Adults previously characterized as having suspected, probable, or confirmed cases of monkeypox during the 2003 Wisconsin monkeypox outbreak provided informed written consent and completed a medical history questionnaire before participation in our previous study (6). Monkeypox confirmation was based on the standard case definition and confirmation by the Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia, USA, during the outbreak (4) or by subsequent immunologic tests (6). Participants were grouped according to vaccination history (Table). Samples from 13 unvaccinated monkeypox-immune persons and 8 vaccinated monkeypox-immune persons were assessed. Of the 8 vaccinated persons, 3 had clinically inapparent cases and were not aware of having been infected with monkeypox because of no noticeable disease symptoms (6). However, these are described as laboratory-confirmed cases of monkeypox on the basis of multiple immunologic assays (6) (Table). Control participants included recent primary smallpox vaccinees (n = 10), revaccinated persons (n = 10) examined 2–4 months postvaccination, long-term immune smallpox vaccinees examined 20–40 years postvaccination (n = 10), and unvaccinated orthopoxvirus-naive persons (n = 12). Heparinized blood was centrifuged over Histopaque-1077 (Sigma, St. Louis, MO, USA), and plasma was collected and stored at –80°C. Studies involving human participants were reviewed and approved by the Institutional Review Board for Oregon Health and Science University.

Monkeypox virus (Zaire strain) and vaccinia virus (Western Reserve strain) were grown in BSC40 cells by using a multiplicity of infection of 0.1 and harvested at 48 h postinfection. Cells were lysed by 3 freeze/thaw cycles in 10 mmol/L Tris, pH 8.0, and used as a virus lysate for preadsorption in ELISA and Western blots. Where indicated, monkeypox and vaccinia viruses were also purified by ultracentrifugation through 36% sucrose at 40,000 × g for 80 min, followed by band extraction after 25%–40% sucrose gradient sedimentation (33,000 × g for 40 min). Protein concentration was measured by the modified Lowry assay (Pierce, Rockford, IL, USA). Purified virus was inactivated for 2 h at 56°C; viral lysates were inactivated with 3% H₂O₂ at room temperature for 2 h. Uninfected BSC40 cell lysate was treated similarly to the infected cells described above.

High protein-binding ELISA plates (Corning-Costar, Corning, NY, USA) were coated with an optimized concentration of H₂O₂-inactived monkeypox-infected BSC40 cell lysate. Plasma samples were preadsorbed with equivalent (6 ×10⁸ PFU/mL) concentrations of H₂O₂-inactived monkeypox or vaccinia whole-cell lysate at a 1:30 dilution (5 μL plasma in 145 μL viral lysate) for 30 min at 37°C. Nonadsorbed samples were similarly treated with ELISA blocking buffer (phosphate-buffered saline containing 5% nonfat dry milk and 0.05% Tween 20). Samples were then added directly to ELISA plates, serially diluted in blocking buffer, and incubated at room temperature for 1 h. As a precaution against human blood-borne pathogens, samples were then treated with 3% H₂O₂ (final concentration) for an additional 30 min. After washing, horseradish peroxidase–conjugated mouse antihuman immunoglobulin (Ig) G monoclonal antibody (clone G18–145; BD Pharmingen, San Diego, CA, USA) was added to the wells. Plates were washed after 1 h and detection reagents were added. Substrate was prepared (o-phenylenediamine; Sigma-Aldrich, St. Louis, MO, USA) and diluted to a concentration of 0.4 mg/mL in 0.05 M citrate, pH 5.0, and H₂O₂ was added (final concentration of 0.01%). Color was developed for at least 20 min before the reaction was stopped by adding 1 M HCl; plates were read at 490 nm.

Antibody titers were determined by log-log transformation of the linear portion of the dilution curve with 0.1 optical density units used as the endpoint, and transformation was performed on final values (6). For an Excel (Microsoft, Redmond, WA, USA) file containing a template of these calculations, please contact the corresponding author, or see the example given in the online Technical Appendix (available from www.cdc.gov/EID/content/14/4/592-Techapp.pdf).

Western blot procedures were performed with the following modifications. Two micrograms of gradient-purified monkeypox or vaccinia virus was separated by 4%–20% Tris-glycine gradient sodium dodecyl sulfate–polyacrylamide electrophoresis (SDS-PAGE) (Invitrogen, Carlsbad, CA, USA) under reducing conditions. Equivalent protein loading was confirmed on representative gels by staining with GelCode Blue (Pierce).

Proteins were electrophoretically transferred to polyvinylidene difluoride membranes (Pierce), and membrane strips with 3 lanes containing a molecular mass standard (SeeBlue Plus 2; Invitrogen), monkeypox, and vaccinia were blocked with phosphate-buffered saline containing 1% Tween 20 and 5% nonfat dry milk. Plasma was diluted 1:20 in uninfected cell lysate or H₂O₂-inactived vaccinia lysate (adjusted to a concentration of 2.5 mg/mL total protein) for 30 min at 37°C. Adsorbed plasma was adjusted to a 1:10,000 dilution in 10 mL of blocking buffer and incubated with membranes overnight in 50-mL conical tubes at 4°C with rocking. After 3 washes in blocking buffer, reactive bands were identified with horseradish peroxidase–conjugated goat antihuman IgG (γ chain specific; Jackson ImmunoResearch, West Grove, PA, USA) by using chemiluminescent detection (SuperSignal West Dura Substrate; Pierce). Plasma from the same monkeypox patient was used as a positive control in each experiment to identify the position of diagnostic bands. Blots were exposed to x-ray film until diagnostic bands were clearly visible, and other films were then overexposed to ensure that any low-intensity bands were given ample opportunity to appear. Plasma from some orthopoxvirus-naive persons did not react with any monkeypox or vaccinia protein bands. In these instances, films were exposed 10× longer than the last readable positive control exposure before a negative result was recorded. Films were scanned and the positions of diagnostic bands were indicated on the basis of the positive control. Analysts scored the vaccinia-preadsorbed Western blots for the 39-kDa, 124-kDa, and 148-kDa diagnostic bands as present only in the monkeypox lane, absent from the monkeypox lane, present in the monkeypox and vaccinia lanes (i.e., experimental equivocal), or technical equivocal caused by nonspecific background. Immunoreactive bands deemed experimental equivocal were counted against the sensitivity or specificity of the assay. Blots containing technical equivocal data were repeated, and analysts rescored bands that were unreadable in the initial screen before determining final sensitivity and specificity.

Orthopoxviruses have highly conserved genomes (16), which results in high levels of antibody cross-reactivity. However, viruses such as monkeypox also contain genes that are absent, mutated, or truncated in vaccinia; these gene products can be used to distinguish between monkeypox and vaccinia infections (6,17–21). The concentration of viral antigen used in the preadsorption step of the ELISA was determined by using high-titer plasma from 2 recently vaccinated vaccinia-immune persons obtained at the peak (day 21) of the anamnestic response with ELISA titers that were ≈10-fold higher than the highest convalescent-phase samples used in the rest of the experiments. Two of the highest titer samples from monkeypox-immune persons (obtained at 2 months postinfection) were also used in these preliminary studies online Technical Appendix (available from www.cdc.gov/EID/content/14/4/592-Techapp.pdf). These samples provided a rigorous test of homologous and heterologous antigens to deplete antiviral antibodies before performing the ELISA. Titration of virus antigen indicated that lysates normalized to contain 6 × 10⁸ PFU equivalents/mL were best for differentiating between vaccinia and monkeypox infections.

Orthopoxvirus-naive persons or persons infected with monkeypox or vaccinia were then tested to establish the diagnostic validity of this approach by using monkeypox-coated ELISA plates (Figure 1, panel A). If plasma from a representative monkeypox-immune person was preincubated with vaccinia antigen, virus-specific antibody titers decreased (from 39,904 ELISA units [EU] to 13,912 EU). However, if the sample was preincubated with an equivalent amount of monkeypox antigen before performing the ELISA, monkeypox antibody titers were further decreased, to 2,542 EU. For the data of the monkeypox-immune person shown in Figure 1, panel A, there was a 5.5-fold difference (13,912 EU divided by 2,542 EU) in the ability of monkeypox antigen to reduce monkeypox-specific antibody levels than vaccinia antigen. For persons who were vaccinia immune but still had contracted clinically apparent monkeypox disease, the difference in postadsorption ELISA titers was typically smaller (3.2-fold difference in the example shown in Figure 1, panel A), but the difference still demonstrated a clear distinction between preadsorption with monkeypox or vaccinia antigens. If a person had been infected with only vaccinia or revaccinated with vaccinia, there was essentially no difference in the ability of monkeypox antigen to deplete antiviral antibodies than an equivalent amount of vaccinia antigen on a monkeypox-coated ELISA plate. In contrast, orthopoxvirus-naive persons were seronegative by ELISA and can be easily distinguished from monkeypox-immune or vaccinia-immune persons because they score below the limits of detection by ELISA (<100 EU) without any preadsorption steps required.

To determine whether the difference in postadsorption ELISA results could be used effectively to distinguish between vaccinia and monkeypox infections, we plotted the results from primary monkeypox-immune persons, previously vaccinated monkeypox immune persons, primary vaccinia-immune persons, and revaccinated vaccinia-immune persons (Figure 1, panel B). We established a diagnostic cutoff of 2.5 by plotting the difference in postadsorption ELISA titers for each group. To establish the diagnostic cutoff, we used persons with laboratory-confirmed cases of primary monkeypox (6) as positive controls and vaccinia-immune persons with no evidence of monkeypox infection as negative controls (Figure 1, panel B). The 2.5-fold difference in postadsorption ELISA titers provided the best balance of high sensitivity (100% positive for primary monkeypox) and high specificity (0% positive for vaccinia immune). Other diagnostic cutoff values resulted in decreased specificity or sensitivity when these 2 divergent groups of positive and negative controls were compared. For example, a diagnostic cutoff of 3.0 reduced the sensitivity of the assay to 93% (12/13) for primary monkeypox but did not affect the specificity (100%). Conversely, a diagnostic cutoff of 2.0 did not affect sensitivity (100%) but decreased specificity from 100% to 90% (i.e., 18/20 vaccinia-immune persons were positive). Vaccinia-immune monkeypox patients typically had lower differences in postadsorption ELISA titers. The lower difference suggests that preexisting immunity to vaccinia may have resulted in a lower induction of monkeypox species-specific antibody responses and is consistent with the results of our previous study (6).

Using the established 2.5-fold diagnostic cutoff, we achieved 100% (13/13) sensitivity for detecting primary monkeypox infection and 63% (5/8) sensitivity for detecting monkeypox infection in vaccinia-immune persons in which monkeypox was a heterologous orthopoxvirus infection. Of the 8 vaccinia-immune persons with monkeypox, 3 were clinically asymptomatic but were previously identified by serologic and cellular techniques (6). However, these cases are more difficult to diagnose on the basis of serologic analysis only (6) and could not be identified by the postadsorption ELISA (Figure 1, panel B; Table). However, if only persons with clinically overt monkeypox infection were included in the analysis, 100% (5/5) of secondary monkeypox infections were identified by this assay (Figure 1, panel B; Table). We also observed 100% specificity; 20/20 vaccinia-immune controls (10 primary and 10 booster smallpox vaccinations) and 12/12 orthopoxvirus-naive persons were negative by this postadsorption ELISA. We report 86% (18/21) overall sensitivity for detecting monkeypox, 100% (18/18) sensitivity for confirming clinically overt monkeypox, and 100% (32/32) specificity with this approach.

To more easily recognize uniquely reactive monkeypox-specific bands by Western blot and simplify interpretation of this serologic diagnostic technique, we added an adsorption step to reduce cross-reactive antibodies to orthopoxvirus. Separation of 2 μg of purified vaccinia and monkeypox by 4%–20% gradient SDS-PAGE resulted in good separation of protein bands across a broad spectrum of molecular masses (Figure 2, panel A) and effective transfer of separated proteins to the polyvinylidene difluoride membrane. Protein-banding patterns of monkeypox virus and vaccinia virus are similar, but a protein band of ≈39 kDa is visible in the monkeypox lane of the GelCode Blue-stained SDS-PAGE that is missing from the vaccinia lane. Incubation of immune plasma in a 20-fold volume excess of virus antigen (1 μL plasma plus 19 μL vaccinia-infected cell lysate at 6 × 10⁸ PFU equivalents/mL) before performing the Western blot effectively reduced the intensity and detection of cross-reactive bands (Figure 2, panel B).

Diagnostic bands with apparent molecular masses of 148, 124, and 39 kDa were identified when monkeypox-immune plasma was used, but were not observed after Western blot analysis with plasma samples from a representative vaccinia-immune person (Figure 2, panel C) or an orthopoxvirus-naive person (Figure 2 , panel D). Because orthopoxvirus-naive persons were seronegative by ELISA (Figure 1, panel A), the Western blot example shown in Figure 2, panel D, required a 1-hour exposure to identify any faint bands after vaccinia antigen adsorption. In contrast, samples from monkeypox-immune and vaccinia-immune persons were exposed to film for 5 s to 2 min for optimal identification of virus-specific banding patterns. We did not observe a correlation between viral protein levels determined by SDS-PAGE (Figure 2, panel A) and immunodominance by Western blot analysis (Figure 2, panel B).

Unblinded analysis of the Western blots was first performed by 2 independent analysts who had access to patients’ medical histories (Figure 3, panel A). Reactivity to the 39-kDa band resulted in 100% sensitivity for identifying primary monkeypox infection and 75% sensitivity for identifying secondary monkeypox infection. However, 20%–30% of the negative control samples (vaccinia primary, vaccinia long-term, and orthopoxvirus naive) also reacted with this band, which resulted in a specificity of only 70%–80% (76% overall specificity). The 124-kDa band showed 96% sensitivity for identifying primary monkeypox infection but only 31% sensitivity for identifying secondary monkeypox infection. However, this protein was more specific than the 39-kDa protein; 0%–10% of negative controls showed reactivity (97% overall specificity). The 148-kDa band showed low diagnostic sensitivity; only 37%–46% of plasma samples from monkeypox-immune persons reacted with it in Western blot analysis. Conversely, the 148-kDa band showed high specificity with 0%–5% of negative control plasma that were positive (i.e., overall specificity >97%). Other bands were also identified as potential diagnostic indicators of monkeypox infection, such as the 98-kDa and 18-kDa bands (Figure 2, panels B and C; online Technical Appendix, http://www.cdc.gov/EID/content/14/4/592-Techapp.pdf). However, further analysis showed that the 98-kDa band had low diagnostic potential (48% sensitivity and 72% specificity), as did the 18-kDa band shown in Figure 2, panel B (60% sensitivity and 66% specificity).

Blinded analysts (n = 4) with no knowledge of infection history were asked to score Western blots to determine the feasibility of this approach under conditions in which background clinical information may not be available (Figure 3, panel B). Unblinded analysts reported somewhat higher sensitivity for each of the diagnostic bands and identified the same or slightly lower specificity when interpreting Western blots with vaccinia-immune or orthopoxvirus-naive plasma samples, but the overall results from blinded and unblinded analysis were similar.