For study population 1, stored samples from the Pittsburgh Men’s Study were examined. MACS was begun in 1984 to characterize the natural history of HIV infection in the United States. All study participants were homosexual and bisexual men >18 years of age who were followed up every 6 months with physical examinations and collection of serum, plasma, and peripheral blood mononuclear cells. To be eligible for our analysis, participants had to have remained in the Pittsburgh Men’s Study for at least 4 years. A subset of 564 study participants was selected for this study.

For study population 2, to search for a correlation between MCV and hepatitis B virus (HBV) infection, a set of 200 samples from participants of a community study of HBV infection in Qidong, People’s Republic of China, was examined. Persons in this study (73 male and 127 female) ranged in age from 12–35 years (median age 30 years). Roughly equal numbers of serum samples were selected for hepatitis B surface antigen (HBsAg) positivity and negativity; samples were matched for participant age and gender. Written informed consent was obtained, and all procedures and protocols were approved by the Institutional Review Board of the University of Pittsburgh.

Blood samples at study entry were examined for MCV IgG positivity. For those participants initially seronegative, a sample from the end of the study (≈4 years later) was tested, and, if MCV seropositive, intervening samples were tested to determine the time of seroconversion. MCV seroconversion was defined as the midpoint between last negative and first positive serum sample. For many of these seroconverters, additional follow-up samples were available in the years after the formal end of our 4-year cohort study. A corresponding study visit was randomly chosen from 82 MCV negative controls for comparison. Specifically, the distribution of visits for MCV seroconversion was used: 10 at visit 2 (V2), 1 each at V3, V4, V8, and V10; 4 at V5; 3 at V6; 8 at V7; and 2 at V9 (total 31). Using a 4:1 match at V2 and a 2:1 match for all other seroconversion visits, we selected82 controls, and their corresponding visit data were compared with the data at the first MCV-seropositive visit for the MCV seroconverters. Thus, MCV seroconverters and controls were temporally matched so that long-term events (e.g., AIDS progression) over the course of the cohort study would not bias the results.

Virus-like particles were produced in human embryonic kidney 293TT cells (25) (a kind gift of Chris Buck) as previously described (26). Viral protein (VP) 1 and VP2 genes were designed according to a silent codon modification scheme (GenBank accession nos. FJ548568–FJ54871) (27) and synthesized by Blue Heron Biotechnology (Bothell, WA, USA) based on MCV339 (accession no. EU375804) (4). BKV VLP produced in a baculovirus system (28) was a kind gift of John T. Schiller.

Serum or plasma samples were tested at 1:100 dilution in a blinded and randomized fashion for MCV VLP reactivity by using Immulon HB2 plates (Thermo Fisher Scientific, Waltham, MA, USA) coated overnight at 4°C with MCV virus-like particles at 100 ng of the protein per well in phosphate-buffered saline (PBS) and blocked with 0.5% nonfat dry milk for 2 hours at room temperature. Paired analysis of serum samples and plasma showed no significant differences on MCV ELISA of 100 μL of serum, diluted with PBS/0.5% milk, added to wells, and incubated at room temperature for 2 hours. Anti-MCV antibody was detected by using horseradish peroxidase–conjugated rabbit anti-human IgG or anti-human IgM (Dako, Glostrup, Denmark) diluted 1:6,000 with PBS/0.5% milk (100 μL incubated for 1 hour). TMB (3.3′,5.5′-tetramethylbenzidine) substrate (Sigma, St. Louis, MO, USA) was used to detect absorbance signal at 405 nm with reference wavelengths of 620 nm after 45 minutes of incubation. Assay optimization by using MCV virus-like particles for protein saturation curves were performed as previously described (16). Each determination was performed in duplicate, and OD values were adjusted by background subtraction by using wells without antigen as previously described (29). A detailed protocol for this assay is available from www.tumorvirology.pitt.edu/mcvtools.html.

Cutoff values were based on results reported previously, which provides a detailed description (16). In brief, in this study, all samples with MCV IgG ELISA reactivity >0.2 OD units were found to have MCV-specific IgG specific for MCV virus-like particles but not heterologous polyomavirus virus-like particles by competition. Samples in the current study with MCV IgG OD values >0.2 units were considered positive without further testing. Previously, we also found human serum samples to have nonspecific reactivity up to 0.05 OD units that showed no specific competition. Thus, patient samples with MCV IgG ELISA <0.05 OD units in our current study were considered negative. Patient samples with MCV IgG ELISA results between 0.05 and 0.2 OD units were subjected to competition with MCV virus-like particles to determine if these titers were specific for MCV infection. If MCV reactivity for the sample was reduced by >50% by MCV virus-like particles competition, the sample was considered positive. Otherwise, the sample was considered negative for MCV antibodies.

Competition experiments were performed by mixing soluble MCV or BK virus–like particles with diluted serum or plasma (100 ng per 100 μL) in 1.5-mL microcentrifuge tubes followed by incubation for 1 hour at room temperature. After incubation, serum samples were directly added to MCV VLP–coated plates, and ELISA was performed as described above.

Testing for HBsAg and hepatitis B core antigen antibody (HBc) were performed by using commercial ELISA kits (Abazyme, Needham, MA, USA, and Abnova, Taipei, Taiwan). Serum samples were diluted 1:100 with PBS and tested in duplicate. Results of the tests were interpreted by using cutoffs determined with negative and positive controls provided by manufacturers.

All analyses were conducted by using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA, USA) and the Epi Info statistical calculator (wwwn.cdc.gov/epiinfo). Continuous data were analyzed by using a nonparametric Mann-Whitney test, and categorical data were analyzed by using a 2-sided Fisher exact test and by χ² tests for trend.

Study participants were 18–69 years old and were mostly white (96.2%). The mean age at baseline for participants was 32.8 years. Blood samples collected at entry were subjected to a battery of laboratory tests as described elsewhere (30,31). Patients were asked about symptoms during the preceding 6-month interval and evaluated for physical signs.

Of 564 MACS participants, at enrollment 447 (79.3%) were positive for MCV IgG (Figure 1). The median MCV ELISA value for positive samples was 0.313 (range 0.013–2.451 OD units). The remaining 117 (20.7%) participants provided specimens that were MCV-antibody negative. A weak but significant increasing trend for MCV seropositivity with age (p = 0.036, χ² test for trend) was found for young adults (25–35 years, n = 278) with a study entry prevalence of 74.8% that rose to 83.7% among men older than 45 years (n = 147) (Figure 2).

Comparing MCV-seropositive and seronegative men, we found no significant differences in HIV, hepatitis C virus, or syphilis status; blood count values; or other immunologic markers (Table 1). No correlation between MCV positivity and reported sexual activity was identified (data not shown). A significant association between MCV seropositivity and chronic hepatitis B virus infection was found at study entry: 31 (7.2%) of 447 MCV-seropositive participants were also positive for HBsAg as compared with 1 (0.8%) of 117 MCV-seronegative participants (odds ratio [OR] 8.1, 95% confidence interval [CI] 1.1–58.8). No significant association was found, however, between MCV and HBV exposure: 231 (51.7%) of 447 MCV-positive men, compared with 55 (47.0%) of 117 of MCV-negative men, were positive for HBc at study entry (OR 1.1, 95% CI 0.9–1.4). No other specific symptoms, signs, or blood test results were significantly associated with MCV positivity at entry, including rashes, diarrhea, fever, respiratory symptoms, or changes in total leukocyte count or cellular subpopulations (not shown).

To determine seroconversion, we tested 108 (92.3%) available plasma samples taken 4–5 years after study enrollment from the 117 initially MCV-seronegative persons. Of the 108 initially MCV-negative persons, 31 (5.5%) seroconverted (Figure 1), resulting in an incidence of 6.62/100 person-years.

MCV seroconversion at longitudinal follow-up visits was determined and infection was defined, for the purposes of this study, to occur at the midpoint between the last MCV-negative and the first MCV IgM- or IgG-positive blood sample. In general, MCV IgM peak levels preceded IgG seroconversion by 1 study visit or were concurrent with IgG seroconversion. Peak IgM levels (OD range 0.12–0.47), when present, were consistently lower than peak IgG levels (OD range 0.18–1.16). Once IgG seroconversion had occurred, none of the subsequent samples from these patients reverted to MCV IgG seronegativity.

Typical patterns for seroconversion are shown in Figure 3. Patient 1 shows the most common pattern, in which IgM peaks immediately before a rise in IgG levels, while the next most common pattern, exemplified by patient 2, shows a concurrent IgM peak with IgG seroconversion. For 6 seroconverters, rises in IgG were not accompanied by a preceding increase in IgM reactivity (patient 3), possibly due to transient IgM peaks that resolved between the 2 blood collections. Finally, for 3 seroconverters, an MCV IgM peak was detected 1–2 years before the rise in MCV IgG titers, which suggests a prolonged period of infection before Ig class switching (patient 4).

To determine signs or symptoms associated with primary MCV infection, the date of seroconversion was identified for each of the seroconverters. Signs, symptoms, and laboratory values reported at the first MCV seropositive visit were then determined. Corresponding study visits (selected as described in Methods) from MCV seronegative controls were then examined. Symptoms, such as fever, rash, weight loss, fever, diarrhea, and cough present at the study visit were similar for MCV seroconverters and the control group (Table 2). Self-reported symptoms lasting >2 weeks during the 6 months before the MCV seroconversion visit were also not significantly different between patients and controls. Moreover, no statistically significant differences in erythrocyte, leukocyte, CD4+, CD8+ cell counts, or other clinical test values were identified between the 2 groups at the first MCV seropositive visit for patients or at comparable visits for controls (Table 3). Although not significant, a greater proportion of the MCV seroconverters were hepatitis B core antibody positive (63.3% vs. 44.4%) and had a marginally lower hemoglobin level (15.5 vs. 16.0 g/dL) as compared with controls. No DNA of the virus was detected by real-time PCR in the plasma for 4 participants at the time of seroconversion (data not shown).

Plasma from 17 of the 31 MCV seroconverters with known dates of infection was available for long-term follow-up (ranging from 7–25 years). Two general patterns of anti-MCV antibody reactivity were seen: 11 (64.7%) patients demonstrated robust MCV seroconversion, with slowly increasing levels of MCV IgG over time (Figure 4). The second pattern (6 patients) revealed a transient increase in MCV IgG >0.2 OD units at the time of seroconversion, which declined over 1–2 years and generally remained <0.2 OD units. Despite this decline, all 6 retained readily detected MCV IgG as measured by MCV VLP competition assays.

Prevalent MCV seropositivity at study entry was only significantly associated with a positive test result for chronic HBsAg carriage. To examine the possible role of MCV infection in influencing chronic HBV infection, we tested 93 HBsAg-positive and 107 negative serum samples collected in a convenience sample from Qidong residents. Among these samples the prevalence of MCV antibody positivity was 66% with mean OD 0.452 (range 0.01–2.19). Rates of MCV positivity for HBsAg positive and negative samples were 61% and 70%, which were not statistically different (Table 4). Among the 107 HBsAg negative participants, MCV prevalence was not statistically significantly different for those exposed (HBc negative, n = 9) and not exposed (HBc negative, n = 84) to HBV infection (75% vs. 71%, respectively).