Antibody Responses to Simian Virus 40 T Antigen: A Case-Control Study of Non-Hodgkin Lymphoma Free

The etiology of non-Hodgkin lymphoma (NHL) is largely unknown. Simian Virus 40, a macaque polyomavirus, can cause leukemia and lymphoma in laboratory rodents (1). Two research groups recently described the detection of SV40 DNA in tumor tissue from ∼40% of U.S. NHL patients (2, 3). Some individuals might have acquired SV40 infection through receipt of poliovirus vaccines from 1955 to 1962, because SV40-contaminated poliovirus vaccines were used widely during this period (4). However, other groups subsequently failed to detect SV40 DNA in NHL tumors (5, 6), leaving the association uncertain. Similar difficulties have been encountered in studies of other tumors hypothesized to be related to SV40 (e.g., mesothelioma and brain tumors; ref. 4).

We recently conducted a U.S. population–based case-control study of SV40 infection and NHL (7). We tested subjects for antibodies to SV40 capsid using a virus-like particle (VLP) enzyme immunoassay (EIA; ref. 8). VLPs (empty capsids formed by spontaneous self-assembly of VP1, the major viral capsid protein) resemble native virions structurally and display B lymphocyte epitopes. In our study (7), only 7% to 10% of cases and 10% to 11% of controls exhibited antibody to SV40 capsid, and the responses were rather weak. In contrast, most individuals had measurable antibody responses to capsids of BK and JC viruses, the polyomaviruses commonly infecting humans. Indeed, based on competitive inhibition experiments, we concluded that the low-level SV40 antibodies in our subjects mostly represented cross-reactive responses against BK or JC virus. These results, and similar results from Spain (9), point to a lack of frequent SV40 infection in NHL patients and in the general population (10).

SV40 transformation and immortalization of cell lines is mediated by T antigen, a viral replication protein that inactivates cellular tumor suppressor proteins p53 and pRb. In animals with SV40-induced cancer, SV40 T antigen is expressed within tumor cells (1, 11-13). As a result, these animals frequently produce high titers of antibody directed against T antigen (i.e., “T antibody”; ref. 11). In contrast, SV40-infected animals without tumors show only infrequent and weaker T antibody responses (11).

We hypothesized that if SV40 causes NHL in humans, antibody against SV40 T antigen should be detectable. This should be true even if SV40 replicates only inefficiently in humans, conceivably leading to weak antibody responses against viral capsid. In the present analysis, we therefore extended our antibody testing of U.S. NHL cases and controls by measuring SV40 T antibody. Results in humans were compared with those in SV40-infected and SV40-uninfected animals, including hamsters with SV40-induced tumors.

The National Cancer Institute-Surveillance, Epidemiology, and End Results case-control study has previously been described (7). Briefly, HIV-uninfected subjects were enrolled between 1998 and 2000 from four U.S. areas: Iowa, and metropolitan Detroit, Los Angeles, and Seattle. Eligible cases with incident NHL (ages 20-74 years) were identified prospectively from the Surveillance, Epidemiology, and End Results registries. Eligible controls were identified using random digit dialing (ages 20-64 years) and Medicare eligibility files (65-74 years), sampled in strata defined by age, sex, race, and study area. The study was approved by institutional review boards at the National Cancer Institute and participating sites.

For the present analysis, we tested whether the frequency of measurable antibody against SV40 T antigen differed among cases and controls. To examine this question efficiently, we measured T antibodies only on a subset of subjects, sampled randomly from strata defined by SV40 VLP serostatus. Specifically, we included all SV40 VLP seropositive subjects and a random sample of SV40 VLP seronegative subjects, stratified by case-control status and BK VLP serostatus (Table 1). It would have been possible to sample subjects completely at random for T antibody testing. However, under the hypothesis that SV40 T antibody responses would be more frequent in SV40 VLP seropositive than in seronegative subjects, the enriched sampling of SV40 VLP seropositive subjects under this two-phase design has greater statistical power to detect an association between T antibody and NHL status (14). We also included sera from 19 SV40 VLP seropositive macaques and 8 SV40 VLP seronegative macaques. Finally, we simultaneously tested specimens from hamsters with tumors induced by s.c. inoculation of SV40-transformed cells (10 unique pooled serum specimens, each created from the sera of three to five animals) and five specimens from SV40-uninfected hamsters (including one pooled serum specimen).

EIA microtiter plates were sensitized with full-length, purified SV40 large T antigen (250 ng/well), produced in sf9 insect cells from a recombinant baculovirus construct (15). Using standard EIA methods, coated plates were incubated with serum samples (diluted 1:100), washed, incubated with peroxidase-conjugated goat antibodies against human immunoglobulin G (for human and macaque samples) or hamster immunoglobulin G, and treated with 2,2′-azino-di-(3-ethylbenzthiazoline-6-sulfonate) hydrogen peroxide solution to colorimetrically measure T antibodies. Samples were assayed in duplicate, and we utilized the mean absorbance measurement for each sample. Results on 17 human samples are missing due to low serum volume.

As an assay cutoff, we chose three standard deviations above the mean absorbance measurement for SV40 VLP seronegative macaques (i.e., 0.04 + 3 × 0.02 = 0.10). Assuming that assay results are normally distributed, this cutoff would classify 99.9% of SV40 VLP seronegative (presumably SV40-uninfected) macaques as T antibody seronegative (i.e., high specificity). Although it is unlikely that even weaker antibody responses measured could be considered seropositive, in sensitivity analyses, we evaluated lower cutoffs.

We fitted a logistic regression model to assess whether SV40 T antibody status was related to case-control status. We used methods developed by Breslow and Holubkov (16) to adjust for two-stage sampling. The odds ratio was additionally adjusted for sex, race, age, and study site.

Cases (n = 85) and controls (n = 95) tested for T antibody in the present analysis resembled the larger groups tested previously for SV40 capsid antibody (7). Among subjects in this analysis, 65% of cases and 51% of controls were male (P = 0.05). Cases and controls resembled each other with respect to age (mean 58 versus 59 years, respectively; P = 0.69) and with respect to race and study site (data not shown). NHLs were classified as follicular (25%), diffuse large B cell (26%), T cell (7%), or other/unknown (42%) subtypes. By site, 60% of NHLs were nodal and 40% were extranodal.

T antibody results are shown in Fig. 1. Hamsters with SV40-induced tumors all had readily measurable SV40 T antibody (median absorbance, 0.99; range, 0.73-1.48), whereas SV40-uninfected hamsters and SV40 VLP seronegative macaques uniformly gave substantially lower results (median absorbance, 0.05; range, 0.03-0.07; and median absorbance, 0.04; range, 0.01-0.08, respectively). NHL cases, controls, and SV40 VLP seropositive macaques resembled these latter two groups, generally showing very low measurable SV40 T antibody (median absorbance, 0.03 0.04, and 0.04, respectively; Fig. 1).

Overall, five cases and five controls (6% versus 5%; Table 1) and two SV40-infected macaques (11%) had T antibody responses classified as seropositive. T antibody responses in humans and macaques were all weaker than seen in hamsters with tumors (Fig. 1). SV40 T antibody seropositivity was unrelated to NHL (adjusted odds ratio, 1.2; 95% confidence interval, 0.3-4.6). SV40 T antibody was detected in cases with varied histologic types of NHL and sites of disease (Table 2). Two T antibody seropositive cases had cutaneous T cell NHL, but only one of these also had SV40 VLP antibody (Table 1); four other T cell NHLs were T antibody seronegative.

Among our controls, we explored whether SV40 T antibody was associated with VLP antibody results, although small numbers prevented strong conclusions (Table 1). SV40 T antibody was present in 6% of SV40 VLP seropositive controls and 3% of SV40 VLP seronegative controls. Interestingly, SV40 T antibody was detected only in BK VLP seropositive controls (7%), a finding that was mirrored in the cases (8%). T antibody was detected in 3 of 69 (4%) JC VLP seropositive controls and 2 of 26 (8%) JC VLP seronegative controls.

Lowering the cutoff for the T antibody EIA classified more subjects as seropositive, but fewer cases than controls were classified as such. With a cutoff of 0.08 absorbance units, 8% of cases and 13% of controls were considered T antibody seropositive. With a cutoff of 0.06 absorbance units, 18% of cases and 24% of controls were considered T antibody seropositive.

We observed a lack of antibody to SV40 T antigen among humans, including individuals with NHL. In animals with SV40-induced tumors, T antigen is present within each tumor cell (1, 12, 13). T antigen expression in tumors leads to a robust antibody response against this oncoprotein, as we showed for hamsters with SV40-induced tumors and as shown previously (11). Even when we lowered our assay's cutoff to increase its sensitivity, T antibody seroprevalence in NHL cases was not higher than in controls.

To our knowledge, only one prior investigation systematically evaluated cancer patients for SV40 T antibody. Shah (17) studied 975 individuals from northern India, where contact with indigenous SV40-infected rhesus might lead to SV40 infections in humans. The study included 26 patients with malignancy of the lymphatic system, 16 with “reticulum cell sarcoma” (now classified as NHL), and 14 with leukemia. Only 5% of cancer patients overall (4-7% of patients with hematologic malignancies) showed SV40 neutralizing antibody, generally at low titer. Notably, no patient had SV40 T antibody as measured by an immunofluorescence assay.

In contrast, investigations that have used sensitive PCR methods to test for SV40 DNA in NHL tumor tissue have produced highly variable results. Studies from the U.S., Europe, and Japan have reported detection rates ranging from 0% to 43% (2, 3, 5, 6, 18-21). Geographic differences across study populations might be relevant, but there were also differences in methods among laboratories, such as the choice of PCR primers and reaction conditions. Some PCR-based studies reporting positive results suffered from methodologic problems, including a lack of masked evaluation of specimens (2, 3, 5, 18-21), lack of appropriate negative controls (5, 18-21), and the amplification of SV40 sequences from negative control tissues (2). Furthermore, the amount of SV40 DNA in PCR-positive specimens seemed quite low (19, 21). Because SV40 is present within each cell in animal lymphomas (1), the reported detection of SV40 DNA at extremely low copy numbers in human tumors could represent a laboratory artifact, e.g., PCR contamination.

Furthermore, there is no epidemiologic evidence linking SV40 to NHL in humans. Between 1955 and 1962, millions of individuals in the U.S. and Europe were exposed to SV40 via contaminated poliovirus vaccines (4). Nonetheless, individuals exposed to SV40-contaminated poliovirus vaccines do not have an elevated NHL risk (22). SV40 infection routes other than vaccination have not been shown. Unlike BK and JC viruses, SV40 does not seem to be shed in human urine nor has it been detected in sewage (23, 24), arguing against person-to-person transmission and frequent infection.

We found SV40 T antibody to be a sensitive and specific marker for SV40-induced malignancy. All five evaluated sera from animals with SV40-related tumors manifested robust T antibody levels using our EIA. In animals, the frequency of T antibody detection and antibody titer increase with tumor size (11, 13). Because NHL is a systemic disease, generally with a large tumor burden at the time of clinical presentation, we would expect our test to have high sensitivity for detecting SV40-related NHLs in humans. Conversely, because only 5% of human controls in our study had measurable T antibody responses, the EIA likely has at least 95% specificity. Interestingly, all T antibody seropositive humans were BK VLP seropositive. This finding could indicate that some low-level SV40 T antibody responses in humans actually represent cross-reacting antibody responses to BK virus T antigen (25).

A limitation of our study was that we did not have archived tumor tissue available, which might have allowed us to compare the SV40 DNA status of tumors with these antibody results. However, methods for SV40 DNA detection are not standardized (4) and, as noted above, issues remain regarding the biological interpretation of PCR results. A further limitation is our small sample size with associated wide confidence limits, although our overall results provide little justification for evaluating a larger collection of NHL samples.

In conclusion, the present study provides evidence that SV40 is not a major cause of NHL in the U.S. Our subjects only rarely had antibody to SV40 T antigen. The frequency of these antibodies did not differ between NHL cases and controls, and the low antibody levels are most consistent with cross-reactivity to BK virus T antigen. Additional serologic studies of patients with NHL and other tumors possibly related to SV40 will be valuable (4).

We thank Ronald Kennedy and Michael Shearer (Texas Tech University Health Science Center, Lubbock, TX) for providing recombinant SV40 T antigen, and Barbara Clayman (Johns Hopkins University) for conducting the serology testing. We gratefully acknowledge the contributions of the study participants and the staff at the SEER centers of Detroit, Iowa, Los Angeles, and Seattle. Supports for this study included the following contracts with the National Cancer Institute: N01-PC-67010, N01-PC-67008, N02-PC-71105, N01-PC-67009, and N01-PC-65064.