herbal medicine use, epstein-barr virus, and risk of...

5
[CANCER RESEARCH 52, 3048-3051, June I. 1992] Herbal Medicine Use, Epstein-Barr Virus, and Risk of Nasopharyngeal Carcinoma1 Allan Hildesheim,2 Sheila West, Ernesto DeVeyra, Maria F. De Guzman, Adonis Jurado, Carol Jones, Joko linai, and Yorio Hinuma3 Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 [A. H., C. J.J; Wilmer Eye Institute, The Johns Hopkins University, School of Medicine, Baltimore, Maryland 21205 ¡S.W.]; Departments of Pathology and Ear, Nose, and Throat, Philippine General Hospital, University of The Philippines, Manila, The Philippines [E. D., M. F. D. G., A. J.J; and Institute for Virus Research, Kyoto University, Kyoto 606, Japan [J. /., Y. H.] ABSTRACT Herbal medicine use is thought to be linked to nasopharyngeal carci noma (NPC) either through its ability to reactivate the Epstein-Barr virus (EBV) or through a direct promoting effect on EBV-transformed cells. To investigate this, 104 histologically confirmed NPC cases and 205 matched controls were studied in The Philippines. Blood was col lected to assess antibody titers against EBV, and an interview was administered which elicited information concerning herbal medicine use and other risk factors for NPC. Subjects strongly positive for anti-EBV antibodies (Epstein-Barr nuclear antigen |EBNA|) (titers - 1:80) were at a 21-fold excess risk of disease (95% confidence interval, 8.4, 51.8). Herbal medicine use was also associated with NPC (relative risk, 2.5; 95% confidence interval, 1.4,4.5). Associations persisted after adjustment for education, smoking, Chinese ancestry, and consumption of salted fish. Exposure to herbal medicines among subjects testing negative/weakly positive for anti-EBNA antibodies was not associated with an elevation in risk (relative risk, 0.6), strong positivity to anti-EBNA antibodies in the absence of herbal medicine use was associated with a significant 16- fold excess risk of disease, and exposure to herbal medicines among subjects testing strongly positive for anti-EBNA antibodies was associ ated with a significant 49-fold excess risk of NPC when cases were compared to controls. Similar results were obtained when other serolog- ical measures of EBV exposure were used. Anti-EBV antibody titers were elevated in herbal medicine users compared to nonusers among cases but not among control subjects. This suggests that, if herbal medicines interact with EBV in the development of NPC, they do not do so by reactivating EBV infection but rather through a direct proliferative effect on EBV-transformed cells. Although the interaction between EBV and herbal medicines is biologically plausible, larger, more detailed studies need to be conducted to validate this preliminary finding. INTRODUCTION EBV4 has been implicated in the etiology of NPC. EBV DNA has been detected in nearly all NPC tumor specimens investi gated to date (1), and recent evidence has demonstrated the transforming ability of the EBV latent membrane protein which is expressed in approximately 65% of NPC tumors (2). Given the ubiquity of this virus relative to the incidence of NPC, however, it is likely that other agents act as cofactors in the etiology of this disease. One such factor, phorbol esters present in extracts from herbal medicines and plants, has been hypoth esized to be linked to NPC either through its ability to reactivate EBV infection (3) or through a direct promoting effect on cells transformed by the EBV (4, 5). A case-control study of NPC was conducted in a Filipino population in The Philippines to examine risk factors for this disease in a country with incidence rates intermediate to that Received 11/22/91; accepted 3/20/92. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' Funded by the Philippine National Research Council. 2To whom requests for reprints should be addressed, at Environmental Epi demiology Branch. National Cancer Institute, Executive Plaza North. Room 443, Bethesda. MD 20892. 3 Present Address: Shionogi Institute for Medical Science, Osaka, Japan. 4The abbreviations used are: EBV, Epstein-Barr virus; NPC, nasopharyngeal carcinoma; VCA, viral capsid antigen; EA. early antigen; EBNA. Epstein-Barr nuclear antigen; RR, relative risk: CI. confidence interval. in Hong Kong (30.0/100,000 among males, 12.9/100,000 among females) and the United States (<1.0/100,000 among males and females). The age-standardized rate of NPC in Rizal Province in The Philippines in the period of 1978-1982 was 4.7/100,000 per year in males and 2.6/100,000 per year in females (6). As part of our case-control study, information was obtained concerning the use of herbal medicines, and EBV antibody status was assessed serologically. This has enabled us to examine the hypothesis that EBV and herbal medicines interact in the development of NPC. PATIENTS AND METHODS Incident cases of histologically confirmed NPC were recruited from the Philippine General Hospital. To be eligible for the study, cases were required not to have had previous treatment for NPC and to have resided on the island of Luzon for at least 6 months. All subjects presenting to the clinic with symptoms compatible with NPC and referred for a nasopharynx biopsy were invited to participate in the study. A total of 234 cases suspicious of NPC were identified during a 24-month period. Of these, 22 (9%) refused to undergo biopsy and were therefore considered ineligible for the study. Of the remaining 212, 104 individuals were pathologically confirmed to have NPC. All consented to participate in the study. Two controls were selected for each confirmed NPC case. One control was selected from the hospital and a second from the commu nity. Hospital controls were individually matched to cases on sex, age (±5 years), and hospital ward type (private versus public). Hospital controls were chosen from the medical/surgical ward on the same day that a suspected case was confirmed by biopsy result. Medical rosters routinely kept by the hospital were used by the study coordinator to select hospital controls. This was done by reviewing the roster and selecting the first eligible subject as a control. Since this study was also designed to examine the effect of diet on risk of NPC, individuals with disorders thought to be associated with changes in dietary patterns were not eligible as controls. Thus, patients with the following diagnoses were excluded from the study: gastrointestinal cancers, peptic ulcer, chronic cirrhosis, and gallbladder disease. A total of 104 hospital controls were selected, and all agreed to participate. Twenty-four hos pital controls had diagnoses of urinary/renal disease, 15 with acute appendicitis, 11 with non-NPC/gastrointestinal cancers, 9 with respi ratory impairment, and the remaining 45 with other conditions. Community controls were individually matched to cases on sex, age (±5years), and neighborhood. Selection of neighborhood controls was achieved by systematically selecting households either to the right or left of the subject's residence until an appropriate match was found. Replacement controls were obtained if the original control selected refused to participate. Thus, a total of 131 community controls were ascertained, of which 101 (77%) agreed to participate. Among those who refused, 76% did so because they refused blood extraction, and the remaining 24% declined all involvement. A personal interview was administered to consenting subjects in their native language, Tagalog, by a trained nurse-interviewer. The interview obtained information concerning sociodemographic characteristics, Chinese ancestry, adult diet, occupation, smoking, and herbal medicine use. "Ever use" of herbal medicines was assessed, and for those who reported use, the names of the medicines used were assessed. Currency of use and total duration of use were not assessed. In addition, 5 ml of venous blood was collected from each consenting subject for anti-EBV antibody determination by immunofluorescence (7). Presence of the 3048 on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Upload: buibao

Post on 22-Mar-2018

217 views

Category:

Documents


2 download

TRANSCRIPT

[CANCER RESEARCH 52, 3048-3051, June I. 1992]

Herbal Medicine Use, Epstein-Barr Virus, and Risk of Nasopharyngeal Carcinoma1

Allan Hildesheim,2 Sheila West, Ernesto DeVeyra, Maria F. De Guzman, Adonis Jurado, Carol Jones, Joko linai,and Yorio Hinuma3

Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 [A. H., C. J.J; Wilmer Eye Institute, TheJohns Hopkins University, School of Medicine, Baltimore, Maryland 21205 ¡S.W.]; Departments of Pathology and Ear, Nose, and Throat, Philippine General Hospital,University of The Philippines, Manila, The Philippines [E. D., M. F. D. G., A. J.J; and Institute for Virus Research, Kyoto University, Kyoto 606, Japan [J. /., Y. H.]

ABSTRACT

Herbal medicine use is thought to be linked to nasopharyngeal carcinoma (NPC) either through its ability to reactivate the Epstein-Barrvirus (EBV) or through a direct promoting effect on EBV-transformedcells. To investigate this, 104 histologically confirmed NPC cases and205 matched controls were studied in The Philippines. Blood was collected to assess antibody titers against EBV, and an interview wasadministered which elicited information concerning herbal medicine useand other risk factors for NPC. Subjects strongly positive for anti-EBVantibodies (Epstein-Barr nuclear antigen |EBNA|) (titers - 1:80) were ata 21-fold excess risk of disease (95% confidence interval, 8.4, 51.8).Herbal medicine use was also associated with NPC (relative risk, 2.5;95% confidence interval, 1.4,4.5). Associations persisted after adjustmentfor education, smoking, Chinese ancestry, and consumption of salted fish.Exposure to herbal medicines among subjects testing negative/weaklypositive for anti-EBNA antibodies was not associated with an elevationin risk (relative risk, 0.6), strong positivity to anti-EBNA antibodies inthe absence of herbal medicine use was associated with a significant 16-fold excess risk of disease, and exposure to herbal medicines amongsubjects testing strongly positive for anti-EBNA antibodies was associated with a significant 49-fold excess risk of NPC when cases werecompared to controls. Similar results were obtained when other serolog-ical measures of EBV exposure were used. Anti-EBV antibody titerswere elevated in herbal medicine users compared to nonusers amongcases but not among control subjects. This suggests that, if herbalmedicines interact with EBV in the development of NPC, they do not doso by reactivating EBV infection but rather through a direct proliferativeeffect on EBV-transformed cells. Although the interaction between EBVand herbal medicines is biologically plausible, larger, more detailedstudies need to be conducted to validate this preliminary finding.

INTRODUCTION

EBV4 has been implicated in the etiology of NPC. EBV DNA

has been detected in nearly all NPC tumor specimens investigated to date (1), and recent evidence has demonstrated thetransforming ability of the EBV latent membrane protein whichis expressed in approximately 65% of NPC tumors (2). Giventhe ubiquity of this virus relative to the incidence of NPC,however, it is likely that other agents act as cofactors in theetiology of this disease. One such factor, phorbol esters presentin extracts from herbal medicines and plants, has been hypothesized to be linked to NPC either through its ability to reactivateEBV infection (3) or through a direct promoting effect on cellstransformed by the EBV (4, 5).

A case-control study of NPC was conducted in a Filipinopopulation in The Philippines to examine risk factors for thisdisease in a country with incidence rates intermediate to that

Received 11/22/91; accepted 3/20/92.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

' Funded by the Philippine National Research Council.2To whom requests for reprints should be addressed, at Environmental Epi

demiology Branch. National Cancer Institute, Executive Plaza North. Room 443,Bethesda. MD 20892.

3 Present Address: Shionogi Institute for Medical Science, Osaka, Japan.4The abbreviations used are: EBV, Epstein-Barr virus; NPC, nasopharyngeal

carcinoma; VCA, viral capsid antigen; EA. early antigen; EBNA. Epstein-Barrnuclear antigen; RR, relative risk: CI. confidence interval.

in Hong Kong (30.0/100,000 among males, 12.9/100,000among females) and the United States (<1.0/100,000 amongmales and females). The age-standardized rate of NPC in RizalProvince in The Philippines in the period of 1978-1982 was4.7/100,000 per year in males and 2.6/100,000 per year infemales (6). As part of our case-control study, information wasobtained concerning the use of herbal medicines, and EBVantibody status was assessed serologically. This has enabled usto examine the hypothesis that EBV and herbal medicinesinteract in the development of NPC.

PATIENTS AND METHODS

Incident cases of histologically confirmed NPC were recruited fromthe Philippine General Hospital. To be eligible for the study, caseswere required not to have had previous treatment for NPC and to haveresided on the island of Luzon for at least 6 months. All subjectspresenting to the clinic with symptoms compatible with NPC andreferred for a nasopharynx biopsy were invited to participate in thestudy. A total of 234 cases suspicious of NPC were identified during a24-month period. Of these, 22 (9%) refused to undergo biopsy and weretherefore considered ineligible for the study. Of the remaining 212, 104individuals were pathologically confirmed to have NPC. All consentedto participate in the study.

Two controls were selected for each confirmed NPC case. Onecontrol was selected from the hospital and a second from the community. Hospital controls were individually matched to cases on sex, age(±5 years), and hospital ward type (private versus public). Hospitalcontrols were chosen from the medical/surgical ward on the same daythat a suspected case was confirmed by biopsy result. Medical rostersroutinely kept by the hospital were used by the study coordinator toselect hospital controls. This was done by reviewing the roster andselecting the first eligible subject as a control. Since this study was alsodesigned to examine the effect of diet on risk of NPC, individuals withdisorders thought to be associated with changes in dietary patterns werenot eligible as controls. Thus, patients with the following diagnoseswere excluded from the study: gastrointestinal cancers, peptic ulcer,chronic cirrhosis, and gallbladder disease. A total of 104 hospitalcontrols were selected, and all agreed to participate. Twenty-four hospital controls had diagnoses of urinary/renal disease, 15 with acuteappendicitis, 11 with non-NPC/gastrointestinal cancers, 9 with respiratory impairment, and the remaining 45 with other conditions.

Community controls were individually matched to cases on sex, age(±5 years), and neighborhood. Selection of neighborhood controls wasachieved by systematically selecting households either to the right orleft of the subject's residence until an appropriate match was found.

Replacement controls were obtained if the original control selectedrefused to participate. Thus, a total of 131 community controls wereascertained, of which 101 (77%) agreed to participate. Among thosewho refused, 76% did so because they refused blood extraction, and theremaining 24% declined all involvement.

A personal interview was administered to consenting subjects in theirnative language, Tagalog, by a trained nurse-interviewer. The interviewobtained information concerning sociodemographic characteristics,Chinese ancestry, adult diet, occupation, smoking, and herbal medicineuse. "Ever use" of herbal medicines was assessed, and for those who

reported use, the names of the medicines used were assessed. Currencyof use and total duration of use were not assessed. In addition, 5 ml ofvenous blood was collected from each consenting subject for anti-EBVantibody determination by immunofluorescence (7). Presence of the

3048

on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HERBAL MEDICINES, EBV, AND NPC

following antibodies was assessed: VCA IgG; VGA IgA; EA IgG; EBNAIgG; EA, diffuse-type IgG; and EA IgA. Titers of antibodies wereexpressed as reciprocals of the maximum dilution of serum (2-foldfrom the initial 1:10 dilution) that gave fluorescence. Samples weretested without knowledge of the case-control status. Strong positivitywas defined as follows: VCA IgG titer > 1:160, VCA IgA titer > 1:10,EA IgG titer > 1:10, EBNA IgG titer > 1:80, EA diffuse-type IgG titer> 1:10, and EA IgA titer > 1:10. These cutoff points were selectedbased on the titer distribution of controls so that as close to half of thecontrol subjects as possible were categorized as strongly positive. Useof other cutoff points did not alter the conclusions.

Anti-EBV antibody geometric mean titers were calculated for casesand controls. The RR, as estimated by the odds ratio, was the measureused to assess associations between risk factors and disease. Significancelevels were determined by use of the 95% CI. Conditional logisticregression was utilized to examine associations between exposures ofinterest and disease, while simultaneously adjusting for potential confounding factors (8). To examine the statistical interaction between twofactors, indicator variables were created for each category of jointexposure and included in the conditional regression model (9). Deviation from an additive interaction was tested for significance utilizingthe test for synergy described by Schlesselman (10). Deviation from amultiplicative interaction was tested for significance by including thecross-product term in the logistic model (11).

Hospital and community controls were compared and found to besimilar with respect to both factors of interest: anti-EBV antibodies

and herbal medicine use. Therefore, these two groups were combinedinto one for purposes of analysis and are henceforth referred to as thecontrol group.

RESULTS

The geometric mean titer of anti-EBV antibodies is presentedin Table 1 for cases and controls. Regardless of the specificantibody test, cases were found to have a higher mean antibodytiter (geometric mean titer ranging from 1:25-1:487) thancontrols (1:5-1:114). The percentage of control samples testingnegative (titer < 1:10) for anti-EBV antibodies ranged from 0-94%, depending on the antibody test used (Table 1). Theequivalent range for cases was 0-20%.

We assessed risk associated with being strongly positive foranti-EBV antibodies by dichotomizing EBV exposure and computing risk estimates, comparing cases to controls. Resultsindicate that individuals strongly positive for anti-EBV antibodies had relative risks ranging from 6-74 when compared tosubjects testing negative/weakly positive (Table 1).

Next, we assessed the association between herbal medicineuse and NPC. Thirty-eight % of cases and 21% of controlsreported having used herbal medicines ¡nthe past. When casesand controls were compared, individuals who reported herbalmedicine use were found to be at a 2.5-fold excess risk of disease(95% CI 1.4, 4.5) compared to individuals who reported neverhaving used herbal medicines. This effect persisted after adjust

ment for potential confounding by education, salted fish consumption, Chinese ancestry, and smoking (RR 2.6; 95% CI 1.4,4.6), factors found in previous studies to be associated withdisease (12-14). The effect was also present when cases werecompared separately to hospital (RR 2.1; 95% CI 1.1, 4.0) andcommunity (RR 2.9; 95% CI 1.5, 5.8) controls.

We attempted to evaluate the specific herbal medicines usedby study subjects to determine whether any specific type wasprimarily involved. This was accomplished by classifying specific herbal medicines reported by subjects into their respectivephylogenetic family. The analysis was complicated by the factthat >40 different types of herbal medicines were reported,most were only used by one or two individuals, and manysubjects reported use of multiple herbal medicines. Of the 40cases and 44 controls who reported ever using herbal medicines,29 cases and 32 controls were able to specify which herbalmedicine(s) they used. We were then able to classify into families all but 6 herbal medicines reported by subjects. A list of themost frequently reported plant families is presented in Table 2.Overall, no specific family of plants appeared to be primarilyresponsible for the association observed between herbal medicine use and NPC. Three cases and no controls reported theuse of plants in the Euphorbiaceae family, and no subjectreported the use of plants in the Thymelaeaceae family, twofamilies of plants known to contain phorbol esters (15).

Table 3 presents results of the analysis which investigatedthe joint effect of EBV and herbal medicines. Only results foranti-EBNA antibodies are presented. EBNA was selected sinceits expression is associated with cell transformation and proliferation (16). However, findings consistent with the ones presented here were also obtained when the other anti-EBV antibody titers were examined. Among subjects testing negative/weakly positive for anti-EBNA antibodies, those who reportedherbal medicine use in the past were not at increased risk ofdisease (RR 0.6; 95% CI 0.1, 3.5), while the converse (strongpositivity for anti-EBNA antibodies in the absence of herbalmedicines) was associated with a 16-fold excess risk of disease(95% CI 5.9, 43.7) when compared to subjects classified asnegative/weakly positive for anti-EBNA antibodies and reporting no herbal medicine use. Subjects who reported herbal medicine use and who were classified as strongly positive for anti-EBNA antibodies were at a 49-fold excess risk of disease (95%CI 13.8, 174.6). This finding suggests a greater than multiplicative interaction between EBV and herbal medicines, althoughthe deviation from the multiplicative model was only marginallysignificant (P = 0.11).

Finally, we examined anti-EBNA antibody titer distributionby herbal medicine use among cases and controls separately.This analysis was performed to determine whether EBV antibody titers were higher among herbal medicine users comparedto nonusers for both study groups. As shown in Fig. 1, an

Table 1 Anti-EBV antibody liters by case status and estimates of risk

Cases Controls

Anti-EBVAntibody:VCA

IgGVCAIgAEAIgGEBNA

IgGEAdiffuse-typeIgGEAIgAGeometric

meantiter1:4871:691:951:1991:631:25%negative"07622015Geometric% strong positive* meantiler929394888085:l

14:8:9:37:5:8%

negative0404049443%strong positiveRRf51606030657U.I14.720.620.873.86.595%CI4.7,

25.95.3,41.46.4,

66.98.4,51.818.1,300.53.1,

13.5" Negativity defined as titers < 1:10.'Strong positivity defined as follows: VCA IgG titer 3: 1:160, VCA IgA titer a 1:10, EA IgG titer > 1:10, EBNA IgG titer > 1:80; EA diffuse-type IgG titer

1:10, and EA IgA titer 2 1:10.c Relative risk estimates for subjects testing strongly positive relative to those testing negative/weakly positive.

3049

on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HERBAL MEDICINES, EBV, AND NPC

Table 2 Plant families most frequently reponed by subjects reporting use ofherbal medicines and respective frequencies, by case status

Familyname:MyrtaceaeLamiaceaeArecaceaeAnonaceaePiperaceaeZingiberaceaeEuphorbiaceaeLythraceaeLeguminosaeMusaceaeTotal

no. ofsubjectsTotalno. who reporteduseTotalno. of subjects re

porting specific herbsNo.

ofcases'93333332221044029No.ofcontrols173111105002054432

'' Several individuals reported use of more than one herbal medicine and may

therefore be listed more than once.

Table 3 Joint effect of EB V and herbal medicinesAdjusted for education, salted fish consumption, and smoking.

Herbal medicine use

EBV: Never Ever

Negative

Positive

1.0°(9/113)'

16.1 [5.9,43.7](55/48)

0.6 [0.1, 3.5]*

(3/30)49.2 [13.8, 174.6]

(37/14)

Test for departure from additive model: P = 0.09Test for departure from multiplicative model: P = 0.11

°RR.*95%CI.' Cases/controls.

upward shift in the antibody titer distribution was observedamong herbal medicine users compared to "never users" for

cases but not for control subjects. Similar findings were observed when antibody titers directed against the other EBVantigens were examined.

DISCUSSION

When specific herbal medicines were examined, no singleplant family appeared responsible for the association observedbetween herbal medicine use and NPC. Although it was expected that many cases would report use of medicinal plantsfrom the Euphorbiaceae and Thymelaeaceae families, both ofwhich are known to contain phorbol esters, only 3 of 40 caseswho were herbal medicine users reported use of medicinal plantsin these families. These results are similar to those reported byYu et al. (21 ), who reported a lack of association between herbalmedicines from the Euphorbiaceae family and NPC. This raisesthe possibility that agents in herbal medicines other than phorbol esters may also be involved in the etiology of NPC. In fact,results from a previous investigation conducted by Zeng et al.(3) suggest that plants from families other than the Euphorbiaceae and Thymelaeaceae families are capable of inducing EBVin vitro.

Of particular note is our finding of an apparent interactionbetween EBV and herbal medicines. Individuals who reportedherbal medicine use in the past but who tested negative/weaklypositive for anti-EBV antibodies were not found to be at anincreased risk of disease, those who reported no previous useof herbal medicines but were strongly positive for anti-EBVantibodies were at a 16-fold excess risk of disease, and thoseindividuals who reported herbal medicine use and who werestrongly positive for anti-EBV antibodies were found to be at a49-fold excess risk of disease when compared to those who

reported being nonusers of herbal medicines and who testednegative/weakly positive for anti-EBV antibodies. Although thepossibility of a chance finding cannot be ruled out given therelatively small size of the present study, the magnitude of theeffect was strong (and marginally significant) and warrantsfurther consideration. Furthermore, the interaction betweenherbal medicines and EBV is biologically plausible, as shown

PROPORTION

Results from the present study conducted on a Filipinopopulation confirm that EBV is strongly associated with NPCin The Philippines (RR 6-74), as has previously been observedin other areas (17, 18). Also, use of herbal medicines was foundto be associated with disease in this population (RR 2.6),independently of educational achievement, smoking, Chineseancestry, and consumption of salted fish. Previous studies whichexamined the association between herbal medicine use and NPCin Chinese populations had conflicting results. While one studyconducted in Taiwan observed an effect of similar magnitudeto that presented here (14), studies conducted by Yu et al. (19-21) in Mainland China failed to detect an independent effect ofherbal medicine use on risk of NPC. Although it is possiblethat the association detected between herbal medicine use andNPC reflects increased use of herbal medicines by cases becauseof their symptoms, the fact that the association was still presentand statistically significant when cases were compared to hospital controls (RR 2.1) suggests that this might not be the case.Nevertheless, an interpretation of causality should be madewith caution. The possibility that herbal medicine use is amarker for a traditional lifestyle which predisposes to NPCcannot be ruled out, despite our attempts to control for possibleconfounding by education, smoking, Chinese ancestry, andconsumption of salted fish. Also, herbal medicines are veryheterogeneous, and it is unlikely that all herbal medicinesincrease risk of NPC.

3050

1:40 1:80 1:160 1:320 1:640 1:1280ANTI-EBNATITER

Q PROPORTION0.5

0.4

1:10 1:20 1:40 1:80 1:160

ANTI-EBNA TITER

1:320 1:640 1:1280

Fig. 1. . I, anti-EBNA antibodies among cases, by herbal medicine use. +,actual data points for herbal medicine users; •¿�actual data points for nonusersof herbal medicines; , curve for herbal medicine users; , curve fornonusers of herbal medicines. It, anti-EBNA antibodies among controls, by herbalmedicine use.

on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HERBAL MEDICINES, EBV, AND NPC

by investigations demonstrating that extracts from herbal medicines and plants are capable of promoting either EBV or EBV-

transformed cells (3, 4). Finally, a previous epidemiologicalstudy conducted in Taiwan (22) revealed an interaction betweenEBV and herbal medicine use which was more than additivebut less than multiplicative.

Our finding that anti-EBV antibody titers were higher amongherbal medicine users compared to nonusers only among casesand not among controls suggests that, if the interaction observed is real, herbal medicines do not interact with EBVthrough an ability to reactivate EBV infection. Rather, it ismore likely that herbal medicines interact with EBV in thedevelopment of NPC through a promotional effect on cellstransformed by the virus. However, it is also possible that theinteraction observed in this study is spurious. It is feasible thatsubjects who are more severely ill are more likely to use herbalmedicines. Also, it has been demonstrated that EBV antibodytiters increase with increasing severity of disease among NPCpatients (16). Therefore, the finding of increased antibody titersamong cases who are herbal medicine users compared to caseswho are not may reflect the increased use of herbal medicinesamong more severely ill individuals rather than a true biologicaleffect.

We were unable to directly test the hypothesis that moreseverely ill NPC patients are more likely to use herbal medicinessince information concerning clinical stage of disease at thetime of diagnosis was not obtained. Also, we are unable toprecisely define the temporal association between EBV andherbal medicine use among control subjects, because no distinction was made between past and current herbal medicineuse. If herbal medicines do indeed reactivate EBV infections,one would expect individuals who are currently using thesemedicines to have higher levels of antibodies than individualswho were users of herbal medicines in the past but who are notcurrent users. Our inability to distinguish these two groups ofindividuals might have masked an association between herbalmedicine use and EBV antibody titers among control subjects.

Other limitations of the present study include the inability toexamine the effect of duration and intensity of use of herbalmedicines on risk of NPC and the relatively small sample sizeof our study. Although 104 cases and 205 controls participatedin the study, the power to detect the observed 5-fold deviationfrom a multiplicative interaction was 60%.

In summary, this study confirms an association of anti-EBVantibodies and herbal medicines with NPC in The Philippines.In addition, an interaction between anti-EBV antibodies andherbal medicines was observed. Although the interaction isbiologically plausible as demonstrated by laboratory studies, weare not able to determine the validity of this finding because ofthe small sample size and the unavailability of detailed information concerning timing and intensity of herbal medicine use.

ACKNOWLEDGMENTS

The authors are grateful to Evelyn Perdi for all blood extraction andinterviewing and Dr. Ernesto Domingo for support and encouragement.We thank Dr. Robert A. DeFilipps of the Smithsonian Institution inWashington, DC, and Dr. James A. Duke of the U.S. Department ofAgriculture in Beltsville, MIX for their assistance in classifying theherbal medicines reported by study subjects.

REFERENCES

1. Evans, A. s.. and Mueller, N. E. Viruses and cancer: causal associations.Annals of Epidemiology, /: 71-92, 1990.

2. Fahraeus, K . Rymo, I Rhim, J. S., and Klein. G. Morphological transformation of human keratinocytes expressing the LMP gene of Epstein-Barrvirus. Nature (Lond.). 345:447-449, 1990.

3. Zeng. Y., Zhong, J. M., Mo, Y. K., and Miao, X. C. Epstein-Barr virus earlyantigen induction in Raji cells by Chinese medicinal herbs. Intervirology, 19:201-204, 1983.

4. Tornei, D. L., Noyes, I., Blocker, D., Holliday, J., and Glaser. R. Phorbolester and Epstein-Barr virus dependent transformation of normal primaryhuman skin epithelial cells. Nature (Lond.), 329: 73-75, 1987.

5. Droms, K. A., and Malkinson, A. M. Phorbol ester-induced tumor promotionby downregulation of protein kinase C. Mol. Carcinogen., 4: 1-2, 1991.

6. Waterhouse, J., and Muir, C. Cancer Incidence in Five Continents, Ed. 4.Lyon, France: International Agency for Research on Cancer, 1982.

7. Imai, J.. and Hinuma, Y. Epstein-Barr virus-specific antibodies in patientswith adult T-cell leukemia (ATL) and healthy ATL virus-carriers. Int. J.Cancer, 31: 197-200, 1983.

8. Lubin, J. H. A computer program for the analysis of matched case-controlstudies. Comput. Biomed. Res., 14: 138-143, 1981.

9. Rothman, K. J. Modern Epidemiology. Boston, MA: Little, Brown, andCompany, 1986.

10. Schlesselman, J. J. Case-Control Studies. Design, Conduct and Analysis.New York: Oxford University Press, 1982.

11. Breslow, N. E., and Day, N. E. Statistical Methods in Cancer Research, Vol.1: The Analysis of Case-Control Studies. Lyon, France: International Agencyfor Research on Cancer, 1980.12. Salted fish and nasopharyngeal carcinoma. Lancet. /'<:841-842, 1989.

13. Mabuchi, K., Bross, D. S., and Kessler. 1.1. Cigarette smoking and nasopharyngeal carcinoma. Cancer (Phila.), 55: 2874-2876, 1985.

14. Lin, T. M., Chen, K. P., Lin, C. C., Hsu, M. M., Tu, S. M., Chiang, T. C,Jung, P. F., and Hirayama, T. Retrospective study on nasopharyngeal carcinoma. J. Nati. Cancer Inst., 51: 1403-1408, 1973.

15. Duke, J. A., and Ayensu, E. S. Medicinal Plants of China, pp. 300-314.Michigan: Algonac, 1985.

16. De-The. G., and Lenoir, G. Comparative diagnosis of Epstein-Barr virus-related diseases: infectious mononucleosis. Burkitt's lymphoma, and naso

pharyngeal carcinoma. In: E. Kurstak and C. Kurstak, Comparative Diagnosis of Viral Diseases, Vol. I. pp. 195-240. New York: Academic Press,1977.

17. Henle. W., Henle. G., Ho, H. C., Burtin, P., Cachin. Y.. Clifford, P., DeSchryver, A., De-The, G., Diehl, V., and Klein, G. Antibodies to Epstein-Barr virus in nasopharyngeal carcinoma, other head and neck neoplasms,and control groups. J. Nati. Cancer Inst., 44: 225-231. 1970.

18. Shanmugaratnam. K. Nasopharynx. In: D. Schottenfeld and J. F. Fraumeni,Cancer Epidemiology and Prevention, pp. 536-553. Philadelphia: W. B.Saunders Company, 1982.

19. Yu, M. C., Huang, T. B., and Henderson. B. E. Diet and nasopharyngealcarcinoma: a case control study in Guangzhou, China. Int. J. Cancer, 43:1077-1082. 1989.

20. Yu, M. C., Ho, J. H. C., Lai, S. H., and Henderson, B. E. Cantonese-stylesalted fish as a cause of nasopharyngeal carcinoma: report of a case controlstudy. Cancer Res., 46: 956-961, 1986.

21. Yu, M. C., Garabrant. D. H.. Huang, T. B., and Henderson, B. E. Occupational and other non-dietary risk factors for nasopharyngca! carcinoma inGuangzhou. China. Int. J. Cancer, 45: 1033-1039, 1990.

22. Lin. T. M., Yang, C. S., Tu, S. M., Chen, C. J.. Kuo. K. C., and Hirayama,T. Interaction of factors associated with cancer of the nasopharynx. Cancer(Phila.). 44: 1419-1423, 1979.

3051

on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1992;52:3048-3051. Cancer Res   Allan Hildesheim, Sheila West, Ernesto DeVeyra, et al.   Nasopharyngeal CarcinomaHerbal Medicine Use, Epstein-Barr Virus, and Risk of

  Updated version

  http://cancerres.aacrjournals.org/content/52/11/3048

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  [email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/52/11/3048To request permission to re-use all or part of this article, use this link

on May 21, 2018. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from