Subjects and Methods

Subjects and vaccination. The Saint Louis University and the University of Massachusetts Medical Schools institutional review boards approved the randomized clinical trial that was conducted at the National Institute of Allergy and Infectious Diseases-funded Vaccine Treatment and Evaluation Unit at Saint Louis University. Subjects were enrolled from 1 April-15 May 2002, after providing written, informed consent. Two groups of healthy adult individuals were recruited, and details of inclusion/exclusion criteria have been described elsewhere [17]. Individuals were eligible for inclusion in the previously vaccinated group (vaccinia nonnaive) if they were 32–60 years of age, had been vaccinated before 1971, and had either a typical vaccine scar or documentation of a previous smallpox vaccination; a second group was vaccinia naive and 18–31 years of age (born 1972–1984).

Ten vaccinia-naive and 20 vaccinia-nonnaive individuals were assigned to receive 1 inoculation of undiluted Dryvax vaccine (∼10⁸ pfu/mL). These participants were part of a larger study to evaluate 4 dilutions of Dryvax in previously vaccinated individuals. All individuals completed the vaccination phase of the study. Vaccination sites were assessed on days 0, 3–5, 6–8, 12–15, and 26–30, and the presence of a major skin reaction was recorded on days 6–8 after vaccination.

Vaccine and diluents. The lyophilized vaccinia vaccine (Dryvax; Wyeth Laboratories; lot no. 4008248) and vaccine diluents (Chesapeake Biological Laboratories; lot no. 1468-1A) were provided by the Centers for Disease Control and Prevention (CDC; Atlanta, GA). Vaccination was performed by use of the methods described by the CDC [8]. In brief, a sterilized bifurcated needle was inserted vertically into a prepared vial of Dryvax vaccine. The area of skin over the deltoid region of the upper arm received 15 strokes of a bifurcated needle containing virus. The site was covered with a nonocclusive sterile dressing.

Blood samples. Approximately 90 mL of citrate-anticoagulated blood was drawn in Becton Dickenson Separator BD tubes before vaccination and on days 3–5, 6–8, 12–15, and 26– 30 after vaccination. Tubes were shipped overnight to the University of Massachusetts, and peripheral blood mononuclear cells (PBMCs) were harvested and cryopreserved within 30 h of phlebotomy, as described elsewhere [9].

CMI assays. Cryopreserved PBMCs collected from individuals before vaccination and on days 3, 7, 14, and 28 after vaccination were thawed and tested in the vaccinia-specific CMI assays (cytotoxic T cell lysis [CTL], interferon [IFN]-γ ELISPOT, and lymphoproliferation). The target cells used in the CTL assay were autologous Epstein-Barr virus (EBV)-transformed B-lymphoblastoid cell lines (BLCLs), which were infected with vaccinia virus 1 day before and labeled with ⁵¹Cr the day of the assay [10]. Responder PBMCs were stimulated with virus-infected autologous PBMCs for 6 days at 37°C and then were added, at various effector/target (E/T) ratios, in 96- well U-bottom plates, in triplicate to the 51Cr-labeled infected autologous BLCLs, for 4.5 h, as described elsewhere [10]. At each E/T ratio (90, 30, and 10), the percentage vaccinia-specific immune lysis was calculated as the difference between percentage lysis of vaccinia-infected targets and percentage lysis of uninfected targets; 30% lytic units (LU) per 10⁶ cells were calculated by use of an exponential fit method [10], using software provided by Proteins International.

A modified ELISPOT assay was used to detect release of live virus-specific IFN-γ by cryopreserved PBMCs, as described elsewhere [7], except that vaccinia stock virus was used to stimulate PBMCs at an MOI of 1.0. The frequency of vaccinia-specific IFN-γ-positive cells per 10⁶ PBMCs was determined, and the results were considered to be positive if the number of spotforming cells (sfcs) per 106 PBMCs in virus-stimulated wells was 2-fold higher than the number of sfcs per 10⁶ PBMCs in the control wells and at least 12 sfcs were present per 10⁶ PBMCs.

Lymphoproliferation assays were performed in replicates of 5, as described elsewhere [9]. For each group of replicates, stimulation index (SI) was calculated by use of the mean of the 3 values between the high and low counts-per-minute wells.

Statistical considerations. The primary objective of this portion of the clinical study was to compare the CMI responses, antibody levels, and skin reactions between vaccinia-naive and vaccinia-nonnaive individuals who received undiluted Dryvax vaccine (vaccinia naive, n = 10; vaccinia nonnaive, n = 19) and had a major skin reaction after the initial vaccination.

Descriptive statistics of CMI responses (CTL, ELISPOT, and lymphoproliferation), neutralizing antibody levels, and local skin reactions (lesion, erythema, and induration) were summarized separately for the vaccinia-naive and vaccinia-nonnaive groups at each of the scheduled study visits. In the present study, all individuals had blood drawn within study guidelines. Blood drawn on days 0 (before vaccination), 3–5, 6–8, 12–15, and 26–30 are presented, using day 3 for blood drawn 3 or 4 days after vaccination, day 7 for blood drawn 7 or 8 days after vaccination, day 14 for blood drawn 13–15 days after vaccination, and day 28 for blood drawn 26–29 days after vaccination. Results were compared between the vaccinia-naive and vaccinia-nonnaive groups by use of the 2-sided Wilcoxon rank sum test. Because of the small sample size and skewness of the data, statistical models were not performed for the purpose of this portion of analysis. The linear regression model (the general linear model procedure in SAS [SAS Institute]) was also used, to examine the association between CMI responses and lesion sizes. The model also controlled for baseline antibody levels. The fitness of regression models was assessed by the F-test of model significance, and strength of association was assessed by R². The nature of the association is explained in terms of fold increase, in the underlying CMI result, per the unit (1 mm) increase in lesion size.

Results

Clinical observations. A total of 30 subjects were vaccinated: 10 (100%) of 10 vaccinia-naive individuals and 19 (95%) of 20 previously vaccinated individuals had major skin reactions after vaccination with undiluted vaccine. Fever was more common in vaccinia-naive individuals (P < .001). Timing of the skin reaction did not differ between the 2 groups, although skin reactions in vaccinia-naive individuals were larger from days 7 to 28, compared with those in vaccinia-nonnaive individuals. On day 14, lesions (P = .003) and erythema (P = .005) were significantly larger in the vaccinia-naive individuals than in the vaccinia-nonnaive individuals. The difference for size of induration was marginally significant (P = .057) (table 1.)

Vaccinia-specific CTL and IFN-γ ELISPOT responses. Figure 1A demonstrates vaccinia-specific CTL activity on days 0, 3, 7, 14, and 28 after vaccination in vaccinia-naive and vaccinia- nonnaive individuals. When the magnitude of the CTL responses, measured as percentage specific lysis, in individuals with major skin reactions in each group were compared, no significant difference was detected between vaccinia-naive and previously vaccinated individuals (table 2.) None of the vaccinia- naive individuals demonstrated vaccinia-specific CTL activity on day 0 (before vaccination), whereas 2 of 20 vaccinianonnaive individuals had positive but very low CTL activity (LU, 5 and 7, respectively; day-14 mean LU for the group, 244).

ELISPOT assays were used to determine the frequency of vaccinia-specific IFN-γ-producing cells in PBMCs obtained before vaccination and on days 3, 7, 14, and 28 after vaccination. No significant differences in baseline (prevaccination) vaccinia-specific IFN-g ELISPOT responses were detected (data not shown). ELISPOT responses occurred by day 14 in all vaccine recipients and by day 7 in some but not all vaccinianonnaive individuals, and the peak number of cells positive by IFN-γ ELISPOT occurred on day 14 after vaccination in both groups. In the 1 individual who failed to develop a major skin reaction, CMI responses were present but reduced, compared with those in subjects with a major skin reaction and IFN-γ ELISPOT response on days 14 and 28 (32 and 43 sfcs/10⁶ PBMCs, respectively) and for the group with positive skin reactions (geometric mean on days 14 and 28, 175 and 56 sfcs/ 10⁶ PBMCs, respectively).

Figure 1B demonstrates that the numbers of vaccinia-specific IFN-γ-producing T cells were significantly different (P = .03) between the vaccinia-naive and vaccinia-nonnaive individuals on day 14 after vaccination (peak response), with vaccinia-naive individuals having a greater response. The data also suggest that there are significantly fewer vaccinia-specific IFN-γ-producing T cells at peak response in vaccinia-nonnaive individuals than in vaccinia-naive individuals (table 2.) Four of the individuals in the vaccinia-nonnaive group had a detectable, positive IFN-g ELISPOT response on day 0. One vaccinia-nonnaive individual developed ELISPOT responses, despite the lack of skin reaction, on day 7 after vaccination. All of the vaccinia-naive and vaccinia- nonnaive individuals who had major skin reactions developed positive ELISPOT responses by day 14.

Induced vaccinia-specific CMI and antibody responses on day 14 after vaccination, among individuals developing a skin reaction, correlated with the size of the major skin-reaction lesion. The highest increase in CMI response occurred for CTL (1.48- fold increase/mm; P < .001) and ELISPOT (1.38-fold increase/ mm; P < .001) (table 3.) In this statistical regression model, the size of skin reaction, in the case of CTL or ELISPOT, did not change, regardless of baseline measure used, indicating that the model, despite the low R2 and the low numbers of individuals, accurately reflected a correlation more highly associated withCTL and ELISPOT responses (table 3.) For subjects with a skin reaction, the magnitude of the IFN-γ-producing T cell response by ELISPOT on day 14 was significantly higher in the vaccinianaive individuals than in the previously vaccinated individuals (P = .026). On the other hand, the day-7 responses were significantly higher in the vaccinia-nonnaive individuals (P = .005), indicating that the T cell memory response occurred earlier in vaccinia-nonnaive individuals but that the peak response was more robust in vaccinia-naive individuals.

Lymphoproliferation. On day 14, lymphoproliferation responses to vaccinia virus were induced in all individuals with a major skin reaction. The vaccinia-nonnaive individual who did not have a skin reaction also demonstrated positive lymphoproliferation response (SI = 9.3) on day 7. Vaccinia-naive individuals had higher lymphoproliferative responses than did vaccinia-nonnaive individuals (figure 1C). Prevaccination SIs after stimulation of cryopreserved PBMCs with live virus indicated no prior exposure in vaccinia-naive individuals; in contrast, 8 (42%) of 19 vaccinia-nonnaive individuals demonstrated lymphoproliferation responses to virus before vaccination (positive response p SI ⩾ 3), indicating a substantial memory response to virus. Day-3 and day-7 lymphoproliferation responses were significantly higher in the vaccinia-nonnaive individuals (P = .001 and P < .001, respectively), again indicating a more rapid CMI kinetic response to virus on revaccination 30 years after primary vaccination. Comparisons of CMI and skin reactions in individuals with prior memory responses, by use of lymphoproliferation (SI ⩾ 3), did not demonstrate any significant differences from those of the vaccinia-nonnaive individuals or trend toward accelerated or boosted responses.

Antibody response and CMI. Neutralizing antibody (NAb) and ELISA antibody responses for the present study have been reported elsewhere [17], but we summarize the NAb data here to illustrate the differences in kinetics of protective CMI and NAb responses between vaccinia-naive and vaccinia-nonnaive individuals (figure 1D). Vaccinia-nonnaive individuals who received the vaccine had the expected secondary antibody response, with 3 of 20, 17 of 20, and 18 of 20 individuals having a ⩾4-fold increase on days 7, 14, and 28 after vaccination, respectively. Positive NAb responses in vaccinia-naive individuals (titers of ⩾1:80 with baseline titers of ⩽8) were shown in 0 of 10, 7 of 10, and 9 of 10 vaccinia-naive individuals on days 7, 14, and 28 after vaccination, respectively.

Summary of CMI and antibody responses. Table 2 shows a summary of the geometric mean values obtained for each assay and the range of values seen across individuals. Individual differences, especially at the early time points, highlight the variation in memory responses seen in previously vaccinated individuals. Table 3 summarizes the differences seen after comparison of the 2 groups. Statistical analyses can be compared with data, as presented in figure 1.

Discussion

In response to infection or vaccination, CD8⁺ T cells undergo burst expansion followed by rapid contraction to memory cells [18]. The magnitude of this CD8⁺ T cell expansion determines memory levels [19, 20]. Previous studies of individuals after vaccination with vaccinia have shown that the initial IFN-γ T cell response is primarily CD8⁺ [7]. Live virus infections with vaccinia serve as a unique model of infection/immunization to study the kinetics of human T cell responses. In the present study, we have reported the first detailed analysis of CMI responses, comparing previously exposed individuals with those naive to vaccinia. Future analysis will need to focus on delineation of the kinetics and the cell types of specific early and later T cell memory populations after live virus vaccination. Such studies should lead to an improved understanding of the elements of achieving induction of long-term protective immune responses.

Our results are interesting from 2 perspectives. First, historically, protective immunity against smallpox, after smallpox vaccination, was presumed to be mediated by neutralizing antibody present in the serum of individuals. Recent studies have suggested that these antibodies are long-lived despite lack of continued antigen exposure [17, 21]. Our results (figure 1) demonstrate that T cell responses in vaccinia-nonnaive individuals coincide and, in some individuals, precede antibody responses by more than a week, suggesting that the induction of T cell memory may be requisite for protective immune response after reexposure to virus. In vaccinia-naive individuals, T cell immune responses appeared before the induction of neutralizing antibody. In addition, T cell responses occurred earlier in vaccinia-nonnaive individuals than in vaccinia-naive individuals (figure 1, day-7 responses), which is consistent with the presence of memory T cells despite the absence of detectable IFN-γ-producing T cells or CTL memory responses on day 0 (before vaccination) in some individuals and which, in our opinion, reinforces the notion that T cell-mediated immune events may be an essential component to eliciting protective immune responses. It is interesting that lymphoproliferation responses were still detected on day 0 in several of the vaccinia-nonnaive individuals but did not correlate with earlier induction of antibody response. However, the limited numbers of subjects in each arm of the present study could have obscured such a correlation.

Second, our results show that, on day 14, the magnitude of the initial burst response was higher in vaccinia-naive individuals than in vaccinia-nonnaive individuals, although the maximal response for vaccinia-nonnaive individuals could have occurred before day 14. The recovery to memory levels by day 28 suggests that the kinetics of long-term memory may be the same for both populations but that both the primary and the secondary memory response kinetics are likely to be 2-phased, with an initial burst phase followed by a second, gradually declining memory phase. Future studies assessing T cell subsets are likely to demonstrate differences in T cell populations in these 2 phases, as well as between vaccinia-naive and vaccinianonnaive individuals. In a previous study by our group, we assessed major skin reaction rates to several dilutions of Dryvax vaccine in vaccinia-naive individuals [9], and lack of major skin reaction was associated with failure to induce T cell responses. This same failure to induce significant T cell immune responses has been reported in the present study in 1 vaccinianonnaive individual, despite the presence of neutralizing antibody indicative of prior immunization. The importance of understanding the individual variation of virus-mediated effector and memory responses has implications for the design and strategies that will be used to test early potential efficacy of candidate vaccines [22].

The analysis of T cell responses after acute viral infection in humans has significant limitations, primarily related to the restriction of studying immune responses in the peripheral blood compartment. However, vaccination with live virus vaccines provides a unique environment to study immune responses to infection. Vaccinia virus inoculated by the skin-scarification technique enables study of immune responses in both the early phase of viral infection and the long-term memory phase. This model can provide new insights into the role of T cells in early viral infections, beyond those from current investigations using other viral models.

Characterization of human T cell responses after viral infection has been done after infection with EBV, HIV, cytomegalovirus (CMV), hepatitis B virus (HBV), and hepatitis C virus [23–29]. These studies have shown that T cells that recognize the specific infecting virus predominate in the blood compartment early after infection. The data presented in the present report delineate the time course of vaccinia-specific T cell responses after vaccination with live vaccinia virus in both vaccinia-naive individuals and individuals receiving secondary inoculation (revaccination). In other viral systems, early T cell phenotypes are characterized by the presence of CD8⁺ T cells that coexpress markers such as CD45RO, HLA-DR, CD27, and Ki67 [27]. A decrease in cell-surface markers, such as CCR7 and CD28 [30], during the course of acute infection suggests that phenotypic changes in the blood compartment can provide clues about components of protective immune response, and, when studied in more-controlled settings—such as clinical trials of live virus—vaccination could lead to an understanding of whether these phenotypic patterns translate to functional differences in humans.

In addition, early responding CD8⁺ T cells secrete IFN-γ in an antigen-specific fashion during infection with viruses such as EBV, CMV, and HBV [23, 28, 29]. Recently, highly specific epitope responses to vaccinia vaccination have been shown in vaccinia-naive individuals who received vaccinia, and the responses to an HLA-A2.01 epitope appear to show a biphasic response of virus-specific circulating T cells [31]. There is a very brisk increase during the first month after vaccination, followed by a decrease to a stable, long-lived memory-level response [31]. Data from the present study and accumulating data from other studies of smallpox vaccination suggest that the antigen exposure in the vaccinia skin-inoculation method may prolong the memory response, compared with that elicited by other nonlatent viruses. This should be considered when comparing data derived from human studies using live virus vaccination with data on both animals and humans derived from studies involving other methods of virus exposure. It is possible that the longer duration of antigen exposure following vaccinia inoculation may drive T cells into memory development and induce distinct phenotypes for resting memory, compared with latency-stage T cells. This speculation is supported by data from chronic infections that result in end-stage effector T cells that have diminished capacity to respond to viral loads induced during reactivation or reinfection [32]. T cells may be involved in maintaining latency, as well as influencing viral cell tropism in determining the development of latent infections, whereas the vaccinia model more closely mimics acute nonpersistent viral infection.

Our results demonstrate that, in both vaccinia-naive and vaccinia-nonnaive individuals, cell-mediated immunity precedes antibody responses. Previous epidemiological data on revaccination during smallpox outbreaks suggest that the initial 4 days after exposure is a crucial period, during which protective immune responses can be reestablished by vaccination. This period may be dominated by induction of memory cellular immune responses and, therefore, critical for the protective effects of vaccination with vaccinia after exposure to smallpox. Differences were observed in the kinetics of induction of CMI responses between vaccinia-naive and vaccinia-nonnaive individuals, which may also explain the epidemiological basis for less-protective responses in vaccinia-naive individuals during this postexposure vaccination period. These results point to the utility of vaccinia vaccination as a means to study human T cell memory responses to viruses and the elements of protective long-term immunity.