVA expressing HIV-1 clade C env, gag, tat, rev, nef, and RT	32 healthy volunteers	82% and 100% ELISPOT and 91% and 100% binding antibody response rates in low- vs high-dose MVA recipients	Local reactogenicity higher in high-dose group; no SAEs, no cardiac events	[147]
MVA expressing env, gag, pol derived from HIV-1 CRF01_AE	48 healthy volunteers	Dose and route dependent cellular and humoral immune responses with 108 IM highest	Local reactogenicity more common with ID and higher dose; no related SAEs	[148]
MVA expressing 5T4 (TroVax)	19 patients with metastatic colorectal cancer undergoing chemotherapy	1/19 had complete remission, 6/19 had partial remissions, 5/19 had stable disease	No related SAEs, no increase in chemotherapy-related side effects	[123,149,150]
	28 patients with metastatic renal cancer	1 patient had partial remission; 14 had stable disease; addition of IFNα to MVA-5T4 increased antibody responses, but decreased cellular immune responses	No related SAEs
	733 patients with metastatic renal cancer	No difference in survival between vaccinees and placebo recipients; trend toward enhanced survival with higher antibody responses to 5T4	No difference in AEs/SAEs between vaccinees and placebo recipients
MVA expressing HPV E2 (MVA-E2)	54 women with CIN II or III	20/34 vaccinees with complete resolution; 16/20 response rate to excision in controls, but 3 relapsed within 1 year	Mild AEs reported	[151,152]
	50 men with intra-urethral flat condylomata	28/30 vaccinees with complete resolution; 13/20 controls responded to 5-FU, but 3 relapsed within 1 year	Mild AEs commonly reported
MVA expressing epitopes derived from melanoma antigens (MVA-Mel3)	14 HLA-A2+ patients with resected melanoma at risk of recurrence	7/13 vaccinees had an epitope-specific tetramer response	Mild local and systemic reactogenicity common	[128,129]
	41 HLA-A2+ patients with unresectable melanoma	71% of vaccinees had Melan-A-specific tetramer responses which correlated with increased median survival; 21% had some degree of clinical response	Dose-dependent mild-to-moderate local and systemic reactogenicity seemed to decrease with subsequent doses; one possible allergic reaction noted
MVA expressing M. tb antigen 85A (MVA-85A)	42 healthy volunteers	Highly immunogenic in BCG-naïve and BCG-vaccinated recipients	Local reactogenicity very common; systemic reactogenicity common	[118,119]
MVA and FPV expressing malaria antigens ME-TRAP	405 children in coastal Kenya	No protection against malaria	No difference in AEs/SAEs between vaccinees and controls; no SAEs related to product	[121,122]
MVA expressing influenza A NP and M1	28 healthy adults	2.5 to 11 fold increase in T cell responses	Local reactogenicity more common with ID administration; systemic reactogenicity higher with high dose IM administration	[153]
MVA expressing EBV EBNA-1 and LMP2 (MVA-EBNA1/LMP2)	37 patients with treated nasopharyngeal cancer and residual EBV DNA load	Study enrolling	Not yet available	clinicaltrials.gov NCT01094405
VACV and FPV expressing CEA, MUC-1, and co-stimulatory molecules (PANVAC-VF)	48 patients with metastatic breast cancer	NCI-05-C-0229 study enrolling	Not yet available	[154] clinicaltrials.gov NCT00179309

Abbreviations: VACV – vaccinia virus; HVTN – HIV Vaccine Trial Network; PACTG – Pediatric AIDS Clinical Trial Group; PSA – prostate-specific antigen; TTP – thrombogenic thrombocytopenic purpura; FPV – fowlpoxvirus; ALT – alanine aminotransferase; ANRS - Agence nationale de recherches sur le sida et les hépatites virales; SAAVI – South African AIDS Vaccine Initiative; HPV – human papillomavirus; CIN – cervical intra-epithelial neoplasia; 5-FU – 5-fluorouracil; M. tb – Mycobacterium tuberculosis; BCG – bacille Calmette-Guérin; ME-TRAP – multiple epitope string and thrombospondin related adhesion protein; NP – nucleoprotein; M1 – influenza matrix protein 1; ID – intradermal; IM – intramuscular; CEA – y antigen; MUC-1 – mucin-1

Three different MVA vaccines have recently been tested in humans, namely TBC-MVA (Therion) [86], ACAM3000 (Acambis/Sanofi) [72,88] and IMVAMUNE/MVA-BN (Bavarian-Nordic) [87,89,91], but only IMVAMUNE has advanced to Phase II studies (Table 2). The clinical trials published to date have demonstrated that MVA is safe and well-tolerated across a wide dose range (10⁶–10⁸ TCID₅₀ in four trials [72,85,87–89], 10⁶ PFU/ml in a fifth [86]) via id., sc. or im. routes with approximately 450 persons enrolled in the various published trials and over 2000 more in as yet unpublished trials [84]. Reported local and systemic reactogenicity have generally been mild and considerably less than with Dryvax or ACAM2000 [84]. In addition, there have been no reports of vaccine-related SAEs, cardiac events or cases of eczema vaccinatum in any published trial with MVA.

MVA is immunogenic in humans with induction of antibodies, NAb and virus-specific T-cell responses following vaccination [72,85–89,98]. Importantly, antibodies induced by MVA inoculation appear to neutralize VARV approximately as well as antibodies induced by Dryvax [98], with titers that exceed those thought to be necessary to prevent clinical smallpox [62,63]. Interestingly, the sc. route seemed superior to the im. route in terms of VARV neutralization [98], and the id. route allowed a tenfold lower dose of MVA to be used in another study [72,88]. Three trials have noted that, in contrast to antibody titers, cellular immune responses induced by MVA have not followed a clear dose-response or route-dependent pattern [72,85,89]. This suggests that for MVA inoculation, humoral and cellular immune responses have distinct priming requirements such that antibody responses follow a dose-reponse trend whereas T cells may require a threshold dose beyond which further antigen only minimally increases responses.

Furthermore, MVA vaccination is able to attenuate the primary cutaneous lesion induced by scarification with Dryvax [81,85,87,88], an important and long-held correlate of protection against clinical smallpox infection [1]. Three published studies [85,86], including our recent study [88], examined take responses in MVA-vaccinated subjects. Frey et al. found that subjects who had been inoculated with MVA at either of four doses or routes (2 × 10⁷ TCID₅₀ sc., 5 × 10⁷ sc., 1 × 10⁸ sc., 1 × 10⁸ im.) had significantly smaller take lesions following Dryvax and shed significantly less VACV from the scarification site [85] compared with subjects who had not received MVA. Similarly, Parrino et al. found that receipt of even a single im. dose of 10⁶ PFU/ml of MVA attenuated the take response and significantly decreased viral shedding at the site of Dryvax scarification [86].

In our trial using ACAM3000, we observed attenuated take responses and decreased viral shedding following Dryvax scarification that correlated with the dose and route of MVA inoculation [88]. Representative photographs are shown in Figure 1 with two placebo recipients experiencing robust takes typical of VACV-naive individuals and an MVA recipient from the 10⁸ sc. group showing a blunted take. Attenuated takes were seen in 72% of MVA recipients and were most significantly attenuated in the 1 × 10⁷ id. and 1 × 10⁷ im. groups [88]. We also measured viral shedding in each scarified individual at each visit and stratified the titers by vaccine group. Individuals who had been vaccinated with 10⁷ im., 10⁷ sc., 10⁷ id. or 10⁸ sc. of MVA all cleared virus from the scarification site significantly more rapidly than placebo recipients [88]. Furthermore, the peak NAb response induced by MVA inoculation correlated with the degree of attenuation of the lesion, as well as both reduced duration and titer after challenge with Dryvax [88].

Figure 1. Clinical take response correlates with viral titer after scarification.

Representative usual take responses are shown from two placebo recipients (A & B) and a blunted take from a subject who received two doses of MVA at 10⁸ sc. (C) as part of a clinical trial with MVA [72]. Subjects who elected to receive scarification with Dryvax were followed for clinical take, graded as 3 (normal take), 2 (reduced take), or 1 (limited take) based on CDC criteria and correlated with progency viral shedding [88].

Photographs courtesy of Lindsey R Baden and Marissa A Wilck.

Taken together, these data indicate that MVA is safe, generally well tolerated and immunogenic in healthy subjects. MVA elicits both antibodies and T cell-mediated immune responses, and vaccination with MVA leads to a blunted take response with less viral shedding at the site of subsequent Dryvax scarification. In prelicensure trials, including the aforementioned published trials, over 2700 subjects were vaccinated with more than 4700 doses of IMVAMUNE, including more than 1000 subjects from groups with contraindications for traditional smallpox vaccines, including HIV-infected and eczematous patients [99]. A Phase III trial of IMVAMUNE is planned to open soon [205].

Vaccinia viruses as vectors

As the global VARV eradication campaign was concluding, technical developments in molecular biology allowed the creation of genetically recombinant organisms and viruses. VACV was among the earliest eukaryotic viruses to be engineered to express heterologous genes [100,101] and the potential for recombinant VACV to be used as a vaccine vector was immediately recognized. VACV is itself the prototypic vaccine and lyophilization allowed for global deployment of VACV in the eradication campaign. The large genomic size of the orthopoxviruses allows them to carry multiple large genetic inserts and poxviruses are exclusively cytoplasmic viruses with essentially no risk of integration into the host genome. Poxviruses also have very low mutation rates and selected deletion of viral genes can enhance both safety and immunogenicity by abrogating VACV immune evasion strategies [102]. While VACV has been engineered as a vector for a variety of infectious diseases and malignancies, the first recombinant VACV vector to enter use was the veterinary vaccine, Raboral V-RG^® (Merial), which expresses the rabies virus glycoprotein G at the VACV Copenhagen thymidine kinase locus [103]. Baits containing liquid vaccine have been deployed in wilderness areas to vaccinate sylvan rabies hosts such as foxes, raccoons and skunks [104]. Oral rabies vaccination programs using Raboral V-RG have led to elimination of rabies from red foxes in several western European countries, near elimination from red foxes in Ontario, control of a rabies outbreak in coyotes in Texas, and containment of raccoon rabies in the eastern USA [105]. While loss of thymidine kinase leads to some attenuation, the recombinant VACV is still replication competent and able to cause inadvertent and potentially serious infections [106,107].

A recombinant VACV expressing the prostate-specific antigen (PSA) in conjunction with T-cell costimulatory factors has been tested in clinical trials. PROSTAVAC-VF is a combination regimen consisting of recombinant VACV and fowlpoxvirus (FPV), which expresses PSA as well as human B7.1 (CD80), ICAM-1 (CD54) and LFA-3 (CD58; collectively called TRICOM) to enhance cell-mediated immune responses to PSA [108]. The recombinant VACV is given as a prime followed by six boosts with the heterologous recombinant FPV, since anamnestic VACV-specific NAb limited responses using the homologous vector [108]. In a recently published clinical trial, 82 VACV-experienced subjects with castration-resistant metastatic prostate cancer received PROSTAVAC-VF, while 40 control subjects received empty poxvirus vectors [108]. Progression-free survival at 6 months (the primary end point) did not differ between the groups at 23% in the PROSTAVAC-VF group versus 25% in the control group. However, overall survival at 3 years (30 vs 17%) and median survival (25.1 vs 16.6 months) were both higher in the PROSTAVAC-VF group [108].

MVA as an attenuated vector for infectious diseases

Safety concerns have greatly limited development of recombinant VACV vectors, particularly for infectious diseases where many of the target populations are young and VACV-naive. However, MVA shares many of the positive attributes of replication-competent poxviruses with considerably enhanced safety and has long been proposed as an improved vector [93]. Numerous preclinical studies have demonstrated safety, immunogenicity and efficacy of MVA-delivered immunogens in animal models of HIV infection, malaria and TB among others (reviewed in [95]). In the previous decade, many of these vaccine concepts entered clinical trials (Table 3) and selected studies are briefly discussed later.

One of the first concepts to be tested employed a DNA prime–MVA boost regimen intended to induce T-cell responses against HIV-1 using a linear construct of HIV-1 consensus clade A epitopes fused to Gag p24/p17, called MVA.HIVA [109–111]. In the first study, local reactogenicity was commonly seen following id. MVA administration, although a single subject having systemic systems may have had a concomitant viral illness [110,111]. Positive IFN-γ ELISpot responses were seen in seven of eight MVA–MVA recipients and eight out of nine DNA–MVA recipients [110]. A subsequent double-blind, placebo-controlled study of 119 volunteers tested two doses of the DNA prime, followed by two doses of 5 × 10⁷ PFU/ml id. of the recombinant MVA boost at either 8 and 12 weeks or 20 and 24 weeks after the DNA prime or placebo [112]. No SAEs related to the vaccine were reported, and again local reactogenicity was commonly noted following the id. administration of MVA [112]. Only 10% of the 119 subjects had positive ELISpot responses to the HIVA immunogen following vaccination, and there was no clear trend to the distribution of the responders based on their group assignment [112]. In a follow-up double-blind, placebo-controlled trial of 24 healthy subjects, higher doses of the DNA prime–MVA boost and MVA prime–MVA boost components were used, but because of a trial delay, the MVA/MVA group received their booster doses between 1 and 9 months after the priming dose [113], complicating assessment of this arm. HIV-1-specific T-cell responses could only be detected in half of the MVA–MVA group using an in vitro culture method to enhance the IFN-γ ELISpot, while all the DNA–MVA recipients had detectable T-cell responses [113]. Taking both groups into consideration, 12 out of 16 vaccinated subjects (four subjects in each group of 12 received placebo) were considered responders, and the T-cell responses were mostly mediated by CD4⁺ T cells. Vaccine-elicited T cells secreted multiple cytokines and chemokines in response to HIV-1 clade A peptide stimulation [113]. Trials using the same immunogens are ongoing in HIV-1 infected patients taking HAART to determine if a DNA prime–MVA boost can lead to better immune-mediated control of chronic HIV-1 infection ([114] and reviewed in [109]).

Another candidate HIV-1 vaccine using a DNA prime–MVA boost regimen has been tested in a Phase I study and is being tested in an ongoing Phase IIA study presently. In the HVTN 065 study, a DNA prime and recombinant MVA boost encoding HIV-1 clade B-derived gag, env, protease and reverse transcriptase (the DNA prime also encoded tat, rev and vpu) and recombinant MVA alone were compared with placebo, via im. injection in 120 healthy volunteers [115]. Mild-to-moderate local reactogenicity was more common with MVA than DNA, but only one subject discontinued vaccination owing to side effects related to vaccination (chest pain and dyspnea 30 min after injection) [115]. HIV-1 specific CD4⁺ T-cell responses were seen in over 75% of subjects who received two doses of DNA and two doses of MVA, compared with less than half of subjects who received three doses of recombinant MVA [115]. A similar, but quantitatively lower, trend was seen with respect to CD8⁺ T-cell responses: 42% in the two-dose DNA–two-dose MVA group versus 17% in the three MVA dose group [115]. Antibody responses, however, were higher in the three-dose MVA group than the DNA prime–MVA boost group, with a trend toward broader neutralization activity in the three-dose MVA group, at least against relatively neutralization-susceptible tier 1 viruses [115]. Interestingly, DNA priming appeared to interfere with the development of VACV-specific cell-mediated immune responses, at least after the first dose of recombinant MVA [115]. The two-dose DNA–two-dose MVA and three dose recombinant MVA regimens are being compared in an ongoing Phase IIA study that aims to enroll 300 participants (HVTN 205 [206]).

Modified vaccinia Ankara has also been evaluated as a vector to deliver the immunodominant Mycobacterium tuberculosis antigen 85A id. to both naive volunteers and volunteers who had previously received the current M. tuberculosis vaccine, bacille Calmette–Guérin (BCG) [116,117]. Local reactogenicity was universal with systemic reactogenicity occurring in a third of vaccinees regardless of whether they were BCG naive or BCG primed [117]. Previous BCG recipients had an anamnestic CD4⁺ T-cell response that was twofold higher than responses in the BCG-naive group and nearly tenfold higher than a control group given BCG alone [117]. A series of follow-up studies are assessing the safety of MVA85A in infants [207], HIV-infected persons [208] and persons latently infected with M. tuberculosis [209].

Malaria has also been the target of recombinant MVA vaccines, and clinical trials have advanced to the Phase IIB stage of development. Recombinant FPV and MVA vectors have been generated which express an immunogen consisting of a linear construct of CD4⁺ and CD8⁺ T-cell and B-cell epitopes (multiple epitope [ME]) from the Plasmodium falciparum pre-erythrocytic stage fused to the thrombospondin-related adhesion protein (TRAP). These were found to be safe, immunogenic, and partially protective against experimental malaria challenge in naive adults [118]. A regimen of two id. doses of FPV-ME-TRAP followed by MVA-ME-TRAP was then studied in 405 healthy children, aged 1–6, in coastal Kenya in a randomized double-blind trial using rabies vaccine as an active comparator [119]. No SAEs related to vaccination were seen, but local reactogenicity was noted in 15–20% of the recombinant poxvirus recipients compared with 2% of those who received rabies vaccine [119]. Modest immunogenicity by IFN-γ ELISpot was noted, but no efficacy with respect to time to first episode of malaria, episodes of fever with any degree of parasitemia or incidence of multiple episodes of malaria was seen during 9 months of follow-up [119]. Immune responses did not correlate with any of the clinical outcomes, and both surveillance parasitemia prevalence and hemoglobin level did not differ between recombinant poxvirus or rabies vaccine recipients [119]. An additional 18 months of clinical follow-up was done on 387 vaccinees, and again no efficacy of vaccination with the poxvirus-ME-TRAP regimen was seen [120].

MVA as a vector for therapeutic immunization against cancer

There is also considerable interest in the use of MVA as a vector to deliver tumor-specific antigens to induce immune responses that may help control malignancies [96] and several of these therapeutic vaccines have advanced to clinical trials (Table 3). The oncofetal antigen 5T4, which is expressed by a number of adenocarcinomas, is a notable target and MVA-5T4 (TroVax) has been shown to be safe and immunogenic, but clinical responses have been disappointing thus far (reviewed in [121]). The use of replication-competent VACV as an oncolytic virus [122,123] is beyond the scope of this article.

The observations that melanomas can regress spontaneously and that antigen-specific infiltrating lymphocytes can be isolated from tumors has led to melanoma being considered as a potential target for many different immunotherapy approaches [124,125]. One strategy has been to combine seven HLA-restricted epitopes from five melanoma antigens (tyrosinase, Melan-A/MART-1, MAGE-A1, MAGE-A3 and NY-ESO-1) into a polyepitope string (Mel3). A DNA prime–MVA boost approach that delivered this polyepitope immunogen was first tested in 14 patients with surgically resected melanoma but a high risk of recurrence [126], half of whom were randomized to a two-dose DNA and two-dose MVA regimen while the other half received four doses of recombinant MVA at 2 week intervals. Mild local and systemic reactogenicity was noted commonly but one subject reported noticeable skin inflammation at three nevi and was found on biopsy to have CD4⁺ and CD8⁺ T lymphocytes infiltrating the nevi [126]. Two of six subjects in the DNA–MVA group had a positive Melan-A/MHC tetramer-specific response (one subject withdrew) while four of seven subjects in the MVA only group had positive responses [126]. However, except for one subject in the DNA–MVA cohort who did not respond to Melan-A but did respond to NY-ESO-1, no other epitope-specific immune responses were detected ex vivo [126].

In a follow-up, open-label, dose-escalation study, progressively higher doses of DNA-Mel3 and MVA-Mel3 were tested in 41 patients with unresectable stage III or IV melanoma [127]. Again, mild-to-moderate local and systemic reactogenicity was commonly seen and seemed to increase in severity with higher doses of MVA, but was noted to decrease with subsequent doses [127]. However, one subject had a syncopal episode 12 h following the second MVA vaccination that was thought to be related to an allergic reaction to the vaccine [127]. Cellular immune responses were evaluable in 36 patients, 24 of whom were tetramer-positive to at least one tumor-specific epitope. DNA priming elicited low-frequency responses in only 25% of subjects while boosting with MVA induced tumor antigen-specific responses to as high as ten of 11 subjects (91%) in the highest-dose (1 × 10⁹ PFU) MVA groups [127]. Responses measured by IFN-γ ELISpot were lower with only 11 of 36 subjects (31%) having a detectable response to at least one epitope. Most subjects (78% overall) seroconverted to MVA with response rates ranging from 50% at 5 × 10⁷ PFU per dose to 100% at 1 × 10⁹ PFU, but these antivector responses did not seem to interfere with progressive boosting in a subset of patients who receive additional off-protocol doses of MVA-Mel3. Clinical responses were seen in only eight of 39 (21%) evaluable subjects: six had disease stabilization for at least 5 months, while two had regression of their original lesions but developed new lesions. Interestingly, of the eight subjects with clinical responses, seven had Melan-A-specific tetramer-positive CD8⁺ T cells and median survival was longer (100 weeks) among subjects who had Melan-A-specific tetramer-positive CD8⁺ T-cell responses compared with nonresponders (37 weeks) [127].

As vector-specific responses were found to be immunodominant in an analysis of the earlier MVA-Mel3 trial [128], genetic modification of vaccinia-derived vectors to enhance immunogenicity of the insert has been proposed [129]. Selective deletions of VACV immunomodulatory or pathogencity genes have also been proposed as a means to increase immune responses to the immunogen [49] without sacrificing safety. Many of these genetically engineered VACV strains have shown promise in preclinical models (reviewed in [49,95]). Of these concepts, only one (NYVAC) has entered clinical trials.

NYVAC: a fourth-generation vector for vaccines against infectious diseases

Although MVA is safe, immunogenic and has been shown to successfully deliver a variety of immunogenic genetic inserts, concern exists that its profound attenuation is responsible for the modest immunogenicity noted in several of the clinical trials discussed. It has been proposed that targeted gene deletion from the parental VACV strains can lead to adequate attenuation for safety purposes, yet enhance immunogenicity of the transgene [49]. While many fourth-generation VACV-derived vectors have been tested in animal models, only NYVAC has been tested in clinical trials thus far. NYVAC was derived from a plaque-purified isolate from the VACV Copenhagen strain by selective deletion of 18 open reading frames that encode for virulence proteins without deletion of genes necessary for expression of transgenes or viral replication in permissive cells [130]. NYVAC replicated to very low levels in human cell lines, did not produce pock lesions or progeny virus on rabbit skin and was much less neurovirulent in newborn mice and less virulent in athymic and pharmacologically immunosuppressed mice than wild-type VACV. Nonetheless, it was able to efficiently express a transgene (rabies glycoprotein) even in relatively nonpermissive cells [130].

The first published trial using NYVAC as a vector tested its capacity to deliver seven genes derived from P. falciparum in an open-label study at two dosages in 49 malaria-naive volunteers [131]. Local reactogenicity was common in all dose groups, but systemic reactogenicity was more common in the high dose group and decreased upon the third vaccination [131]. VACV-specific NAbs were elicited to only low titers in VACV-naive volunteers, but prior recipients had anamnestic responses to recombinant NYVAC [131]. Antibody responses to the malaria antigens were generally higher in the high dose group and both VACV-naive and VACV-experienced vaccinees responded, although response rates were higher in the VACV-naive subgroup [131]. Cell-mediated immune responses were also generally higher in the high dose group [131]. Experimental challenge with P. falciparum was administered to 19 low dose and 16 high dose vaccinees, and responses were compared with eight control subjects. While there was a delay in time to parasitemia in vaccinees, only one subject from the low-dose group was protected from parasitemia, suggesting that the vaccine had limited efficacy [131]. This study demonstrated that NYVAC was safe, well-tolerated and could be used successfully as a vector for heterologous immunization, although pre-existing anti-VACV immunity inhibited antibody responses to the insert [131].

A subsequent series of studies has evaluated NYVAC as a vector for HIV-1-derived antigens. In the first study, EV01, 24 healthy volunteers were randomized to either NYVAC expressing HIV-1 clade C-derived gag, pol, env and nef (n = 20) or placebo (n = 4) [132]. Mild-to-moderate local and systemic reactogenicities were noted in most participants, with no SAEs, and few AEs thought to be ‘possibly related’ to the vaccine [132]. Only ten NYVAC-C recipients had evaluable cells for testing cell-mediated immune responses, and only five of these subjects had detectable IFN-γ ELISpot responses to the HIV-1 inserts [132]. Interestingly, IgM responses to clade C Env were detected in more vaccinees than IgG responses and one subject’s IgM could neutralize autologous virus [133].

In a follow-up study (EV02) NYVAC was preceded by a DNA vaccine with the same insert in a prime–boost regimen with 23 healthy volunteers randomized to DNA/NYVAC-C and 17 to NYVAC-C [134]. Two individuals discontinued vaccination, one because of a vasovagal reaction with the first DNA injection and the second because of an asymptomatic rise in alanine aminotransferase, evaluated as ‘possibly related’; aside from this, local and systemic reactogenicity was mild-to-moderate [134]. Based on an intention-to-treat analysis, 83% of subjects in the DNA/NYVAC-C group responded to vaccination by IFN-γ ELISpot compared with 35% in the NYVAC-C group, while a per-protocol analysis showed a 90 versus 40% response rate [134]. Cell-mediated immune responses remained detectable 6 months following the completion of vaccination in 80% of the DNA/NYVAC-C group versus 13% in the NYVAC-C group [134]. While most volunteers were VACV-naive prior to the trial, post hoc analysis showed little effect of prior smallpox vaccination on NYVAC-C immunogenicity, although the magnitude of the peak ELISpot reponses were lower in VACV-experienced male subjects [134]. Antibody responses followed a similar trend to cell-mediated immune responses, as 75% of DNA/NYVAC-C recipients and 27% of NYVAC-C recipients raised IgG to gp140. However, no neutralization activity was detected and all recipients except one had seroreverted by week 48 [134]. A more detailed assessment of the T-cell responses induced by the vaccines showed that the elicted T cells were primarily Env specific, were predominantly CD4⁺ (although CD8⁺ T cells were elicited in about half of responders), and the T-cell responses consisted of polyfunctional memory cells that were capable of secreting multiple effector cytokines upon HIV-1 peptide stimulation [135]. Based on these results, a larger study of 147 volunteers, EV03/ANRS Vac20, was performed, and while analysis is ongoing, preliminary reports indicate that the regimen of three DNA booster injections followed by one NYVAC injection is superior to two DNA injections followed by two NYVAC injections with a 91 versus 80% response rate [136]. A recombinant NYVAC with an HIV-1 clade B-derived insert is also being testing in a heterologous prime–boost regimen with a recombinant adenovirus serogroup 5 (rAd5) encoding the same clade B insert (HVTN 078) [210]. This trial is of particular interest since NYVAC-C induced primarily CD4⁺ T-cell responses and rAd5 vectors induce primarily CD8⁺ T-cell responses [137]. This regimen might offer a ‘balanced’ CD4⁺ and CD8⁺ T-cell response that has been proposed as potentially important for vaccine efficacy [138].

Expert commentary & five-year view

In the past 5 years, the traditional calf lymph-derived VACV vaccine (Dryvax) has been replaced by a tissue culture-derived vaccine (ACAM2000) [46]. The clinical trials that led to FDA approval of ACAM2000 suggest that it is able to efficiently induce takes and both humoral and cell-mediated immune responses in naive recipients [46]. There appears to be some decreased efficacy in boosting responses in recipients who received Dryvax in the past [46], but it is not known if this will translate into decreased efficacy against smallpox. As the risk of a smallpox outbreak appears low, the clinical efficacy of ACAM2000 may never be known. As the military smallpox vaccination program continues [139] we will gain more insights into the relative risks of ACAM2000 with respect to accidental transmission and other rare side effects of live VACV vaccination.

In the next 5 years, IMVAMUNE will probably be considered for approval as an alternative smallpox vaccine for individuals in whom live VACV vaccination (such as ACAM2000) would be contraindicated. It is unlikely that MVA will be recommended for universal vaccination, as the risk of VARV release, while not known, is probably quite low. Furthermore, while immune correlates of protection against smallpox are not well understood, MVA is decidedly less immunogenic than VACV. However, there are substantial numbers of military, healthcare and laboratory personnel who work with, or are at risk for contact with, pathogenic orthopoxviruses, but cannot currently be vaccinated because of concomitant medical conditions such as eczema or immunosuppression. MVA may prove a reasonable alternative to ACAM2000 in these individuals and could possibly serve as a priming vaccine [84], but as yet no trial has studied MVA priming inoculation followed by ACAM2000 via scarification in eczematous or immunocompromised hosts. NYVAC has not been specifically studied as an alternative smallpox vaccine, but does induce VACV-specific responses when used as a vector.

Over the next 5 years, an increasing number of clinical trials will be reported in which MVA and NYVAC have been used as vectors for immunogens for other infectious diseases and cancer. Safety concerns will likely limit nonattenuated VACV vector use to populations expected to have been previously vaccinated (such as with prostate cancer for the coming decades). MVA and NYVAC have not been directly compared head-to-head but preclinical studies in murine and simian models suggest that they may induce distinct facets of the immune system [140,141]. Furthermore, analysis of published clinical trial data suggests that both MVA and NYVAC are useful as booster vaccines when preceded by a DNA vaccine. Ongoing clinical trials with MVA and NYVAC will help clarify if antivector immunity impairs responses to the immunogen of interest in both therapeutic and prophylactic vaccine approaches. The results of the Thai canarypox prime–protein boost HIV vaccine trial (RV144) [142] suggest that poxvirus efficacy may be augmented by recombinant protein boosting and future studies will undoubtedly examine this hypothesis with MVA and NYVAC. Depending on the results of pending heterologous prime–boost studies, future trials may involve novel combinations of DNA vaccines, recombinant poxviruses, recombinant adenoviruses and protein boosts.

Key issues.

• Vaccination with calf lymph-derived live vaccinia virus (VACV) via dermal inoculation (scarification) is responsible for the eradication of smallpox (variola virus, VARV).

• Accidental or criminal release of VARV remains a threat and has stimulated renewed interest in smallpox vaccination.

• A cell-culture grown VACV vaccine has been approved for use in humans (ACAM2000) and elicits similar immune responses to historical VACV vaccines, but may carry similar risks of complications.

• Attenuated smallpox vaccines based on VACV have been assessed but not yet approved for clinical use.

• Modified vaccinia Ankara (MVA) is the leading attenuated smallpox vaccine candidate and appears safe, well-tolerated and immunogenic.

• VACV-derived viruses can be used as vectors to deliver a wide range of immunogens in vivo.

• MVA and a fourth-generation VACV-derivative, NYVAC, are under investigation as vectors for HIV, TB, malaria and cancer vaccines.