INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Epstein-Barr virus (EBV), a gammaherpesvirus widespread in human populations, is usually carried as a lifelong asymptomatic infection yet is etiologically linked to a number of different malignancies (reviewed in reference 37). It was the study of one such malignancy, the endemic (African) form of Burkitt's lymphoma (BL), which first identified two distinct types of EBV isolates (1, 9), originally called A and B and now designated types 1 and 2. These are essentially homologous across large stretches of the genome but are distinguished by linked polymorphisms in the latent genes encoding the nuclear antigens EBNA2, -3A, -3B, and -3C (1, 9, 41). The virus type can therefore be determined at the DNA level by PCR amplification across these polymorphic loci (2, 41, 47) and at the protein level by using monoclonal antibodies (MAbs) or human sera with type-specific reactivities against these antigens (40, 45). Furthermore, within each broad type, individual virus strains can be distinguished by screening across other defined polymorphisms in the viral genome (4, 30, 34, 35) and by determining the strain-specific "EBNAprint," i.e., the precise sizes of the virus-coded EBNA1, -2, -3A, -3B, and -3C proteins as visualized in immunoblots (19, 53). Studies on endemic BL tumors and subsequently on virus isolates rescued as in vitro-transformed lymphoblastoid cell lines (LCLs) from healthy donors suggested that type 1 and type 2 strains were of roughly equal prevalence in at least some parts of equatorial Africa (57). By contrast, the much rarer cases of EBV-positive sporadic BL occurring in children in the Western world almost all carried a type 1 virus (14, 58). This again appeared to reflect the situation within the general population in these areas, where, based on virus isolations, type 1 strains were prevalent and <10% of individuals carried type 2 virus (20, 53). The balance of evidence from such work also strongly suggested that healthy virus carriers harbored a single virus strain; multiple infection either with different virus types or with different strains of the same type appeared to be extremely rare (20, 53; reviewed in reference 17).

The generality of these conclusions has subsequently been challenged by observations of AIDS patients in Western communities. Such patients show an unusually high risk both of Burkitt-like lymphoma, arising relatively early in the course of AIDS and (like sporadic BL) EBV genome positive in roughly 30% of cases (22, 48), and of immunoblastic lymphoma arising in late-stage AIDS and (like posttransplant lymphoma) EBV genome positive in the majority if not all cases (22, 32). Unexpectedly, both of these AIDS-associated lymphomas and also a third EBV-associated malignancy seen in this patient group, Hodgkin's disease, were found to carry a type 2 EBV strain in 25 to 50% of EBV genome-positive cases (6, 10, 14, 21, 36, 46). These observations, and an earlier report detecting coresident EBV genotypes in an AIDS lymphoma patient (26), prompted several studies of EBV infection in HIV-positive patients as a whole, analyzing the types of virus that were detectable in the oropharynx and circulating B-cell pool either directly by PCR analysis or by virus rescue in vitro (7, 29, 31, 44, 50, 51, 54, 55). This work clearly showed that the prevalence of type 2 virus infection in such patients was at least 30%, substantially higher than had been apparent from most studies on the general population from which these patients were drawn. Interestingly, detailed analysis of such in vitro isolates revealed that most of the HIV-positive patients with detectable type 2 EBV also carried a coresident type 1 strain, while another 25% of the patients carried multiple viruses, all of type 1 (54).

The full biological significance of these findings is still not known. One interpretation is that the EBV carrier state in HIV-positive patients faithfully magnifies that which exists generally in the immunocompetent population and that the full range of resident strains are simply easier to detect in this immunocompromised setting because of the higher overall viral load. If that were the case, earlier studies (20, 53) must have seriously underestimated the true prevalence of type 2 virus in the general Caucasian population (3, 47); this is possible in that type 2 strains may be subdominant in vivo and are rescued less efficiently than type 1 strains in in vitro transformation assays (38). In seeking to address these issues, we noted at the outset that almost all of the work to date on HIV-positive patients, including our own recent analysis (54, 55), had been confined to male homosexual cohorts. We reasoned that if the above-described arguments were true, then a similar pattern of virus isolations should be obtained from other T-cell-immunocompromised patient groups. In the present work we have used the same methods as employed with the HIV-positive homosexual cohort (54) to screen an equally large cohort of hemophilic patients who, over a decade earlier, had acquired HIV iatrogenically from contaminated Factor VIII concentrates.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Patients. The study group comprised 48 HIV-positive hemophilic patients. These included 31 adult males (mean age, 39 years; range, 22 to 71) attending outpatient clinics at the Queen Elizabeth Hospital, Birmingham (designated QEH), or at Churchill Hospital, Oxford (OX), and 17 adolescent males (mean age, 16; range, 13 to 20) attending outpatient clinics at the Birmingham Children's Hospital (BCH). All patients had been infected with HIV in the early 1980s from contaminated batches of factor VIII concentrate; they were studied between 1993 and 1996, some 10 to 15 years following HIV infection, at a time when almost all were significantly immunocompromised and showed depressed CD4⁺ T-cell counts (median count, 150/mm³; range, 0 to 820). A control group consisted of 38 HIV-negative adolescent hemophilic patients, again sampled via outpatient clinics at Birmingham Children's Hospital and spanning the same age range as the 17 HIV-infected adolescent patients described above. Heparinized blood (20 to 30 ml) and throat washings were taken and processed as described previously (54), with most patients being sampled on two separate occasions over the 3-year study period. Blood samples provided plasma for serological studies and mononuclear cells for virus isolation and in vitro regression assays, while throat washings provided cell-free filtrates for virus isolation. Data thus obtained from the HIV-positive and HIV-negative hemophilic patient groups are compared with corresponding results from 40 HIV-positive male homosexuals studied in parallel (54) and from 56 HIV-negative adult control donors analyzed in an earlier study (53).

Immunological assays. Plasma preparations were first heat inactivated at 56°C for 30 min and then stored at 20°C until testing for immunoglobulin G antibody titers to the EBV capsid antigen (VCA) by standard immunofluorescence assay (53). When yields of peripheral blood mononuclear cells (PBM) from EBV-seropositive patients were sufficiently high, an aliquot of cells was screened for in vitro regression of B95.8 strain EBV-induced transformation to provide a measure of EBV-specific cytotoxic T-lymphocyte (CTL) memory function; results were expressed as the minimum seeding of PBM per 0.3-ml microtest plate well required to observe a 50% incidence of regression in replicate wells, as described previously (35). To allow comparison between patient groups, individual regression titer end points were classified as being within the normal range for most healthy EBV carriers (<4.5 × 10⁵ PBM/well), weak compared to those for most healthy carriers (4.5 × 10⁵ to 6 × 10⁵ PBM/well), or undetectable (>6 × 10⁵ PBM/well).

Virus isolation. The methods for isolation of virus from blood and from throat washings were exactly those used in the parallel study of HIV-positive male homosexuals (54). Briefly, PBM were cultured in 0.3-ml microtest plate wells at a range of cell dilutions (six replicates/seeding) in RPMI 1640 medium and 10% (vol/vol) fetal calf serum, initially supplemented with 0.1 µg of cyclosporin A per ml to inhibit T-cell activation, and individual wells developing spontaneously transformed foci of outgrowth were expanded to provide LCLs carrying the patient's endogenous EBV isolate(s). Throat washings were similarly assayed for transforming virus by using PBM from cord blood or from adult EBV-seronegative donors as a source of indicator cells. Note that these transformation assays were each conducted over a 12-week period; all wells which developed microscopically visible foci within that time were expanded irrespective of growth rate; hence, the time from initial seeding to LCL cryostorage could be up to 6 months. Occasionally wells developed transformed foci which failed to expand, and in such cases the culture was used to provide a DNA sample only.

Identification of virus isolates. The methods of genomic and EBNAprint analyses were essentially as described in earlier work (28, 53, 54). Briefly, DNA from LCL preparations was subjected to PCR amplification with primer-probe combinations across type-specific polymorphisms within the EBNA2 and EBNA3C genes (41), across the 33-bp repeat region of the latent membrane protein 1 (LMP1) gene (34), and across a locus in LMP1 which in some virus strains displays a 30-bp deletion relative to the B95.8 prototype (34). Reference EBV strains included in the genomic analysis were B95.8 (type 1, 4.5 copies of the 33-bp LMP1 repeat, undeleted at the 30-bp LMP1 locus) and AG876 (type 2, 4 copies of the 33-bp LMP1 repeat, deleted at the 30-bp LMP1 locus). Note that although it is possible for heterogeneity across the 33-bp repeat region to be generated during replication of a single EBV strain (49), in practice we found that the great majority of LCL clones established by B-cell transformation with a reference laboratory strain retained the same number of repeats as shown by the cell lines from which the reference virus was prepared (53a).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

EBV immune status of hemophilic patients. We first set out to determine the status of the HIV-positive hemophilic patients with respect to prior EBV infection, by using anti-VCA antibodies as a serological indicator, and to measure the prevailing level of EBV-specific T-cell immunity in EBV-infected patients, by using the in vitro regression assay. For this part of the study, the 17 adolescent patients are considered separately from the rest of the cohort so as to allow their direct comparison with an adolescent control group of 38 HIV-negative hemophilic patients. As shown in Table 1, roughly two-thirds of these adolescent patients had a history of EBV infection, both in the HIV-positive cohort and in the HIV-negative control group. However anti-VCA titers were fivefold higher in HIV-positive patients than in HIV-negative patients, an elevation consistent with that reported for other T-cell-immunocompromised cohorts and probably reflecting increased EBV antigenic load in vivo (52). Furthermore, the level of EBV-specific T-cell immunity was markedly reduced in the HIV-positive patients, with six of the seven individuals analyzed failing to register an end point in the in vitro regression assay; by contrast, regression end points for EBV-infected hemophilic patients in the HIV-negative control group were mostly within the normal range.

EBV isolations from HIV-positive hemophilic patients. Virus isolates were prepared from the blood and/or throat washings of all 39 HIV-positive hemophilic patients (11 adolescent and 28 adult) who had serological evidence of prior EBV infection. A mean of 19 independent EBV isolates (range, 4 to 39) per patient was obtained. The different patterns of results are described below.

(i) Patients with a single detectable EBV strain. In 25 of the 39 patients studied, all EBV isolates rescued from a single individual were of a single unique strain. These findings are illustrated here with reference to BCH10, a patient from whom a total of 34 independent EBV isolates (14 from blood and 20 from throat washings) were rescued in vitro from two occasions of testing 2 years apart. Figure 1 presents the results of EBV genomic analysis of four blood-derived isolates (isolates 1 to 4) and seven throat washing-derived isolates (isolates 1* to 7*); these are representative of the data from all 34 isolates from this patient. Every virus isolate gave a type 1-specific PCR amplification product at both the EBNA2 and EBNA3C polymorphic loci, and there was no evidence of any coresident type 2 virus. Furthermore, every one of these type 1 isolates was identical at two well-characterized polymorphisms (the 33-bp repeat and 30-bp deletion loci) within the LMP1 gene sequence; thus, as shown in Fig. 1, each isolate had six copies of the 33-bp repeat and displayed a 30-bp deletion. Note that this and all ensuing PCR assays included as reference virus isolates the prototype 1 B95.8 strain (4.5 copies of the 33-bp repeat and undeleted at the 30-bp locus) and the prototype 2 AG876 strain (4 copies of the 33-bp repeat and deleted at the 30-bp locus) (34).

View larger version (66K): [in this window] [in a new window]   FIG. 1.   PCR analysis of EBV strains carried by blood-derived LCLs 1 to 4 and throat washing-derived LCLs 1* to 7* from patient BCH10. For virus typing, EBNA2 gene amplification was carried out with a common 5' primer and type-specific 3' primers, and EBNA3 gene amplification was carried out with common 5' and 3' primers; in each case, the products were probed separately with type-specific probes. For strain detection, LMP1 gene amplifications were carried out across the 33-bp repeat and 30-bp deletion loci. Assays included reference DNA samples from the type 1 EBV-carrying B95.8 cell line (4.5 copies of the 33-bp repeat, undeleted [undel] at the 30-bp locus), the type 2 EBV-carrying AG876 cell line (4 copies of the 33-bp repeat, deleted [del] at the 30-bp locus), and the EBV-negative (EB-ve) BJAB cell line. All BCH10-derived isolates were type 1 and were identical by strain-specific marker analysis. (Longer exposures of the LMP1 deletion gel showed that isolate 2* was deleted at the 30-bp locus.)

View larger version (74K): [in this window] [in a new window]   FIG. 2.   EBNAprint analysis of EBV strains carried by blood-derived LCLs 1 to 4 and throat washing-derived LCLs 1* to 6* from patient BCH10. Protein extracts were separated by SDS-PAGE, and the immunoblots were then probed with the type 1 EBNA3-specific MAb E3CA10, the EBNA2-specific MAb PE2, and the EBNA1-specific MAb IH4. Immunoblots included reference protein samples from the type 1 EBV-transformed LCL C2+ OBA, the type 2 EBV-transformed LCL C2+ BL16, and the EBV-negative (EB-ve) BJAB cell line. All BCH10-derived isolates gave identical EBNAprints.

(ii) Patients with multiple EBV strains. The remaining 14 patients gave evidence of infection with more than one EBV strain. For example, patient BCH3, who could be studied on a single occasion only and from whom only five viral isolations were made, was interesting in that one type 1 strain was detected in the blood and two other distinct type 1 strains were detected in the throat. Figure 3 presents the PCR analysis of genomic markers, from which it is clear that all five isolates were type 1 at both the EBNA2 and EBNA3C loci. However, three different patterns (shown by blood isolates 1 and 2, throat isolates 1* and 2*, and throat isolate 3*) were observed from analysis of the polymorphic LMP1 loci. These different viral identities were fully confirmed by the immunoblot analysis illustrated in Fig. 4. Although all isolates encoded EBNA2 proteins which migrated to similar positions within the type 1 size range, the EBNA1 and EBNA3C immunoblots revealed three different EBNAprints. In the case of EBNA1, the three resident viral strains could be distinguished by EBNA1 size and by the fact that the EBNA1 encoded by throat isolates 1* and 2* lacked the MAb IH4 epitope and had to be detected by using a polyvalent human serum, AM, with strong anti-EBNA1 reactivity. In the case of EBNA3C, the strains were again distinguishable on the basis of their different-sized EBNA3C proteins and by the absence of the MAb E3CA10 epitope in the protein encoded by throat isolate 3*.

View larger version (45K): [in this window] [in a new window]   FIG. 3.   PCR analysis of EBV strains carried by blood-derived isolates 1 and 2 and throat washing-derived isolates 1* to 3* from patient BCH3. Assays were conducted with controls as described for Fig. 1. All BCH3-derived isolates were type 1 but could be distinguished into three groups (1, 2 v 1*, 2* v 3*) by strain-specific marker analysis.

View larger version (51K): [in this window] [in a new window]   FIG. 4.   EBNAprint analysis of EBV strains carried by blood-derived isolates 1 and 2 and throat washing-derived isolates 1* to 3* from patient BCH3. Immunoblots were conducted with controls as described for Fig. 2; additional blots were probed with polyclonal human (HU) sera from a type 1 virus-infected donor, MS (known to be reactive to type 1 EBNA3C), and from donor AM (known to be selectively reactive to EBNA1). The BCH3-derived isolates could be classified into three groups (1, 2 v 1*, 2* v 3*) from their distinct EBNAprints.

View larger version (98K): [in this window] [in a new window]   FIG. 5.   PCR analysis of EBV strains carried by blood-derived isolates 1 to 7 and throat washing-derived isolates 1* to 9* from patient QEH5. Assays were conducted with controls as described for Fig. 1. All QEH5-derived isolates were type 1 and undeleted at the LMP1 30-bp locus but showed different numbers of LMP1 33-bp repeats. Note that some lanes display more than one PCR product in the LMP1 repeat analysis, implying that in such cases the LCL is a mixture of cells carrying different virus strains.

View larger version (95K): [in this window] [in a new window]   FIG. 6.   EBNAprint analysis of EBV strains carried by blood-derived isolates 1 to 5 and throat washing-derived isolates 2* and 7* to 9* from patient QEH5. Immunoblots were conducted with controls as described for Fig. 2; an additional blot was probed with the LMP1-specific MAb pool CS1-4. The two main EBNAprint patterns are illustrated by isolates 1, 3, 5, and 8* and by isolates 2, 4, and 2*, respectively.

View larger version (110K): [in this window] [in a new window]   FIG. 7.   PCR analysis of EBV strains carried by blood isolates 1 to 12 and throat washing-derived isolates 1* to 9* from patient BCH7. Assays were conducted with controls as described for Fig. 1. All LCLs carried type 1 virus, but LCLs 5, 8, and 12 also contained cells carrying a type 2 virus; among the type 1 isolates, one was dominant in the blood, but a second type 1 virus, distinguishable by LMP1 repeat size, was detectable in the throat (isolates 1* to 4*).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

This work set out to resolve a number of apparently contradictory findings in the literature regarding the number and type of EBV strains present within virus carriers. On the one hand, the analysis of healthy EBV-infected individuals in Western countries suggests that most if not all carry a single virus strain only and that this is of type 1 in at least 90% of cases (reviewed in reference 17). Such conclusions are based on detailed virus isolation studies (20, 53), including prospective sampling of individual donors over several years (53), and on the fact that familial transmission of particular virus strains can be traced retrospectively between generations (19). On the other hand, virus isolation studies of HIV-positive patients, largely involving male homosexual cohorts, show that at least half of such patients carry multiple EBV strains and that the incidence of type 2 virus infection exceeds 30% (7, 31, 44, 54, 55). This prompts a number of questions. In particular, do the data for T-cell-immunocompromised patients magnify what is occurring (albeit undetected) in healthy virus carriers, or are they specific to the immunocompromised patient group? If the latter, to what extent are the atypical features of EBV carriage in HIV-positive homosexuals a consequence of immune impairment per se and to what extent do they reflect the increased risk of hematogeneous viral transmissions associated with a homosexual lifestyle? We sought to address these questions by applying exactly the same EBV isolation protocols to a second HIV-positive cohort, namely, hemophilic patients who had become iatrogenically infected with HIV over a decade earlier from contaminated batches of Factor VIII concentrate.

The preliminary assays of anti-VCA antibody titers and of EBV-specific T-cell immunity in these hemophilic patients (Table 1) were important for the following reasons. First, they showed that approximately one-third of adolescent hemophilic patients, whether from the HIV-positive study group or HIV-negative controls, had no serological evidence of prior EBV infection, a value in line with that expected for this age group in the general population of the United Kingdom (25). Likewise, adult HIV-positive hemophilic patients had a 90% incidence of EBV seropositivity, again mirroring that seen in the general adult population (53). Hence, one can discount the possibility that Factor VIII preparations can act as a source of EBV infection; it is most likely, therefore, that hemophilic patients acquire EBV via the natural oral route. Second, using elevated anti-VCA titers and depressed virus-specific T-cell responses as surrogate markers of immune T-cell impairment (52), we found that the HIV-positive hemophilic patients as a group were at least as immunocompromised as the HIV-positive homosexuals examined in a parallel study (54). This is further supported by comparing general criteria of immune status, such as CD4⁺ T-cell counts and the proportion of patients with symptoms of AIDS (see footnotes a and b of Table 3). Hence, any differences between the two groups in their range of resident EBV strains cannot be ascribed to differences in host immune competence. The two studies were also equally rigorous in that here we screened 745 independent virus isolates from 39 hemophilic patients, compared to 560 isolates from 35 homosexual patients in the parallel work (54). Given these numbers, any differences in results are also unlikely to have arisen through sampling error alone.

Table 3 summarizes the overall data from the two studies in terms of the numbers of individuals with single or with multiple EBV strains and the numbers with type 1 and/or type 2 virus infection. These allow us to draw conclusions on two major issues. The first issue is whether overt multiple infections are a feature of all T-cell-immunocompromised groups. We were concerned that the frequency of multiple infection previously seen in the HIV-positive homosexual patients might reflect the atypical acquisition of additional EBV strains via a hematogeneous route from sexual contacts. Likewise, allograft recipients, another immunocompromised group for whom multiple EBV infections have been reported (18, 20), might also have acquired virus by an atypical route, namely, from B cells in the allograft itself (16, 23). In fact, we found that a substantial proportion (36%) of HIV-positive hemophilic patients also carry two or more detectable EBV strains, and, as argued earlier, these almost certainly have been acquired by the natural route. As an additional control to ask whether hemophilia per se is a predisposing factor, we also screened several HIV-negative hemophilic patients without finding evidence of coresident EBV strains (55a). The data therefore suggest very strongly that immune T-cell impairment is the major determinant of the overt multiple EBV infections seen in HIV-positive individuals. It is also interesting that both here (Table 2) and in the homosexual cohort (54), multiple infections with virus strains of the same type are more frequently found in the throat than in the blood; possible reasons for this are discussed elsewhere (54).

It remains a moot point, however, whether the acquisition of multiple EBV strains is genuinely restricted to immunocompromised individuals or is merely easier to detect in such cases because of the elevated virus load in vivo. In this context, some studies based on the direct PCR amplification of ex vivo samples, either across polymorphisms in the EBNA1 gene (4) or in the BamHI-F region of the genome (30), suggest that different strains coexist in the blood and/or throat of healthy immunocompetent carriers. On the other hand, in vitro isolation studies indicate the dominance of a single transforming strain in the great majority of such carriers, with any observed heterogeneity among isolates from a single individual being the result of intrastrain recombination at repeat sequences in the viral genome (20, 53). Final resolution of this issue will require a combination of these two approaches and must take into account the possible involvement of transformation-defective viruses that would not be rescued in vitro. Nevertheless, the power of in vitro isolation methods should not be underestimated. In the present work the detection of coresident EBV strains in HIV-positive hemophilic patients was not dependent on establishing unusually large numbers of virus isolates per patient; the mean numbers of isolates from individuals with overt multiple infection were actually lower than those for individuals with only one detectable strain (see Table 2, footnote a). Furthermore, in most cases where coinfection occurred, the different resident strains were each represented several times in the panel of LCLs derived from that individual, and so their presence was not difficult to detect. The uniformity of virus strains observed among independent virus isolates from healthy carriers (20, 53) is therefore more striking when viewed in this light.

The second issue being addressed by this study is whether type 2 virus strains are prevalent in all T-cell-immunocompromised groups. Our data strongly suggest that this is not the case. Thus, we found that just 10% of HIV-positive hemophiliacs had detectable type 2 EBV infection, and in only one patient was a type 2 virus the dominant strain. This is significantly different (P < 0.05) from what was seen in HIV-positive homosexuals, where >30% of patients carried a type 2 virus, often as the dominant or equidominant strain rescuable in vitro (54). Note that although the slower growth of type 2 virus-transformed LCLs can bias against type 2 virus detection in in vitro isolation assays (38), our protocols are specifically designed to minimize this problem. Whatever residual bias exists, however, must apply equally to the studies of both HIV-positive cohorts. We conclude, therefore, that there is a real difference in type 2 virus prevalence between the hemophilic and homosexual patient groups. It is well established that homosexual groups can harbor a greater range of infectious agents than are endemic in the general population. By contrast, apart from iatrogenic infections from Factor VIII preparations, hemophilic patients tend to mirror the general population in their viral flora. The example of human herpesvirus 8, a sexually transmitted gammaherpesvirus, is especially pertinent here; the incidence of human herpesvirus 8 infection in HIV-positive homosexuals is markedly higher than that seen in HIV-positive hemophilic patients or in the general population (12, 27).

We interpret our findings as showing that type 2 strains of EBV, a type most commonly seen in equatorial Africa, have become endemic in homosexual communities in the West, whereas they remain relatively rare in the general population as a whole. The previously reported high incidence of type 2 virus infection (7, 29, 31, 44, 50, 51, 54, 55) and of type 2 virus-positive malignancies (6, 10, 14, 21, 36, 46) among HIV-positive patients almost certainly reflects the predominance of male homosexuals in these patient cohorts, especially during the first 10 years of the AIDS epidemic (5). It is also necessary to reexamine the notion, to which these earlier findings gave rise, that type 2 viruses enjoy some kind of selective advantage over type 1 strains in the immunologically compromised host. There is no a priori reason for this, and it seems to us more likely that both viruses are able to take equal advantage of impaired T-cell surveillance. The implication of our findings is that the relative prevalence rates of type 1 and type 2 viruses, at least in the general population in the United Kingdom, are in line with those first suggested from in vitro isolation studies of healthy individuals, where the great majority carried type 1 virus and only 5 to 10% carried type 2 (53). Interestingly, an extensive survey of virus isolates from Dutch bone marrow transplant recipients led to similar conclusions regarding type 2 prevalence (20). Such observations are also in line with the European data from three EBV-associated tumors, posttransplant lymphoma, sporadic BL, and Hodgkin's disease, arising in HIV-negative patients; these tumors were found to carry a type 1 strain in the great majority of EBV genome-positive cases (8a, 10, 13, 24, 42, 58). This preponderance of type 1 viruses cannot be due to nonpathogenicity of type 2 EBV, since type 2 strains are well represented in these same categories of malignancy arising in (predominantly homosexual) AIDS cohorts (6, 10, 14, 21, 36, 46). The question of type 2 virus prevalence in other Western societies remains to be resolved and will require in vitro isolation work to complement the results of direct PCR amplification assays. It is possible that in North America, where a greater proportion of the population is of African descent, type 2 prevalence may be higher (47). Even here, however, the data from EBV-associated tumors arising in HIV-negative patients strongly suggest that type 1 viruses are predominant (8, 11, 14, 29a). It will be important to map the epidemiology of type 1 and type 2 virus infections more accurately on a worldwide basis so that we can begin to understand how EBV coevolution with the human species led to the emergence of these distinct virus types.