Int. J. Cancer: 70, 375382 (1997) r 1997 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de lUnion Internationale Contre le Cancer

EPSTEIN-BARR VIRUS-ASSOCIATED HODGKINS DISEASE: EPIDEMIOLOGIC CHARACTERISTICS IN INTERNATIONAL DATA

Sally L. GLASER1*, Ruby J. LIN1, Susan L. STEWART1, Richard F. AMBINDER2, Ruth F. JARRETT3, Pierre BROUSSET4, Gorm PALLESEN5, Margaret L. GULLEY6, Gulfaraz KHAN7, Jane OGRADY8, Michael HUMMEL9, Maria Victoria PRECIADO10, Hans KNECHT11, John K.C. CHAN12 and Alexander CLAVIEZ13 1Northern California Cancer Center, Union City, CA 2Johns Hopkins School of Medicine, Baltimore, MD 3LRF Virus Centre, University of Glasgow, UK 4Ho pitaux de Toulouse, France 5Aarhus Kommunehospital, Denmark 6University of Texas Health Science Center, San Antonio, TX 7Tufts University, Boston, MA 8Oxfordshire Health Authority, Oxford, UK 9Freie Universita t, Berlin, Germany 10Hospital de Nin os Dr. Ricardo Gutierrez, Buenos Aires, Argentina 11University of Massachusetts Medical Center, Worcester, MA 12Queen Elizabeth Hospital, Kowloon, Hong Kong 13Universita ts-Kinderklinik, Kiel, Germany
Hodgkins disease (HD) has long been suspected to have an infectious precursor, and indirect evidence has implicated Epstein-Barr virus (EBV), a ubiquitous herpesvirus, as a causal agent. Recent molecular studies using EBER in situ hybridization or latency membrane protein-1 (LMP-1) immunohistochemistry have identied EBV latent infection in up to 50% of HD tumors. However, the epidemiologic features of these cases have not been examined in detail. To explore the epidemiology of EBV-positive HD so as to understand the role of EBV in HD etiology more clearly, this project accumulated patient data from 14 studies that had applied these EBV assays to HD tumors. With information on age at diagnosis, sex, ethnicity, histologic subtype, country of residence, clinical stage and EBV tumor status from 1,546 HD patients, we examined risk for EBV-positive disease using logistic regression. Forty percent of subjects had EBV-positive tumors, and EBV prevalence varied signicantly across groups dened by the study variables. Odds ratios (OR) for EBV-associated HD were signicantly elevated for Hispanics vs. whites (OR 5 4.1), mixed cellularity vs. nodular sclerosis histologic subtypes (OR 5 7.3, 13.4, 4.9 for ages 014, 1549, 501 years), children from economically less-developed vs. more-developed regions and young adult males vs. females (OR 5 2.5). These ndings suggest that age, sex, ethnicity and the physiologic effects of poverty may represent biologic modiers of the EBV association and conrm that this association is strongly but variably linked to histologic subtype. The data augment biologic evidence that EBV is actively involved in HD pathogenesis in some cases but describe epidemiologic complexity in this process. Int. J. Cancer, 70:375382, 1997. r 1997 Wiley-Liss, Inc.

Hodgkins disease (HD) is a malignant lymphoma whose unusually heterogeneous clinical, histologic, and epidemiologic characteristics have suggested either a single disease entity with a complex host response or 2 or 3 etiologically distinct conditions. For either interpretation, an infectious precursor has long been proposed. This hypothesis was prompted by clinical symptoms of HD such as cyclic fevers and night sweats, by the morphologic appearance of reactive tissue surrounding the malignant (ReedSternberg) cells and by epidemiologic ndings of a bimodal age-incidence curve, geographic variation in the incidence in young persons and childhood social-class risk factors consistent with HD as an uncommon outcome of delayed infection in young adults (Kaplan, 1980; MacMahon, 1966; Correa and OConor, 1971; Gutensohn and Cole, 1981). Considerable evidence has pointed to an etiologic role for Epstein-Barr virus (EBV), a ubiquitous herpesvirus that has been associated with a number of lymphoid and epithelial malignancies.

Epidemiologic studies have shown that a history of mononucleosisa clinical manifestation of EBV infectiontriples the risk of HD, and that HD patients have higher antibody titers than control subjects to the EBV antigens indicating primary infection and viral reactivation (Mueller, 1987). One study found that persons with serologic evidence of EBV infection had 2.5- to 4-times the risk of developing HD as persons without infection (Mueller et al., 1989). Biologic plausibility for an association between EBV and HD is suggested by observations that EBV immortalizes human B-lymphocytes and renders them tumorigenic in immunodecient mice, that expression of an EBV gene in immortalized rodent cell lines results in transformation both in vitro and in vivo and that EBV is directly tumorigenic in some primates (Wang et al., 1985; Moorthy and Thorley-Dawson, 1993; Wang et al., 1988; Kaye et al., 1993; Cleary et al., 1985). A role for EBV in HD pathogenesis has been further supported by the identication of latent EBV infection in a proportion of HD tumors, based on Southern blot DNA hybridization (Weiss et al., 1987; Staal et al., 1989; Anagnostopoulos et al., 1989), in situ hybridization and antigen detection (Weiss et al., 1989; Wu et al., 1990; Pallesen et al., 1991a; Herbst et al., 1991), and polymerase chain reaction (PCR) (Herbst et al., 1990). Ultimately, assays for abundantly expressed viral transcripts (EBERs) and immunohistochemical assays for the latency membrane protein-1 (LMP-1), both of which can be applied to archived tissue specimens, have provided direct evidence of EBV within Reed-Sternberg cells or their variants in as many as 40 to 50% of HD patients (Pallesen et al., 1991a; Gulley et al., 1994; Hummel et al., 1992; Herbst et al., 1992; Khan et al., 1992, 1993; Deacon et al., 1993; Chang et al., 1993; Ambinder et al., 1993; Brousset et al., 1993; Zhou et al., 1993; Delsol et al., 1992, 1993; Poppema et al., 1994; OGrady et al., 1994; Tomita et al., 1996; Zarate-Osorno et al., 1995; Chan et al., 1995; Preciado et al., 1995a and b; Claviez et al., 1994; Quintanilla-Martinez et al., 1996; Pinkus et al., 1994; Murray et al., 1992; Armstrong et al., 1992; Weiss et al., 1991; Joske et al., 1992). These latter molecular studies suggest a new classication for HD tumors as EBV-positive or EBV-negative. Distinctive epidemiologic features for these 2 subtypes, particularly features accounting
*Correspondence to: Northern California Cancer Center, 32960 AlvaradoNiles Road, Suite 600, Union City, CA 94587, USA. Fax: 510-429-2550. e-mail: sglaser@nccc.org Received: August 5, 1996.

376

GLASER ET AL.

for some of the established variation of HD by age, sex, ethnicity and histologic subtype, would help validate this classication, further support a pathogenetic function for EBV in HD, and provide clues regarding additional risk factors. However, detailed epidemiologic characterization of EBV-classied HD has been limited. Some studies have reported EBV-positive HD to be more common in males, in 1 of the 4 histologic subtypes, in children and older adults, in Hispanic populations, and in economically lessdeveloped countries (Pallesen et al., 1991a; Gulley et al., 1994; Chang et al., 1993; Ambinder et al., 1993; Khan et al., 1993; Zarate-Osorno et al., 1995; Quintanilla-Martinez et al., 1996; Armstrong et al., 1993; Jarrett et al., 1991; Weinreb et al., 1996). However, these ndings, frequently based on small case series from populations with differing HD risks, vary widely and often conict. Certain studies suggest EBV is independently associated with age, whereas others fail to support such an association (Gulley et al., 1994; Khan et al., 1993; Quintanilla-Martinez et al., 1996; Jarrett et al., 1991). Across studies, EBV-positive HD has been detected in 34 to 96% of males and 17 to 83% of females (Gulley et al., 1994; Chang et al., 1993; Zhou et al., 1993; Vestlev et al., 1992; Pinkus et al., 1994). In white populations, EBV prevalence has varied within and across histologic subtypes (nodular sclerosis, 1350%; mixed cellularity, 5096%; lymphocyte predominance, 0100%; lymphocyte depletion, 0100%) (Pallesen et al., 1991a; Herbst et al., 1991, 1992; Hummel et al., 1992; Khan et al., 1993; Delsol et al., 1993; Poppema et al., 1994; Vestlev et al., 1992; Pinkus et al., 1994; Murray et al., 1992; Armstrong et al., 1992; Weiss et al., 1991). Among Hispanics, it has ranged from 37 to 100% (Gulley et al., 1994; Chang et al., 1993; Ambinder et al., 1993; Zarate-Osorno et al., 1995; Quintanilla-Martinez et al., 1996; Armstrong et al., 1993). Because of the correlations among the age-, sex-, ethnic-, and histologic subtype-specic patterns of HD and the international variation in its incidence, exploration of the epidemiologic features of EBV-classied HD requires multivariate analysis of a large and diverse case series. Although individual laboratory studies have often lacked the requisite sample size, when combined they include adequate numbers of patients to permit such an analysis. Therefore, the purpose of this project was to aggregate patient data from studies that had used reliable assays for EBV in order to explore in some detail the epidemiologic characteristics of EBV-associated HD.
MATERIAL AND METHODS

Investigators were invited to participate in this study if they had published results of EBV testing in HD tumors using EBER in situ hybridization and/or LMP-1 assays before early 1995. For each patient included in their publications or tested subsequently, authors were asked to contribute data on age at diagnosis, sex, ethnicity, histologic subtype, clinical stage, year of diagnosis, country of residence at diagnosis, human immunodeciency virus (HIV) status and outcome of EBER and/or LMP-1 tests. Of the 23 research groups approached, 12 submitted case information [Jarrett et al. (Armstrong et al., 1992, 1993); Brousset et al. (Brousset et al., 1993; Delsol et al., 1992, 1993); Pallesen et al. (Pallesen et al., 1991a and b; Vestlev et al., 1992; Zhou et al., 1993; Levine et al., 1994); Gulley et al. (1994); Khan et al. (Khan et al., 1992, 1993); OGrady et al. (1994); Hummel et al. (1992); Ambinder et al. (1993); Preciado et al. (Preciado et al., 1995a and b; Knecht et al. (Knecht et al., 1991; Joske et al., 1992); Chan et al. (1995); and Claviez et al. (1994)]. This was augmented with one case series published in its entirety (Zarate-Osorno et al., 1995) and one unpublished series. In total, data were available on 1,566 HD patients, including 368 who were not previously part of published work. All tumors represented initial diagnoses, had been reviewed originally for uniform diagnosis and histologic subtyping, and had been preserved in xative before being assayed for EBV. Twenty HIV-positive subjects were excluded from analyses because of the very high proportion of such patients with EBV-positive tumors

and the distinctive clinical behavior of HIV-associated HD (Tirelli et al., 1995). The presence of EBV in Reed-Sternberg cells had been investigated for the 1,546 remaining patients; 505 were classied by EBER alone, 482 by LMP-1 alone, and 559 by both methods. In this last group, agreement between the 2 techniques was very high (k 5 0.929). Therefore, HD tumors were considered EBV-positive if assay results in Reed-Sternberg cells or their variants were positive by either method. HD tumors were considered EBVnegative if both assays were negative or, when only one assay was used, its result was negative. Eleven patients were EBER-positive but LMP-1-negative; 7 patients, all with the mixed cellularity (MC) subtype, were LMP-1positive but EBER-negative. For descriptive statistics, age at diagnosis was grouped into 5-year categories (04, 59 . . . 80 or over) and into 4 larger groups (014, 1539, 4054, 55 or over) consistent with the age-incidence patterns of HD. Classication of histologic subtype used the 4 standard Rye-conference categories (nodular sclerosis [NS], lymphocyte predominance [LP], MC, and lymphocyte depletion [LD]). Although nodular LP HD may be an entity that is distinct from other HD, in this study the 38 LP cases designated as nodular and the 11 described as diffuse were combined with the 88 LP cases not subclassied. Ethnicity was categorized as white, black, Asian, Hispanic, and other. Year of diagnosis was grouped as 19521979, 19801984, 19851989 and 19901994. To approximate the childhood socioeconomic environment shown to affect HD incidence (Correa and OConor, 1971), regional economic development was estimated from the 1970 per capita gross national product (GNP) for countries with adult or mixed-age cases and from the 1988 per capita GNP for countries with childhood case series. Childhood socioeconomic environment was classied as higher or lower based on a cut-point of the 30th percentile in GNP world rank (Smith-Morris, 1990). With this approach, the United States, United Kingdom, European countries and Hong Kong were classied as having higher economic development; India, China and countries in Central and South America and the Middle East were classied as having lower economic development. To evaluate the representativeness of our study series, we compared the age and sex distributions for selected regional subgroups of our cases with HD cases from appropriate populationbased cancer registries. Study subjects from the United Kingdom, Denmark, France and China (Peoples Republic and Hong Kong) were compared with registry cases diagnosed in 19831987 in these areas (Parkin et al., 1992). United States study cases, an ethnically diverse group primarily from the San Francisco Bay Area, San Antonio, Texas and Georgia, were compared with 19881992 population-based HD cases from California, a similarly varied population (Perkins et al., 1995). Statistical analyses were undertaken using SAS version 6.11 (SAS Institute, 1989). Differences across component case series and across case subgroups dened by study variables were evaluated with chi-square statistics for frequency distributions and Wilcoxon rank-sum tests for means. Because study cases were neither population-based nor randomly selected, comparisons of characteristics between EBV-positive and EBV-negative cases potentially would be confounded by the compositions of the component case series, particularly as some had been assembled on the basis of age, ethnicity and/or country for their increased likelihood of being EBV-positive (Gulley et al., 1994; Ambinder et al., 1993; Zarate-Osorno, 1995; Preciado et al., 1995a and b; Claviez et al., 1994; Armstrong et al., 1993). Therefore, we focussed our analyses on exploring the risk of EBV-positive HD among subsets of cases dened by each of the independent variables. To control for confounding, we modeled this risk using unconditional logistic regression (Breslow and Day, 1980). For the models, age was grouped as 014, 1549 and 501 years; ethnicity as white, Hispanic and other; and histologic subtype as NS, MC and other to

EPIDEMIOLOGY OF EBV-ASSOCIATED HODGKINS DISEASE

TABLE I HD PATIENT CHARACTERISTICS BY COMPONENT CASE SERIES Case series Number Mean age (years) % Male % White % European % NS % MC % EBV1

377

Jarrett et al., (Armstrong et al., 1992, 1993) Brousset et al., (Brousset et al., 1993; Delsol et al., 1992, 1993) Pallesen et al., (Pallesen et al., 1991a,b; Zhou et al., 1993; Vestlev et al., 1992; Levine et al., 1994) Gulley et al., (1994) Khan et al., (1992, 1993) OGrady et al., (1994) Hummel et al., (1992) Glaser and Ambinder, not previously published Ambinder et al., (1993) Preciado et al., (1995a,b) Knecht et al., (Joske et al., 1992; Knecht et al., 1991) Zarate-Osorno et al., (1995)1 Chan et al., (1995) Claviez et al., (1994)
1Data

304 266 236

31.4 39.6 37.0

60.2 63.5 59.8

88.8 96.2 88.1

88.8 100 88.1

53.9 37.4 60.9

32.7 50.2 29.4

35.2 35.0 44.9

171 130 97 94 65 39 37 36 27 23 21

36.3 30.4 38.5 37.8 38.9 9.8 8.2 39.0 32.7 39.3 10.4

66.1 66.1 65.0 87.2 0 47.4 78.4 69.4 48.2 78.3 57.1

11.8 91.9 100 100 66.2 35.9 100 94.4 0 0 100

0 97.7 100 100 0 0 0 94.4 0 0 100

61.2 57.7 44.3 51.1 81.5 48.7 24.3 55.6 48.2 69.6 52.4

34.7 24.6 35.1 36.2 7.7 41.0 56.8 33.3 25.9 21.7 42.9

46.8 30.8 33.0 45.7 23.1 53.9 46.0 58.3 66.7 65.2 47.6

from publication only.

maximize statistical power and facilitate interpretation. Backward elimination was used to delete variables from the initial model, producing a nal model that was tested for goodness-of-t by a restricted chi-square test. The models evaluated the effects of age group, sex, ethnicity, histologic subtype and regional economic level on the risk of EBV-positive HD, controlling for country, clinical stage and component case series. Because earlier reports suggested that prevalence of EBV-positive HD in age groups varied by histologic subtype, sex and regional economic development, the initial model included two-way interactions between age, histology, sex and economic level.
RESULTS

Differences across case series Table I presents selected characteristics of the patient populations for each of the 14 component case series. The numbers of subjects per series ranged from 21 to 304. These case series differed signicantly (p # 0.01) in subjects mean age and in their distributions by sex, ethnicity, histologic subtype, country at diagnosis, economic level, year of diagnosis and EBV status, which varied from 23.1 to 66.7% positive. There were no differences in the distributions of clinical stage across the 5 case series reporting stage data (Khan et al., 1992, 1993; OGrady et al., 1994; Zarate-Osorno et al., 1995; Preciado et al., 1995a and b; Claviez et al., 1994). Epidemiologic features of HD of the study sample Study subjects were diagnosed between 1952 and 1994 (median year, 1989). Age at diagnosis ranged from 2 to 95, with a mean of 33.7 years. Figure 1 illustrates the excess of male cases in childhood and the young adult incidence peak characteristic of HD in incidence data (Perkins et al., 1995). The combined case series comprised a wide range of racial and ethnic groups, including 149 Hispanics from the United States, Mexico, Honduras, Costa Rica and Colombia; 26 Brazilians and 20 East Indians, among others. Patients resided in 24 countries, although 1,150 were from the United Kingdom, France, Denmark, or the United States. Table II presents the numbers of study cases by age, sex, ethnicity, and histologic subtype. A total of 58.1% of the subjects was male, with a male excess in all groups except blacks and young adults with NS. Ethnic groups differed signicantly in their distributions by age and histologic subtype. Children comprised approximately 13% of whites and Hispanics, 26.9% of Asians, and 60.6% of

FIGURE 1 Relative frequency distribution of Hodgkins disease study subjects by age and sex.

others. The proportion of NS cases ranged from 75.7% in blacks to 49.3% in others, whereas MC occurred among 42.3% of Hispanics but only 13.5% of blacks. There was also a signicant difference by regional economic level; the 221 cases from less-developed regions were more likely than the 1,322 from more-developed areas to be younger than age 15 (46.3% vs. 10.5%), have the MC subtype (46.6 vs. 32.5%), and be of non-white background (83.3% vs. 10.1%). Study patients had lower mean ages than population-based cases from corresponding areas (United Kingdom, 34.2 vs. 41.8 years; Denmark, 38.3 vs. 44.3 years; France, 36.9 vs. 40.2 years; China, 32.2 vs. 40.5 years; United States, 34.3 vs. 37.1 years). These differences primarily reected greater proportions of registry cases older than age 50. Age distributions were similar between study and registry cases only for Danish males and French females. Sex distributions were similar between study and registry patients only among those from the United Kingdom, Denmark, and France. Predictors of EBV-positive HD A total of 618 study subjects (40.0%) had EBV-positive HD. However, the proportions of cases with EBV-positive tumors differed signicantly (p # 0.001) across age, sex, ethnicity, histologic subtype, country of residence and regional economic level. Figure 2 shows that the highest percentages of EBV-positive cases in each 5-year age group occurred in children younger than 10 and

378

GLASER ET AL.
TABLE II DISTRIBUTION OF STUDY SUBJECTS BY HISTOLOGIC SUBTYPE, ETHNICITY, AGE GROUP AND SEX

Ethnicity Age group 014

Whites 1539 4054 551 014

Blacks 1539 4054 551 014

Asians 1539 4054 551 014

Hispanics 1539 4054 551 014

Others 1539 4054 551

NS LP MC LD NS LP MC LD

33 10 47 2 30 2 13 3

160 42 120 11 198 10 31 1

52 25 47 5 36 9 8 1

26 13 70 8 42 8 38 4

1 4 1

10 1 1 10

1 1 1

1 2 1 1

Males 4 5 1 6 3 1 Females 1 10 2 2

4 1 1

4 1 2 1 3

4 1 10 3 3

22 16 1 25 1 7 3

4 1 6 4 4

4 1 8 2 7 1 9 2

14 3 15 1 4 6

9 2 5 3 1

3 1 1 1

1 1

FIGURE 2 Percentage of EBV-positive patients in each 5-year age group.

FIGURE 3 Percentage of EBV-positive patients in each histologic subtype- and age-specic group.

in adults older than 80; the lowest percentages were in young adults, particularly those 15 to 29 years old. Males were nearly twice as likely to have EBV-associated tumors as females (47.7% vs. 29.2%). EBV-positive disease affected 35.9% of whites and 16.2% of blacks, but 60 to 65% each of Asians, Hispanics, Indians and persons of other ethnic origin. The MC subtype included the largest proportion of EBV-positive disease (70.4%), and the LP subtype included the smallest (16.0% overall, with 13.2% for nodular LP and 36.4% for diffuse LP); 23.2% of NS and 54.9% of LD tumors were EBV-positive. The prevalence of EBV-positive disease ranged widely among countries, from a maximum of 87.5% in Saudi Arabia (n 5 8) to a minimum of 30.8% in the United Kingdom (n 5 394). Persons from less-developed countries were almost twice as likely to have EBV-positive HD as those from more-developed regions (63.4% vs. 36.0%). With cases stratied by age, the distribution of EBV prevalence also varied signicantly (p , 0.001) across histologic subtypes, ethnic groups, and economic levels. Figure 3 shows that in all age groups, NS and particularly LP had relatively low percentages of EBV-positive cases, whereas MC was associated with high levels. Despite this variation, EBV prevalence was uniformly higher in children than in young adults for all histologic subtypes. Across ethnic categories, the respective EBV prevalence in the 4 age groups (014, 1539, 4054, 551 years) was higher among Asians (92.9%, 38.1%, 50.0%, 81.8%), Hispanics (85.7%, 49.3%, 63.2%, 66.7%) and others (74.4%, 40.0%, 66.7%, 100%) than among whites (46.2%, 29.6%, 37.6%, 48.3%) or blacks, who had relatively low proportions at all ages (16.7%, 13.0%, 33.3%, 20.0%). However, in almost all ethnic groups, the highest percentages with EBV-positive tumors occurred in children or in persons older than

55, and the lowest occurred consistently in young adults. For cases from economically less-developed regions, the percentages of EBV positivity were higher than for cases from more-developed regions in all 4 age groups (70.0%, 50.7%, 75.0%, 70.4% vs. 46.9%, 29.2%, 38.2%, 50.0%, respectively); however, young adults in both groups experienced the lowest EBV prevalence. Logistic regression analysis identied age group, sex, ethnicity, histologic subtype and regional economic level as signicant, independent predictors of the risk of EBV-associated HD. After the model controlled for the effects of the other factors, Hispanics had 4 times the risk of EBV-positive disease as whites (odds ratio [OR] 5 4.1; 95% condence interval, 1.89.6). The risks of EBV-positive HD associated with histologic subtype, economic level, and sex were modied by age (Table III). MC tumors were substantially more likely to be EBV-positive than NS tumors for persons of all ages but particularly for young adults, for whom the risk was 13-fold. The odds of HD being EBV positive were signicantly elevated in economically less-developed than in more-developed regions for children but not for older persons. Compared with males, females had half the risk of having EBV-positive disease at young adult ages. Table IV shows how the OR associated with age varied by economic level, histologic subtype and sex. In less-developed regions, the higher risk of EBV-positive disease in children than young adults occurred for both the NS and MC subtypes but was especially pronounced for NS (OR 5 10.0 and 14.3 for males and females). In moredeveloped regions, children with NS were at slightly higher risk of having EBV-positive disease than young adults, although at a signicant level only for girls (OR 5 2.2). For persons older than 50 compared with young adults, the effect of age on EBV positivity was less marked and did not vary with regional economic level.

EPIDEMIOLOGY OF EBV-ASSOCIATED HODGKINS DISEASE

379

However, older women with NS had a nearly 4-fold higher EBV risk than young adult women, and older men with MC had approximately half the EBV risk of younger men.
DISCUSSION

Earlier studies described epidemiologic features of EBVassociated HD, but these reports were based on EBV assays of differing sensitivities and often involved small study samples. This project has assembled the largest case series in which EBV latent gene products in Reed-Sternberg cells were evaluated by the 2 standard reliable assays. The size and diversity of this series have permitted us to explore epidemiologic patterns of EBV-positive HD among relevant case subgroups and to examine risks of EBVpositive disease in these subgroups free of the effects of major confounders. Although the case series was neither populationbased nor randomly selected, the age- and sex-specic distribution (Fig. 1) and the comparisons to regional registry data show that it was reasonably representative of population-based HD cases with respect to these demographic variables. Analyses of our data have conrmed several epidemiologic features of EBV-associated HD and identied characteristics that were signicant predictors of risk after control for co-variables. Histologic subtype was the strongest risk factor for HD being EBV positive. In our case series, EBV gene products were present in approximately 75% of MC tumors but in only 25% of NS cases, and persons with MC had very high odds of having EBV-positive disease compared with persons with NS. This elevated risk was not explained by the higher frequency of MC in groups likely to have EBV-positive tumors (children, Hispanics, residents of poorer countries). However, the particularly high odds for EBV-positive MC in young adults does reect the low likelihood of an EBV association for NS in this age group. Differences between MC and NS in extent and patterns of EBV association add to clinical and other epidemiologic evidence that these histologic subtypes represent distinct entities (Cozen et al., 1992).
TABLE III ADJUSTED ODDS RATIOS (OR) AND 95% CONFIDENCE INTERVALS (CI) FOR EBV-POSITIVE HD ASSOCIATED WITH HISTOLOGIC SUBTYPE, REGIONAL ECONOMIC DEVELOPMENT AND SEX Variable Comparison Age stratum (years) Adjusted OR 95% CI

Histologic subtype MC vs. NS1 Regional economic Less vs. more1 development Sex Female vs. male1

014 1549 501 014 1549 501 014 1549 501

7.3 13.4 4.9 6.0 0.9 0.8 0.6 0.4 1.2

3.814.2 9.019.9 2.88.7 2.018.0 0.42.3 0.23.0 0.31.1 0.30.6 0.72.0

1Referent

group.

EBV-positive HD occurred in a large proportion of Hispanic patients, who were 4 times as likely to have EBV-positive tumors as whites, after control for the other factors. Prevalence and risk of EBV-associated disease were also elevated in other non-white groups except blacks. Although the numbers of non-white cases were too small to permit detailed evaluation of EBV patterns, our ndings do demonstrate that ethnicity affects risk of EBVassociated HD independent of age, sex, histologic subtype and nationality. Ethnic differences in EBV tumor association may be due to unmeasured confounders (e.g., individual socioeconomic status) or may reect variation in genetic susceptibility, as with HLA type (which affects risk of HD and differs among ethnic groups) (Gutensohn, 1982; Klitz et al., 1994). We have found that young adult females were half as likely as males to have EBV-positive HD and were especially unlikely to have EBV-positive NS, as illustrated in Table IV by the higher risks for females of EBV-positive NS in childhood (OR 5 14.3, 2.2) and at older ages (OR 5 3.6, 3.9) relative to young adulthood. This age-specic protective effect of female gender is consistent with a role for female reproductive experience in the development of EBV-positive HD, notably the NS subtype. Although there is no direct evidence regarding reproductive risk for EBV-associated malignancies in general or HD in particular, experimental data do support an interaction of pregnancy-mediated immunosuppressive mechanisms (e.g., via glucocorticoids) with expression of EBV gene products, suggesting that such a biologic mechanism is plausible (Sargent, 1993; Glaser et al., 1995). EBV-associated HD was infrequent in young adults, especially those with the NS subtype, and was more common in children and older adults. This variation of EBV positivity with age and histologic subtype is consistent with the multiple-etiology hypothesis, which states that the cause of HD differs by age group. The nding also implies that EBV is unlikely to play a primary role as the infectious precursor to young adult HD predicted by earlier epidemiologic studies (Gutensohn and Cole, 1981; Jarrett, 1992; Glaser and Jarrett, 1996). The substantially increased risks of EBV-positive NS and MC in children vs. young adults suggest that timing of infection greatly affects the association of EBV with HD tumors, with early age at infection strongly predicting EBVpositive disease for both these histologic subtypes. The variation in magnitude of this age effect with regional economic level also points to the importance of socioeconomic conditions and thus conceivably the virulence of and/or susceptibility to infection in predicting the association of EBV in HD. For NS, the higher risk of EBV positivity in children even in developed regions may primarily reect the low risk of EBV association in NS for young adults. However, the presence of EBV in some NS tumors and the apparent importance of host factors such as age and economic circumstance in predicting this association illustrate the complexity of the relation of this virus with the histologic expression of HD. Our combined data conrm several previously described epidemiologic patterns of EBV-positive HD. Because the case series

TABLE IV ADJUSTED ODDS RATIOS (OR) AND 95% CONFIDENCE INTERVALS (CI) FOR EBV-POSITIVE HD ASSOCIATED WITH AGE Strata Variable Comparison Regional economic development Histologic subtype Males Adjusted 95% CI OR Females Adjusted 95% CI OR

Age

014 vs. 1549 years1 Less developed More developed 501 vs. 1549 years1 Less developed More developed

NS MC NS MC NS MC NS MC

10.0 5.5 1.6 0.9 1.2 0.4 1.3 0.5

2.639.7 1.323.3 0.8 3.0 0.4 1.7 0.4 3.9 0.1 1.4 0.7 2.4 0.3 0.8

14.3 7.9 2.2 1.2 3.6 1.3 3.9 1.4

3.459.9 1.736.0 1.0 4.7 0.5 2.7 1.111.3 0.4 4.3 2.2 7.1 0.8 2.8

1Referent

group.

380

GLASER ET AL.

contributing to this project included a majority of subjects from whom the earlier ndings were derived, conrmation was anticipated. Given the signicant differences in patient characteristics among these series (Table I), many of the discrepant ndings in EBV prevalence from previous reports are likely due to epidemiologic heterogeneity. Recent publications on EBV in HD case series not included in this analysis also present results similar to ours. Among 50 Japanese patients with HD, Tomita et al. (1996) detected EBV-positive tumors in 64%a rate similar to that found in our study for Asians (mostly Chinese)and reported signicantly elevated 5- to 7-fold risks of EBV-positive HD for the MC subtype compared with all others, for males vs. females, and for persons older than 40 vs. those younger. In 183 United States patients of unspecied ethnicity, Pinkus et al. (1994) found EBV-positive tumors in 69% of MC, 13% of NS, and no cases of LP and in signicantly more males than females (33.7 vs. 17.4%). Quintanilla-Martinez et al. (1996) reported that the percentage of 50 adult Mexicans with EBV-positive tumors was high (70%) and did not vary signicantly by age, sex or histologic subtype; these ndings may largely reect the higher risk of EBV-associated HD in Hispanics, given the absence of an independent risk of EBVpositive HD with regional economic development in adults in our case series. An international study of 277 pediatric HD patients described a protective effect of increasing age in childhood on risk of EBV-positive disease, consistent with the pattern in Figure 2, and signicant inter-country differences in risk (Weinreb et al., 1996). However, this analysis was not controlled for ethnicity or regional economic status, which may partly explain the association of EBV positivity with nationality. Our results have revealed signicant and often large epidemiologic differences in EBV-positive and EBV-negative HD, but they also have shown that the epidemiologic characteristics considered here do not discriminate neatly between the 2 virus-dened subtypes of HD. To some extent, such incomplete concordance may be a consequence of misclassication in our data, given that patient information, histologic diagnosis, and laboratory assays for this project were not obtained under a single protocol. Although the observed differences in EBV prevalence by histologic subtype are unlikely to be due to histologic misclassication, some of the relatively few EBV-positive NS cases might represent a subcategory of NS, such as the LD variant, with a higher likelihood of being EBV associated. Inconsistencies and errors in ethnic categorization are a possible source of bias; however, because U.S. Hispanics (one group subject to considerable misclassication) were identied both by

medical record report and surname linkage, bias in this group should be reduced (West et al., 1995). Finally, inter-laboratory differences in technical procedures or assay interpretation also have the potential to affect our results (Pallesen et al., 1993). More likely, the lack of complete denition of EBV-associated HD by our study factors reects a complex interplay of environmental and host characteristics that are only partially understood and are inadequately measured by our study variables. In some ways, this complexity is reminiscent of that characterizing Burkitts lymphoma, which is also variably linked with EBV. As with HD, the association with EBV varies with age, incidence, and geography. Burkitts lymphoma is uniformly EBV-positive in equatorial Africa, Papua New Guinea, and parts of South America, where it is a common cancer in children; however, it is only infrequently EBV-linked elsewhere, occurring as a relatively rare disease that affects older persons (Magrath, 1991; Rickinson and Kieff, 1996; Gutierrez et al., 1992). However, unlike HD, the highly EBVpositive form of Burkitts lymphoma has been strongly associated with additional factors (e.g., holoendemic malaria). Moreover, irrespective of EBV association, Burkitts lymphomas share a common histologic appearance and chromosomal translocations. Findings regarding the association of EBV in HD in this large international case series suggest that the host factors of age, sex, ethnicity and physiologic effects of poverty may represent biologic modiers of the involvement of EBV in HD pathogenesis and conrm that the association is strongly linked to histologic subtype. However, the imprecision of each of these factors in predicting risk of EBV-positive HD reveals an epidemiologic complexity in the relation between virus and disease that exists over and above the effects of small sample sizes and demographic differences of the component case series. Ultimately, better knowledge of the biology of this enigmatic lymphoma and identication of co-factors for EBV association will be required to generate a more exact epidemiologic characterization and etiologic grasp of EBV-positive HD.
ACKNOWLEDGEMENTS

The authors acknowledge the contribution of case series by Dr. G. Barboa, Director de Patologia, Bogata, Colombia; Dr. V. Napoli, Department of Pathology, Grady Health System, Atlanta, GA; and K. Swaminathan, Department of Pathology, Postgraduate Institute of Basic Medical Sciences, Madras, India. They thank Ms. R. Leung, Ms. E. Satariano, M.P.H., and Mr. J. Hsu, M.P.H., for their contributions to the project.

REFERENCES

AMBINDER, R.F. and 14 OTHERS, Epstein-Barr virus and childhood Hodgkins disease in Honduras and the United States. Blood, 81, 462467 (1993). ANAGNOSTOPOULOS, I., HERBST, H., NIEDOBITEK, G. and STEIN H., Demonstration of monoclonal EBV genomes in Hodgkins disease and Ki-1 positive anaplastic large cell lymphoma by combined Southern blot and in situ hybridization. Blood, 74, 810816 (1989). ARMSTRONG, A.A., ALEXANDER, F.E., PINTO PAES, R., MORAD, N.A., GALLAGHER, A., KRAJEWSKI, A.S., JONES, D.B., ANGUS, B., ADAMS, J., CARTWRIGHT, R.A., ONIONS, D.E. and JARRETT, R.F., Association of Epstein-Barr virus with pediatric Hodgkins disease. Amer. J. Pathol., 142, 16831688 (1993). ARMSTRONG, A.A., WEISS, L.M., GALLAGHER, A., JONES, D.B., KRAJEWSKI, A.S., ANGUS, B., BROWN, G., JACK, A.S., WILKINS, B.S., ONIONS, D.E. and JARRETT, R.F., Criteria for the denition of Epstein-Barr virus association in Hodgkins disease. Leukemia, 6, 869874 (1992). BRESLOW, N.E. and DAY, N.E., Statistical Methods in Cancer Research. Vol. 1. The Analysis of Case-Control Studies. IARC, Lyon (1980). BROUSSET, P., ROCHAIX, P., CHITTAL, S., RUBIE, H., ROBERT, A. and DELSOL, G., High incidence of Epstein-Barr virus detection in Hodgkins disease and absence of detection in anaplastic large-cell lymphoma in children. Histopathology, 23, 189191 (1993). CHAN, J.K.C., YIP, T.T.C., TSANG, W.Y.W., LAU, W.H., WONG, C.S.C. and MA, V.W.S., Detection of Epstein-Barr virus in Hodgkins disease occurring in an Oriental population. Hum. Pathol., 26, 314318 (1995).

CHANG, K.L., ALBUJAR, P.F., CHEN, Y.Y., JOHNSON, R.M. and WEISS, L.M., High prevalence of Epstein-Barr virus in the Reed-Sternberg cells of Hodgkins disease occurring in Peru. Blood, 81, 496501 (1993). CLAVIEZ, A., TIEMANN, M., PETERS, J., KREIPE, H., SCHNEPPENHEIM, R. and PARWARESCH, R., The impact of EBV, proliferation rate, and bcl-2 expression in Hodgkins disease in childhood. Ann. Hematol., 68, 6166 (1994). CLEARY, M.L., EPSTEIN, M.A., FINERTY, S., DORFMAN, R.F., BORNKAMM, G.W., KIRKWOOD, J.K., MORGAN, A.J. and SKLAR, J., Individual tumors of multifocal EB virus-induced malignant lymphomas in tamarins arise from different B-cell clones. Science, 228, 722724 (1985). CORREA, P. and OCONOR, G.T., Epidemiologic patterns of Hodgkins disease. Int. J. Cancer, 8, 192201 (1971). COZEN, W., KATZ, J. and MACK, T.M., Risk patterns of Hodgkins disease in Los Angeles vary by cell type. Cancer Epidemiol. Bio. Prev., 1, 261268 (1992). DEACON, E.M., PALLESEN, G., NIEDOBITEK, G., CROCKER, J., BROOKS, L., RICKINSON, A.B. and YOUNG, L.S., Epstein-Barr virus and Hodgkins disease: transcriptional analysis of virus latency in the malignant cells. J. exp. Med., 177, 339349 (1993). DELSOL, G., BROUSSET, P., CHITTAL, S. and RIGAL-HUGUET, F., Correlation of the expression of Epstein-Barr virus latent membrane protein and in situ hybridization with biotinylated BamHI-W probes in Hodgkins disease. Amer. J. Pathol., 140, 247253 (1992).

EPIDEMIOLOGY OF EBV-ASSOCIATED HODGKINS DISEASE

381

DELSOL, G., MEGGETTO, F., BROUSSET, P., COHEN-KNATO, E., AL SAATI, T., ROCHAIX, P., GORGUET, B., RUBIN, B., VOIGT, J.J. and CHITTAL, S., Relation of follicular dendritic reticulum cells to Reed-Sternberg cells of Hodgkins disease with emphasis on the expression of CD21 antigen. Amer. J. Pathol., 142, 17291738 (1993). GLASER, R., KUTZ, L.A., MACCALLUM, R.C., and MALARKEY, W.B., Hormonal modulation of Epstein-Barr virus replication. Neuroendocrinology, 62, 356361 (1995). GLASER, S.L. and JARRETT, R.F., The epidemiology of Hodgkins disease. Baillie ` res Clin. Haematol., 9, 401416 (1996). GULLEY, M., EAGAN, P.A., QUINTANILLA-MARTINEZ, L., PICADO, A.L., SMIR, B.N., CHILDS, C., DUNN, C.D., CRAIG, F.E., WILLIAMS, J.W. JR. and BANKS, P.M., Epstein-Barr virus DNA is abundant and monoclonal in the ReedSternberg cells of Hodgkins disease: association with mixed cellularity subtype and Hispanic American ethnicity. Blood, 83, 15951602 (1994). GUTENSOHN, N., Social class and age at diagnosis of Hodgkins disease: New epidemiologic evidence for the two-disease hypothesis. Cancer Treat. Rep., 66, 689695 (1982). GUTENSOHN, N. and COLE, P., Childhood social environment and Hodgkins disease. N. Engl. J. Med., 304, 135140 (1981). GUTIERREZ, M.I., BHATIA, K., BARRIGA, F., DIEZ, B., MURIEL, F.S., DE ANDREAS, M.L., EPELMAN, S., RISUENO, C. and MAGRATH, I.T., Molecular epidemiology of Burkitts lymphoma from South America: differences in breakpoint location and Epstein-Barr virus association from tumors in other world regions. Blood, 79, 32613266 (1992). HERBST, H., DALLENBACH, F., HUMMEL, M., NIEDOBITEK, G., PILERI, S., MULLER-LANTZSCH, N. and STEIN, H., Epstein-Barr virus latent membrane protein expression in Hodgkin and Reed-Sternberg cells. Proc. nat. Acad. Sci. (Wash.), 88, 47664770 (1991). HERBST, H., NIEDOBITEK, G., KNEBA, M., HUMMEL, M., FINN, T., ANAGNOSTOPOULOS, I., BERGHOLZ, M., KRIEGER, G. and STEIN, H., High incidence of Epstein-Barr virus genomes in Hodgkins disease. Amer. J. Pathol., 137, 1318 (1990). HERBST, H., STEINBRECHER, E., NIEDOBITEK, G., YOUNG, L.S., BROOKS, L., MULLER-LANTZSCH, N. and STEIN, H., Distribution and phenotype of Epstein-Barr virus-harboring cells in Hodgkins disease. Blood, 80, 484 491 (1992). HUMMEL, M., ANAGNOSTOPOULOS, I., DALLENBACH, F., KORBJUHN, P., DIMMLER, C. and STEIN, H., EBV infection patterns in Hodgkins disease and normal lymphoid tissue: expression and cellular localization of EBV gene products. Brit. J. Haematol., 82, 689694 (1992). JARRETT, R.F., Hodgkins disease. Baillie ` res Clin. Haematol., 5, 5779 (1992). JARRETT, R.F., GALLAGHER, A., JONES, D.B., ALEXANDER, F.E., KRAJEWSKI, A.S., KELSEY, A., ADAMS, J., ANGUS, B., GLEDHILL, S., WRIGHT, D.H., CARTWRIGHT, R.A. and ONIONS, D.E., Detection of Epstein-Barr virus genomes in Hodgkins disease: Relation to age. J. clin. Pathol., 44, 844848 (1991). JOSKE, D.J.L., EMERY-GOODMAN, A., BACHMANN, E., BACHMANN, F., ODERMATT, B. and KNECHT, H., Epstein-Barr virus burden in Hodgkins disease is related to latent membrane protein gene expression but not to active viral replication. Blood, 80, 26102613 (1992). KAPLAN, H.S., Hodgkins Disease (2nd ed.) Harvard University Press, Cambridge (1980). KAYE, K.M., IZUMI, K.M. and KIEFF, E., Epstein-Barr virus latent membrane protein-1 is essential for B-lymphocyte growth transformation. Proc. nat. Acad. Sci. (Wash.), 90, 91509154 (1993). KHAN, G., COATES, P.J., GUPTA, R.K., KANGRO, H.O. and SLAVIN, G., Presence of Epstein-Barr virus in Hodgkins disease is not exclusive to Reed-Sternberg cells. Amer. J. Pathol., 140, 757762 (1992). KHAN, G., NORTON, A.J. and SLAVIN, G., Epstein-Barr virus in Hodgkin Disease. Relation to age and subtype. Cancer, 71, 31243129 (1993). KLITZ, W., ALDRICH, C.L., FILDES, N., HORNING, S.J. and BEGOVICH, A.B., Localization of predisposition to Hodgkins disease in the HLA class II region. Amer. J. hum. Genet., 54, 497505 (1994). KNECHT, H., ODERMATT, B.F., BACHMANN, E., TEIXEIRA, S., SAHLI, R., HAYOZ, D., HEITZ, P. and BACHMANN, F., Frequent detection of Epstein-Barr virus DNA by the polymerase chain reaction in lymph node biopsies from patients with Hodgkins disease without genomic evidence of B- or T-cell clonality. Blood, 78, 760767 (1991). LEVINE, P.H., PALLESEN, G., EBBESEN, P., HARRIS, N., EVANS, A.S. and MUELLER, N., Evaluation of Epstein-Barr virus antibody patterns and detection of viral markers in the biopsies of patients with Hodgkins disease. Int. J. Cancer, 59, 4850 (1994).

MACMAHON, B., Epidemiology of Hodgkins disease. Cancer Res., 26, 11891200 (1966). MAGRATH, I.T., African Burkitts lymphoma. History, biology, clinical features, and treatment. Amer. J. Ped. Hematol./Oncol., 13, 222246 (1991). MOORTHY, R.K. and THORLEY-DAWSON, D.A., Biochemical, genetic, and functional analyses of the phosphorylation sites on the EBV-encoded oncogenic latent membrane protein LMP-1. J. Virol., 67, 263745 (1993). MUELLER, N., Epidemiologic studies assessing the role of the Epstein-Barr virus in Hodgkins disease. Yale J. Biol. Med., 60, 321327 (1987). MUELLER, N., EVANS, A., HARRIS, N.L., COMSTOCK, G.W., JELLUM, E., MAGNUS, K., ORENTREICH, N., POLK, B.F. and VOGELMAN, J., Hodgkins disease and Epstein-Barr virus: Altered antibody pattern before diagnosis. N. Engl. J. Med., 320, 689695 (1989). MURRAY, P.G., YOUNG, L.S., ROWE, M. and CROCKER, J., Immunohistochemical demonstration of the Epstein-Barr virus-encoded latent membrane protein in paraffin sections of Hodgkins disease. J. Pathol., 166, 15 (1992). OGRADY, J., STEWART, S., ELTON, R.A. and KRAJEWSKI, A.S., Epstein-Barr virus in Hodgkins disease and site of origin of tumour. Lancet, 343, 265266 (1994). PALLESEN, G., HAMILTON-DUTOIT, S.J., ROWE, M. and YOUNG, L.S., Expression of Epstein-Barr virus (EBV) latent gene products in tumour cells of Hodgkins disease. Lancet, 337, 320322 (1991a). PALLESEN, G., HAMILTON-DUTOIT, S.J. and ZHOU, X., The association of Epstein-Barr virus (EBV) with T cell lymphoproliferations and Hodgkins disease: Two new developments in the EBV eld. Adv. Cancer Res., 62, 179239 (1993). PALLESEN, G., SANDVEJ, K., HAMILTON-DUTOIT, S.J., ROWE, M. and YOUNG, L.S., Activation of Epstein-Barr virus replication in Hodgkin and ReedSternberg cells. Blood, 78, 11621165 (1991b). PARKIN, D.M., MUIR, C.S., WHELAN, S.L., GAO, Y.T., FERLAY, J. and POWELL, J. (eds.). Cancer Incidence in Five Continents, Vol. 6. International Agency for Research on Cancer, Lyon (1992). PERKINS, C.I., MORRIS, C.R., WRIGHT, W.E. and YOUNG, J.L., Cancer Incidence and Mortality in California by Detailed Race/Ethnicity, 198892. California Department of Health Services, Cancer Surveillance Section, Sacramento (1995). PINKUS, G.S., LONES, M., SHINTAKU, I.P. and SAID, J.W., Immunohistochemical detection of Epstein-Barr virus-encoded latent membrane protein in Reed-Sternberg cells and variants of Hodgkins disease. Mod. Pathol., 7, 454461 (1994). POPPEMA, S. and VISSER, L., Epstein-Barr virus positivity in Hodgkins disease does not correlate with an HLA A2-negative phenotype. Cancer, 73, 30593063 (1994). PRECIADO, M.V., DE MATTEO, E., DIEZ, B. and GRINSTEIN, S., Epstein-Barr virus (EBV) latent membrane protein (LMP) in tumor cells of Hodgkins disease in pediatric patients. Med. Ped. Oncol., 24, 15 (1995a). RGUEZ, J. and GRINSTEIN, PRECIADO, M.V., DE MATTEO, E., DIEZ, B., MENA S., Presence of Epstein-Barr virus and strain type assignment in Argentine childhood Hodgkins disease. Blood, 86, 39223929 (1995b). QUINTANILLA-MARTINEZ, L., GAMBOA-DOMINGUEZ, A., GAMEZ-LEDESMA, I., ANGELES-ANGELES, A. and MOHAR, A., Association of Epstein-Barr virus latent membrane protein and Hodgkins disease in Mexico. Mod. Pathol., 8, 675679 (1996). RICKINSON, A.B. and KIEFF, E., Epstein-Barr Virus. In: Fields, B.N., Knipe, D.M., Howley, P.M., Chanock, R.M., Melnick, J.L., Monath, T.P., Roizman, B. and Straus, S.E., (eds). Fields Virology, (3rd ed.), pp. 23972446, Lippincott-Raven, Philadelphia (1996). SARGENT, I.L., Maternal and fetal immune responses during pregnancy. Exp. clin. Immunogenet., 10, 85102 (1993). SAS INSTITUTE, SAS/STAT Users Guide, Version 6, (4 ed.) Vol. 2, SAS Institute, Cary, NC (1989). SMITH-MORRIS, M. (ed.)., The Economist Book of Vital World Statistics. Random House, New York (1990). STAAL, S.P., AMBINDER, R., BESCHORNER, W.E., HAYWARD, G.S. and MANN, R., A survey of Epstein-Barr virus DNA in lymphoid tissue. Frequent detection in Hodgkins disease. Amer. J. clin. Pathol., 91, 15 (1989). TIRELLI, U., ERRANTE, D., DOLCETTI, R., GLOGHINI, A., SERRAINO, D., VACCHER, E., FRANCESCHI, S., BOIOCCHI, M. and CARBONE, A., Hodgkins disease and Human Immunodeciency Virus infection: Clinicopathologic and virologic features of 114 patients from the Italian Cooperative Group on AIDS and tumors. J. Clin. Oncol., 13, 17581767 (1995). TOMITA, Y., OHSAWA, M., KANNO, H., HASHIMOTO, M., OHNISHI, A.,

382

GLASER ET AL.

NAKANISHI, H. and AOZASA, K., Epstein-Barr virus in Hodgkins disease patients in Japan. Cancer, 77, 186192 (1996). VESTLEV, P.M., PALLESEN, G., SANDVEJ, K., HAMILTON-DUTOIT, S.J. and BENDTZEN, S.M., Prognosis of Hodgkins disease is not inuenced by Epstein-Barr virus latent membrane protein. (Letter) Int. J. Cancer, 50, 670671 (1992). WANG, D., LIEBOWITZ, D. and KIEFF, E., An EBV membrane protein expressed in immortalized lymphocytes transforms established rodent cells. Cell, 43, 831840 (1985). WANG, D., LIEBOWITZ, D., WANG, F., GREGORY, C., RICKINSON, A., LARSON, R., SPRINGER, T. and KIEFF, E., Epstein-Barr virus latent membrane protein (LMP) alters lymphocyte morphology, adhesion, and growth: deletion of the amino terminus abolishes activity. J. Virol., 62, 41734184 (1988). WEINREB, M. and 22 OTHERS, The role of Epstein-Barr virus in Hodgkins disease from different geographical areas. Arch. Dis. Child., 74, 2731 (1996). WEISS, L.M., CHEN, Y.Y., LIU, X.F. and SHIBATA, D., Epstein-Barr virus and Hodgkins disease. A correlative in situ hybridization and polymerase chain reaction study. Amer. J. Pathol., 139, 12591265 (1991).

WEISS, L.M., MOVAHED, L.A., WARNKE, R.A. and SKLAR, J., Detection of Epstein-Barr viral genomes in Reed-Sternberg cells of Hodgkins disease. N. Engl. J. Med., 320, 502506 (1989). WEISS, L.M., STRICKLER, J.G., WARNKE, R.A., PURTILO, D.T. and SKLAR, J., Epstein-Barr viral DNA in tissues of Hodgkins disease. Amer. J. Pathol., 129, 8691 (1987). WEST, D., GLASER, S., HORN-ROSS, P.L., STEWART, S. and SWALLEN, K., Misclassication of Hispanic ethnicity in a cancer registry data. Amer. J. Epidemiol., 110, S52 (1995). WU, T.-C., MANN, R.B., CHARACHE, P., HAYWARD, S.D., STAAL, S., LAMBE, B.C. and AMBINDER, R.F., Detection of EBV gene expression in ReedSternberg cells of Hodgkins disease. Int. J. Cancer, 46, 801804 (1990). ZARATE-OSORNO, A., ROMAN, L.N., KINGMA, D.W., MENESES-GARCIA, A. and JAFFE, E.S., Hodgkins disease in Mexico. Prevalence of Epstein-Barr virus sequences and correlations with histologic subtype. Cancer, 75, 13601366 (1995). ZHOU, X.G., HAMILTON-DUTOIT, S., YAN, O.H. and PALLESEN, G., The association between Epstein-Barr virus and Chinese Hodgkins disease. Int. J. Cancer, 55, 359363 (1993).