Human Papillomavirus-related Serological Markers of ...cebp.aacrjournals.org/content/cebp/3/4/341.full.pdf · Human Papillomavirus-related Serological Markers of ... cervical carcinoma

Vol. 3, 341-347, June 1994 Cancer Epidemiology, Biomarkers & Prevention 341

Human Papillomavirus-related Serological Markers ofInvasive Cervical Carcinoma in Brazil’

Yeping Sun,2 Jos#{233}Eluf-Neto, F. Xavier Bosch,Nubia Mu#{241}oz,Margaret Booth, Jan M. M. Walboomers,Keerti V. Shah, and Raphael P. Viscidi

Department of Immunology and Infectious Diseases, Johns HopkinsUniversity School of Hygiene and Public Health, Baltimore, Maryland21205 JY. S., K. V. 5.1; Department of Preventive Medicine, Faculdade deMedicina, Universidade de Sao Paulo, 455 Sao Paulo-SP, CEP 01 246,Brazil J. E-N.J; Unit of Field and Intervention Studies, International

Agency for Research on Cancer, Lyon 69372, France IX. B., N. Ml;Epidemiological Monitoring Unit, Department of Epidemiology andPopulation Sciences, London School of Hygiene and Tropical Medicine,

London, WC1 E 7HT, United Kingdom [M. B.); Department of Pathology,Free University Hospital, Amsterdam 1081 HV, the Netherlands

U. M. M. WI; and Department of Pediatrics, Division of PediatricInfectious Diseases, Johns Hopkins University School of Medicine,

Baltimore, Maryland 21205 JR. P. V.1

Abstrad

Masked sera from 1 94 cases and 21 7 controlsparticipating in a case-control study of cervical cancerin Brazil were examined for antibodies to humanpapillomavirus (HPV) 1 6 E6 and E7 byradioimmunoprecipitation assay. Radiolabeled full-lengthE6 and E7 proteins expressed by in vitro transcriptionand translation in rabbit reticulocyte lysate were used asantigens. The antibody prevalences in cases and controlswere: 54.1 % versus 6% for E6; 30.4% versus 4.6% forE7; 63.4% versus 1O.1% for either E6 or E7; and 21.1%versus O.5% for both E6 and E7. The corresponding oddsratios were 35 ([95% confidence interval (Cl)J, 15-83),10 (95% Cl, 4-25), 28 (95% CI, 13-61) and 87 (95%CI, 1 0-736). The most marked contrast between casesand controls was observed for sera with high antibodytiters (cpm > 6000) with an odds ratio of 239 (95% CI,29-1 946) for E6 or E7. Seroreadivity in cases waspartially type specific; women who had HPV-1 6 DNA inthe genital trad had higher antibody prevalence ratesthan those who were negative for HPV DNA. Readivityto the E6 protein was associated with the stage ofdisease; the antibody prevalence was 62.7% in caseswith stages ll-IV and 31 .0% in cases with stage I(P < 0.005). HPV-16 serology and HPV polymerasechain reaction were compared as markers for invasivecervical cancer. The sensitivity and specificity estimateswere: 63.4% and 89.9% for HPV-16 serology; 53.8%

Received 1 2/1 3/93; revised 3/22/94; accepted 3/22/94.‘ The work was supported by USPHS Grant ROl -CA565 14 from the NationalCancer Institute, Conselho Nacional de Desenvolvimento Cientfico e Tec-nologico-Brazil CNPq JEN-204453/88.7, International Agency for Research

on Cancer CNPq 4041 21/89.6-MP, and Funda#{231}#{224}ode Amparo a Pesquisa doEstado de S#{227}oPaulo 90/2319-9.2 To whom requests for reprints should be addressed, at The Johns HopkinsUniversity School of Hygiene and Public Health, Department of Immunology

and Infectious Diseases, 615 North Wolfe Street, Baltimore, MD 21205.

and 94.9% for HPV-16 DNA; 84.4% and 82.7% for anyHPV DNA; and 90.7% and 75.5% for a combination ofHPV-16 serology and any HPV DNA. The specificity forhigh E6 or E7 titers was 99.5%. The HPV-16 E6 and E7serological assays confirm the association of HPVs withcervical cancer previously established by HPV DNAassays. The serological assays may be useful as anadjund to pathological diagnosis and potentially ofvalue for clinical management of HPV-associatedcervical cancers.

Introduction

The predominant role of HPVs3 in the etiology of invasive

cervical carcinoma has been recognized by epidemiologi-cal (1) as well as experimental data (2). Nearly all of the

epidemiological studies have relied on the detection ofHPV DNA in the genital tract as a manker of HPV exposurein their comparisons of case and control women. In mostrecent studies, 70-90% of women with cervical cancer, but

only 5-1 7% of control women, have been reported to haveHPV DNA in the genital tract (1 ). HPV-1 6 has been found

to be the predominant virus in cervical cancer in most partsof the world.

The prevalence of HPV DNA in the genital tract isindicative of either current virus infection or virus-associ-ated abnormality. In contrast, serological markers may beexpected to reflect the entire past history of HPV exposuresof the subject and thus provide additional tools for theinvestigation of cervical cancer. Serological studies of HPVinfections have sought to define markers of HPV infectionand HPV-associated neoplasia. A wide variety of antigens(bacterially expressed or in vitro translated proteins, syn-thetic peptides, viral capsids, etc.) have been used for thispurpose. For most antigens, the specificity or reproducibil-ity of the observed neactivities have been difficult to estab-lish (3). However, one observation has been consistentlyconfirmed in numerous studies. Antibodies to HPV-16 E7,

and to a lesser extent, to HPV-1 6 E6 are found more fre-quently in women with HPV-associated cervical cancer

than in control women (4-10).The main objectives of the present study were to assess

the relationship of HPVs with cervical cancen in Brazil usingHPV-specific serological markers and in addition, to com-

pare serology with the detection of HPV DNA in the genitaltract as a marker for invasive cervical cancer. The serolog-ical markers were also examined for their correlation with

the clinical stage of disease and their HPV type specificity.

We prepared E6 and E7 proteins of HPV-1 6 by in vitro

3 The abbreviations used are: HPV, human papillomavirus; PCR, polymerase

chain reaction; TT-RIPA, transcription and translation radioimmunoprecipi-tation assay; OR, odds ratio; CI, confidence interval.

on June 4, 2019. © 1994 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

http://cebp.aacrjournals.org/

342 HPV Serology of Cervical Cancer in Brazil

transcription and translation, a technique which conservesconformational epitopes and allows for the production offull-length proteins. Antibodies were measured by nadioim-munoprecipitation assay of 35S-labeled proteins. All of the

case and control sema collected in an epidemiological studyof HPV and invasive cervical cancer in Brazil were tested in

a blinded fashion.

Materials and Methods

Study Subjeds. The study design was described in detail ina previous report (1 1 ). In brief, between June 1 990 and June1991, women with a diagnosis of invasive cervical cancerand women selected as controls were recruited from sevenhospitals in Sao Paulo city. The cases were women between

25 and 79 years of age whose diagnosis was confirmed byhistopathology and who had no previous treatment for the

disease. Controls were enrolled from five general hospitalsfrom which the cases were recruited. Controls were fre-quency matched to cases in S-year age groups. Womenwho were admitted for treatment of a gynecological con-

dition or who had a hysterectomy or conization were mel-igible as controls. Women with a psychiatric illness wereineligible as cases or controls. Evidence of a gynecological

on cytological abnormality detected on examination afterrecruitment was not a criterion for exclusion.

All study participants had a pelvic examination per-formed by a gynecologist. At that time, exfoliated cells fromthe ecto- and endocervix were collected for virologicalanalyses. The HPV diagnosis was made, as described, byPCR, using a combination of general and type-specificprimers (1 2). Briefly, the specimens were screened first forthe presence of any HPV using general primers 5 and 6which permit the detection of a broad spectrum of genitalHPVs at the subpicogram level (1 3). The general primer-PCR-positive samples were subjected to HPV type specificPCR using a mixture of HPV-6, HPV-1 6, HPV-33, HPV-1 1-,

HPV-1 8-, and HPV-31 -specific primers. Nucleotide se-

quences of primers and probes used, as well as conditions,have been published (14). To analyze the quality of target

DNA for PCR purposes, samples were also analyzed by PCRusing human g3-globmn gene-specific primers. Only g3-glo-

bin-positive samples were included in the analysis.Serum specimens were collected at the time of the

gynecological examination and stored at -20#{176}C.Two hundred six women with invasive cervical cancer

and 238 control women were eligible; 199 and 225 agreed

to participate. Reasons for nonparticipation were refusal(3 cases, 1 1 controls), death (3 cases), and inability tolocate (1 case, 2 controls). Five cases and eight controlshad no serum collected. The number of participants in-

cluded in the analyses were 194 cases and 217 controls.The 1 94 cases were categorized in clinical stages follow-ing the Federation Intemnationale des Gynaecologistes etObstetmistes classification scheme as stage I (n = 42),stage II (n - 64), and stages lII-IV (n = 70). Eighteen

cases had no stage information.

Detedion of Antibodies to HPV 1 6 E6 and E7 Proteins byTT-RIPA. Antibodies were detected by radioimmunopre-cipitation of full-length HPV-16 E6 and E7 proteins pre-pared by in vitro transcription and translation as describedpreviously (1 5). The specificity of the test has been estab-lished using rabbit antisera to HPV-16 E6 and E7 proteinsand sodium dodecyl sulfate-polyacrylamide gel electro-phoresis analysis of precipitated proteins (15). Briefly,

pGem-1 plasmids containing the E6 or E7 open readingframe of HPV-1 6 (1 6, 1 7) (gifts of Dr. Peter Howley) werelinearized downstream of the T7 promoter with appropriaterestriction enzymes. E6 and E7 mRNA were transcribedwith T7 RNA polymerase in the presence of cap analogue(S’7meGpppG) (Strategene, La Jolla, CA). One pg of RNA

was used to program a micrococcal-nuclease-treated rabbitreticulocyte lysate (Promega, Madison, WI) in the presence

of 40 pCi of E35Slmethionine (Amersham, ArlingtonHeights, IL). Five p1 of serum and S p1 of rabbit reticulocytelysate in 200 p1 of RIPA buffer (10 m�i Tris-HCI, pH 8-140

mM NaCI-2.5 mg/100 ml NaN3-0.1% Nonidet P-40) wereincubated for 3 h at 4#{176}C.Protein A-Sepharose beads (SigmaChemical Co., St. Louis, MO) were added, and the reactionwas further incubated for 1 2 h at 4#{176}C.The beads were

washed twice each with 500 p1 of RIPA buffer, TSA solution(RIPA buffer without Nonidet P-40), and 10 m�i Tris-HCI,

pH 6.8. The beads were then resuspended in 1 00 p1 of 60mM Tris-HCI (pH 6.8), 2 g/l 00 ml sodium dodecyl sulfate,

and 100 mMdithiothreitol, boiled for S mm, and quickly

chilled on ice. The supernatant was transferred to a scintil-lation vial containing 10 ml of Aquasol-2 (DuPont NewEngland Nuclear, Boston, MA), and cpm were measured in

a liquid scintillation counter.

Statistical Analysis. The distributions of cpm in cases andcontrols were compared using the Mann-Whitney test (18).In addition, the sera were categorized as antibody positiveor negative, using as cutoff values the mean plus three SDs

of the control sera after the outliers in the control sera were

excluded (6). Frequencies of antibody-positive sera in dif-ferent groups were compared by the x2 test (1 8). To estimatethe association between a positive serological test and cer-vical cancer, ORs and 95% CIs were calculated as approx-

imations of relative risks using unconditional logistic regres-sion analysis (1 9). ORs were estimated after adjustment for

age, as well as for other variables (socioeconomic status,number of Papanicolaou smear examinations, parity, num-

ber of sexual partners, age at first intercourse, and durationof oral contraceptive use) which have been previously

found to be potential confoundems of the association be-tween HPV and cervical cancer (20). The correlation be-tween E6 and E7 antibodies was expressed as a Pearsoncorrelation coefficient. A test for trend was performedwhere necessary using the method of Mantel (1 8). Thesensitivity and specificity ofthe serological and DNA assays

for invasive cervical cancer were calculated using the

pathological diagnosis as the gold standard.

Results

Antibodies to HPV-1 6 E6 and E7 Proteins in Cases andControls. The distributions of cpm values among cases andcontrols for HPV-16 E6 and E7 are shown in Fig. 1, A andB. The cpm values of case sera against E6 and E7 weresignificantly higher than those of control sera (P = 0.0001

for both E6 and E7). High cpm values (>6000) were foundalmost exclusively in case sera, in 33.5% of the cases andnone of the controls for E6, and in 23.7% of the cases and

0.5% ofthe controlsfor E7 (Paz 0.0001 for both E6 and E7).The cutoff cpm values for categorizing sera as antibodypositive were calculated to be 221 8 for E6 and 2461 for E7.Table 1 shows the antibody prevalence in cases and con-trols to E6, E7, and various combinations of E6 and E7, andthe corresponding ORs for cervical cancer in relation to a

positive serological test. The antibody prevalence was sig-



E

E

A

�

. � �

controls cases

B

100000

10000

1000

100

,00000

10000

1000

100

controls cases

Fig. 1. A, distribution of cpm values for reactivity to HPV-1 6 E6 proteinamong cervical cancer cases and controls. Horizontal line, cutoff value)cpm = 22 1 8) for positivity. B, distribution of cpm values for reactivity toHPV-1 6 E7 protein among cervical cancer cases and controls. Horizontal

line, cutoff value )cpm = 2461 ) for positivity.

Table 1 Prevalence )%) of antibodies to HPV-1 6 E6 and E7 among casesand controls and associated odds ratios for cervical cancer

a OR adjusted for age, socioeconomic status, number of Papanicolaousmears, parity, number of sexual partners, age at first intercourse, and

duration of oral contraceptive use.

Cancer Epidemiology, Biomarkers & Prevention 343

. --, 1fgr’r:. �.:i�11�1�n ‘,�1:r’

Cases

n = 194

Controls

n = 217

OR

)95% Cl)”

Seroreactivity

E6 54.1 6.0 35 (15-83)

E7 30.4 4.6 10 (4-25)

E6 or E7 63.4 10.1 28 (13-61)

E6 and E7 21.1 0.5 87 (10-736)

Titers > 6000 cpm

E6 33.5 0

E7 23.7 0.5 83)10-689)

E6 or E7 40.7 0.5 239 (29-1 946)

E6andE7 16.5 0

nificantly higher in cases as compared to controls for anti-bodies to E6, E7, E6 or E7, and E6 and E7, with ORs rangingfrom 1 0 to 87. The ORs were even higher for sema with high

antibody titers to E6 or E7 (Table 1).

Correlation between Antibodies to HPV-1 6 E6 and E7.Among cases, with an antibody prevalence of 54.1 % for E6and 30.4% for E7, 1 55% of the sera would be expected tohave antibodies to both E6 and E7 if there was no correla-tion between antibodies to the two proteins. The observed

value of 21 .1 % for antibodies to both E6 and E7 (Table 1)was higher than the expected value but the difference wasnot significant (P > 0.10). However, there was a positive

correlation between cpm values for E6 and E7 among the1 23 sema which were antibody positive to either protein(�2 0.454; 95% Cl, 0.301-0.584; P = 0.0001).

Among controls, only 22 sera were antibody positive to

either protein, and all but one had low cpm values (<6000).

Only one serum was antibody positive to both E6 and E7.

There was no significant correlation between the preva-lence of antibodies to, or the cpm values of, the two

proteins.

Relationship between E6 and E7 Antibodies and SelectedRisk Fadors for Cervical Cancer. The distributions of an-tibodies to HPV-16 E6 and E7 proteins were examined by

the key variables previously identified as risk factors for

cervical cancer in Brazil (age, smoking, number of sexual

partners, age at first intercourse, oral contraceptive use, and

parity) (1 1 ). Among cases, antibody prevalences to E6 and

E7 were high but they showed no relationship with any ofthe above risk factors. Among controls, antibody pmeva-

lences to both proteins were low. Antibodies to E7, but not

to E6, decreased with age (P < 0.005; Table 2). Antibody

prevalence to the E7 protein increased with the number ofsexual partners but the trend was not statistically significant.

No clear relationship was observed between antibody prey-alence in the controls and age at first intercourse, use of oralcontraceptives, smoking, or parity (data not shown).

Relationship between Antibodies to HPV-l 6 E6 and E7 andHPV DNA Diagnoses. It was of interest to know if antibod-ies to HPV-1 6 E6 and E7 proteins were specific for infectionwith HPV-16. Therefore, distributions of cpm values and

antibody prevalence were examined by three HPV diagnos-tic categories: HPV-16 positive; other HPV positive; andHPV negative (Fig. 2, A and B). For E6, the cpm values of

the HPV-16-positive cases were significantly higher than

that of other HPV-positive (P = 0.001 ) as well as HPV-negative cases (P = 0.004). For E7, the cpm values ofHPV-1 6-positive cases were significantly higher than that ofHPV-negative cases (P = 0.025) but not that of other HPV-

positive cases (P = 0.1 9). Table 3 shows the prevalences ofantibodies to E6 and E7 in case sera by HPV diagnosis.

Analysis of frequencies revealed significant differences be-tween the diagnostic categories in reactivity to E7, E6 and

E7, and E6 or E7 (P< 0.01 for all comparisons), but not inreactivity to E6 (P = 0.1 03). The greatest difference betweenthe diagnostic categories was observed for sena with a hightiter of antibody to E6 (P < 0.0001 ). For each protein andcombination of proteins, the highest seroreactivity was ob-served among HPV-1 6-positive cases, and the reactivity ofother HPV-positive and HPV-negative cases were not sig-nificantly different.

Among controls, antibodies to either E6 or E7 wereobserved in 1 of 1 0 subjects who were HPV-1 6 positive, in3 of 22 subjects who were positive for other HPVs, and in

1 5 of 1 60 subjects who were HPV negative. The antibodyfrequencies between diagnostic categories among controlswere not significantly different.

Relationship between E6 and E7 Antibodies and ClinicalStages of Cervical Cancer. The conrelation between sero-reactivity of cervical cancer patients to HPV-1 6 E6 and E7protein and clinical stage of disease was examined. Fig. 3,A and B show the distributions of cpm values of case serafor E6 and E7 by clinical stages (stage I, stage II, and stageslll-IV). The cpm values of E6 in more advanced stages weresignificantly higher than that in stage I (P = 0.01 for stage Iversus stage II; P = 0.005 for stage I versus stages lll-IV), buttheme was no significant difference in the distributions ofcpm values between stage II and stages Ill-lV (P = 0.61 ). Incontrast, the distributions of cpm values of E7 did not show

any significant difference between clinical stages. Table 4shows the prevalence of antibodies to HPV-1 6 E6 and E7,



100000 � A

C

EC-ci)

U,

C

0(a

%�

: �;10000

1000

.�c

t

ci)

C

� 10000.

ci)aU,

CJ

� 1000.

100.

HPV-16 otherHPV noHPV

Fig. 2. A, distribution of cpm values for seroreactivity to E6 among case

sera by HPV DNA diagnosis. Horizontal line, cutoff value (2218) for posi-tivity. B, distribution of cpm values for seroreactivity to E7 among case seraby HPV DNA diagnosis. Horizontal line, cutoff value (2461 ( for positivity.


Table 2 Prevalen cc of antibodies to H PV-1 6 E6 and E7 amo ng cases and contro Is by age and numbe r of sexual partners

Cases Controls

Risk factorsn=194 n�217

Number

tested

E6 )+)“

%

E7 )+)%

Number

tested

E6 )+)

%E7 )+)

0/

Age

25-34 12 66.7 41.7 13 7.7 23.1

35-44 43 46.5 25.6 47 6.3 8.5

45-54 60 55.0 28.3 63 6.4 1.6

55-64 46 54.3 37.0 58 6.9 3.4

65+ 33 57.6 27.3 36 2.8 0

P)�2fontrend) >0.10 >0.10 >0.10 <0.005

Number of sexual partners

0 7 14.3 01 103 50.5 29.1 141 5.0 5.7

2-3 61 59.0 34.4 57 70 04+ 30 56.7 26.7 12 8.3 16.7

P)�2fortrend) >0.10 >0.10 >0.10 >0.10

,, )+), positive.

;��! .:�i;; �:

100HPV-16 otherHPV noHPV

100000. B

j.�:.� a �p

#{149}:�?�#{176}�‘

and of high titers of antibodies to E6 and E7 in case sera by

clinical stage of disease. Prevalence of antibodies to E6 inmore advanced stages was significantly higher than that in

stage I (stage I, 31 .0%; stage II, 62.5%; and stages lll-IV,62.9%; x2 for trend = 9.06; P< 0.005), but it did not differbetween stage II and stages Ill-IV (P = 0.996). High cpmvalues to E6 were found more frequently in advanced stages

of disease (stage I, 1 9.0%; stage II, 35.9%; and stages lIl-IV,42.9%; x2 for trend = 6.1 7; P < 0.02). For E7, neither the

Table 3 Antibod y prevalence (%) in case sera b y HPV DNA diagnoses”

HPV-16

n�’96OtherHPVb

n=S7NoHPVn=29

E6 61 .5 49.1 41 .4 0.103

E7 42.7 21.1 13.8 0.002

E6 or E7 72.9 59.6 41 .4 0.006

E6andE7 31.3 10.5 13.8 0.006

E6 cpm > 6000 51 .0 1 9.3 1 0.3 <0.0001

E7cpm>6000 35.4 12.3 13.8 0.002

.‘ 1 2 cases that were )3-globulin negative were excluded from this analysis.bOther HPV, HPV-18 (n = 16); HPV-31 )n = 2); HPV-33 (n = 2); HPV-

31/33 )n = 2); and HPV unknown )n = 35).

prevalence of antibodies nor the magnitude of cpm valuesshowed any association with clinical stages.

The above analyses were repeated for the 88 caseswhich had HPV-1 6 DNA in the genital tract. The resultswere similar to those obtained with all 1 76 cases andshowed that antibodies to E6, but not to E7, increased in the

advanced stage of disease (data not shown).

Comparison of Antibodies to HPV-1 6 E6 and E7 with HPVDNA Diagnosis as Markers of Cervical Cancer. HPV DNAdiagnoses of genital tract specimens were available for thestudy subjects. Therefore, it was possible to compareHPV-1 6 E6 and E7 antibodies in the serum with HPV-1 6

DNA or any HPV DNA in the genital tract as markers forinvasive cervical cancer. Both serology and HPV DNAdiagnosis showed a significant association between a pos-itive test and invasive cervical cancer (OR = 28 for anti-bodies to E6 or E7, 71 for HPV-16 DNA, and 33 for anyHPV DNA) (1 1 ). Assuming that all cervical cancer cases are

associated with HPV, the assays were evaluated as diag-nostic tests for invasive cancer (Table 5). The HPV-1 6 se-rological assay detected more cases than HPV-1 6 DNA

diagnosis but fewer than the generic HPV DNA diagnosis.The combination of serology and DNA diagnosis resulted in

the detection of HPV-associated cervical cancer in morecases (91 %) than HPV DNA diagnosis alone (84%). HPV-1 6DNA assay had the highest specificity (94.9%), followed byHPV-1 6 serology (89.9%), and any HPV DNA assay

(82.7%). HPV-1 6 serology for high antibody titers to E6 orE7 had a sensitivity of 40.7% and a specificity of 99.5%.



Table 5 Comparison of serology and HPV DNA diagnosis as marker

for cervical cancerci)

ciccEC-ci)0.U,

Cci0LU

ci)

ciC

EC-ci)aU)

Cci00

10000.

1000.

100.

100000�

10000

1000.

100.

.:t � � �

stage I stage (I stage )l(-(V

B -.

� � .

Stage ) Stage II Stage (((-(V

Fig. 3. A, distribution of cpm values for seroreactivity to E6 among case

sera by clinical stages. Horizontal line, cutoff value (cpm = 221 8) forpositivity. B, distribution of cpm values for seroreactivity to E7 among case

sera by clinical stages. Horizontal line, cutoff value (cpm = 2461 ) forpositivity.

Table 4 Antibody prevalence (%( by clinical stages(Federation Internationale des Gynaecologistes et Obstetristes)

a Eighteen cases that have no information on stage were excluded from this

analysis.


100000 A

. .Seroreactuvity

Clinical stage”PfortrendI

n=42

II

n=64

Ill-tV

n=70

E6 31.0 62.5 62.9 <0.005

E7 26.2 28.1 32.9 >0.10

E6 cpm > 6000

E7cpm>6000

19.0

21.4

35.9

20.3

42.9

27.1

<0.02

>0.10

Discussion

The E6 and E7 proteins of HPVs are responsible for thetransforming activities of oncogenic HPVs, and they are

expressed consistently in cells of HPV-associated cancers(1 6, 1 7, 21-26). They are also expressed in the course oftransient cytolytic HPV infections which do not lead to

malignancy. Previous investigations have shown that anti-bodies to E6 and E7 proteins are found in small proportionsof normal controls and cervical intraepithelial neoplasiagrade Ill cases, and that the prevalences and titers of these

antibodies increase in women who have invasive cervical

carcinoma (4-8, 10, 27, 28).

The major aims of our study were to assess the rela-

tionship of HPVs with cervical cancer using serological

Sensitivity Specificity

E6 or E7 abs’ 63.4 89.9

E6 or E7 abs (high titers) 40.7 99.5

HPV-16 DNA 53.8 94.9

Any HPV DNA 84.4 82.7

HPV-16 abs or any HPV DNA 90.7 75.5

a Abs, antibodies.

markers and to evaluate antibodies to E6 and E7 proteins asmarkers for invasive cervical cancer. We tested all of thesera collected in a case-control study of invasive cervical

cancer in Brazil for reactivity to in vitro translated HPV-1 6E6 and E7 proteins. We chose E6 and E7 proteins of HPV-1 6because HPV-1 6 was the predominant type in cervicalcancers in the Brazil study and accounted for 54% of thecases (1 1 ). The serum reactivities of case and control serawere analyzed in two ways: (a) by a comparison of the

distributions of the cpm values; and (b) by a comparison of

the antibody prevalence rates calculated on the basis ofappropriate cutoff values. The results of these two analyseswere consistent with one another and showed marked dif-

ferences between cases and controls. The ORs for the as-

sociation of cervical cancer seroreactivity to E6, E7, eitherE6 on E7, and both E6 and E7 ranged from 10 to 87. The

differences between cases and controls were most impres-sive where the frequencies of sera with high cpm values

were compared; 40.7% of case sera, as compared to 0.5%of control sera, had high titers to either protein, giving anOR of 239 (95% Cl, 29-1 946). These findings support andconfirm the association of HPV with invasive cervical can-cer that was previously established by case-control studiesusing HPV DNA diagnosis.

Higher prevalences of E6 and E7 antibodies in invasivecervical cancer cases as compared to controls have beenobserved in studies which used a variety of E6 and E7immunological reagents. However, as compared to otherreagents, in vitro translated proteins showed a greater abil-ity to distinguish between cases and controls. The differ-ence between reagents was best illustrated by tests with theE6 protein. In our previous study with E6 peptides, theantibody prevalences in cases and controls were 1 6 and4%, respectively (6). In a recent report, reactivity to bacte-

rially expressed E6 proteins was observed in 33% ofthe serafrom cervical cancer patients and 3% of control sera (1 0). In

comparison, the reactivity to E6 in TT-RIPA was 54%among cases (62% among HPV-1 6 DNA-positive cases)

and 6% among controls in the present study, and 56%among cases of HPV-1 6-associated cervical cancers and1 .7% among controls in Colombia and Spain (1 5). The verylikely reason for the greater reactivity of sera in IT-RIPA is

the potential of the assay to measure antibodies to confor-mational epitopes.4 The high prevalence of antibody to E6protein indicates that the protein is expressed and is seen by

the immune system of cancer patients. The immunogenicityof the HPV-1 6 E6 protein in cancer patients has implica-tions for vaccine development and suggests that current

4 Y. Sun, K. V. Shah, M. Muller, N. Mu#{241}oz,F. X. Bosch, and R. P. Viscidi,unpublished data.




efforts to develop E7-based vaccines should be extended toinclude the E6 protein.

Among controls, the antibody prevalence to E6 and E7was low and the positive sera had low titers. These reac-tivities are very likely the result of past infections withHPV-16 or related viruses and should therefore be more

frequently found in women who are at increased risk for

genital HPV infections. Antibody prevalence to E7 in-creased with the number of sexual partners but the trendwas not statistically significant. The low antibody preva-

lence may have reduced the ability to detect associations.The decrease in E7 antibody prevalence by increase in agemight suggest that antibody titers decline with age. Sinceour controls were hospitalized patients, it is possible thatthe findings do not apply to the general population. It willbe possible to evaluate the specificity of the low levels of E6

and E7 antibodies by examining ifthese specimens are also

reactive with HPV-16 capsids. A serological assay withHPV-16 capsids, which has been described recently (29),

appears to be a marker for past infections.Theme was a degree of type specificity in the E6 and E7

antibody response when it was examined by comparisonsofthe response in cases in three diagnostic categories: those

who had HPV-1 6 DNA in the genital tract; those who hadother HPV DNA; and those who had no HPV DNA in thegenital tract. The type specificity of the E7 response was

evidenced both in a comparison of antibody prevalencerates (P - 0.002) and the prevalence of sera with high titers(P = 0.002). With respect to E6, significant differences inantibody prevalence rates in the three diagnostic categorieswere found only in sera with high titers (P< 0.0001). It ispossible that sera with high titers to E6 and E7 in casewomen who were diagnosed as HPV negative (for example,four high E7 values in Fig. 2B) represent misclassification bythe DNA assays of cervical scrapes. For two of these fourcases, tumor biopsies, in addition to cervical scrapes, were

tested for HPV DNA; HPV-16 DNA was found in one andthe other was HPV negative.

Antibodies to E6 appeared to be a marker for an ad-vanced stage of disease. The prevalence as well as titers ofantibodies to E6 increased significantly in the higher gradesof disease. These results suggest that antibodies to HPV-1 6E6 might be a marker of invasiveness or tumor mass. It ispossible that E6 antibody assays would be useful as a prog-nostic marker and for monitoring response to treatment inwomen with invasive cancer, but this remains to be evalu-ated in prospective studies. Antibodies to E7 were not re-lated to disease stage in our study.

For epidemiological and clinical studies of HPV-asso-

ciated invasive cervical cancer, a test with a high degree ofsensitivity and specificity for HPV and invasive cancer mayhave some utility as an adjunct to pathology for the diag-nosis of cancer, and as a means for determining the pro-portion of cancers associated with HPV. In our study, wehad the opportunity to compare HPV serology and HPV

PCR for their correlation with invasive cervical cancer.HPV-1 6 serology was more sensitive than the HPV-1 6 DNAassay, probably because serology, but not the DNA diag-nosis, identifies infection with closely related viruses. The54% of cases that were positive by HPV-1 6 DNA PCRprobably represent all, or nearly all, of the cancers that areHPV-1 6 associated. As compared to PCR using a genericprobe, HPV-1 6 serology was less sensitive, probably be-

cause serology cannot identify infections with distantly me-

lated oncogenic viruses. The most sensitive measurement

was a combination of DNA diagnosis and HPV-1 6 serology.The sensitivity may increase if serological assays using E6and E7 proteins of other oncogenic HPVs were used. Thespecificity of the various assays ranged from 95% forHPV-1 6 DNA to 76% for the combination of any HPV DNA

and HPV-1 6 serology. The presence of HPV DNA in a smallproportion of controls is not unexpected and probably rep-

resents women with active or chronic HPV infection. Thelow prevalence of antibody in controls most likely results

from past infection with HPV-1 6 or related viruses.Serology for HPV-transforming proteins, in conjunc-

tion with HPV DNA diagnosis, appears to be a promisingstrategy to identify HPV-associated invasive cancers. Futurestudies should clarify the type specificity of the antibodyresponses and establish more precisely the relative perfor-mance of serology and DNA diagnosis for the detection ofinvasive cervical cancer. The 99.5% specificity of high titersof E6 or E7 antibody responses for invasive cancer suggeststhat the assays may be useful in epidemiological studies

examining the role of HPVs in other cancer, such as vulvar,penile, and oral carcinomas. The serological tests shouldalso be evaluated as prognostic indicators and for monitor-ing of response to treatment.

References1 . Munoz, N., and Bosch, F. x. HPV and cervical neoplasia: review of

case-control and cohort studies. In: N. Munoz, F. x. Bosch, K. V. Shah, andA. Meheus (eds.), The Epidemiology of Human Papillomavirus and CervicalCancer, pp. 251-261. IACR Scientific Pub. No. 119. Lyon, France: Interna-tional Agency for Research on Cancer, 1992.

2. zur Hausen, H. Human papillomaviruses in the pathogenesis of anogeni-tal cancer. Virology, 184:9-13, 1991.

3. Viscidi, R. P., and Shah, K. V. Immune response to genital tract infectionwith human papillomavirus. In: J. I. Gallin and A. S. Fauci (eds.), Advances

in Host Defense Mechanisms, Vol. 8, pp. 239-260, New York: Raven Press,Ltd., 1993.

4. Jochmus-Kudielka, I., Schneider, A., Braun, R., Kimmig, R., Koldovsky,U., Schneweis, K. E., Seedorf, C., and Gissmann, L. Antibodies against thehuman papillomavirus type 16 early proteins in human sera: correlation ofanti-E7 reactivity with cervical cancer. J. NatI. Cancer Inst., 81: 1698-1 704,

1989.

5. KOcheI, H. G., Sievert, K., Monazahian, M., Mittelstadt-Deterding, A.,Teichmann, A., and Thomssen, R. Antibodies to human papollomavirus

type-i 6 in human sera as revealed by the use of prokaryotically expressed

viral gene products. Virology, 182: 644-654, 1991.

6. Muller, M., Viscidi, R. P., Sun, Y., Guerrero, E., Hill, P. M., Shah, F.,Bosch, F. X., Mu#{241}oz,N., Gissmann, L., and Shah, K. V. Antibodies to HPV-1 6E6 and E7 proteins as markers for HPV-16-associated invasive cervicalcancer. Virology, 187: 508-514, 1992.

7. Kanda, T., Onda, T., Zanma, S., Yasugi, T., Furuno, A., Watanabe, S.,Kawana, T., Sugase, M., Ueda, K., Sonoda, T., Susuki, S., Yamashiro, T.,

Yoshikawa, H., and Yoshiike, K. Independent association of antibodies

against human papillomavirus type 16 E1/E4 and E7 proteins with cervical

cancer. Virology, 190: 724-732, 1992.

8. Onda, T., Kanda, T., Zanma, S., Yasugi, T., Watananbe, S., Kawana, T.,

Ueda, K., Yoshikawa, H., Taketani, Y., and Yoshiike, K. Association of theantibodies against human papillomavirus 1 6 E4 and E7 proteins with cervicalcancer positive for human papillomavirus DNA. tnt. J. Cancer, 54: 624-628,1993.

9. Jha, P. K., Beral, V., Peto, J., Hack, S., Hermon, C., Deacon, J., Mant, D.,

Chilvers, C., Vessey, M. P., Pike, M. C., Muller, M., and Gissmann, L.Antibodies to human papillomavirus and to other genital infectious agentsand invasive cervical cancer risk. Lancet, 341: 1116-1118, 1993.

10. Ghosh, A. K., Smith, N. K., Stacey, S. N., Glew, S. S., Connor, M. E.,

Arrand, J. R., and Stern, P. Serological responses to HPV 1 6 in cervical

dysplasia and neoplasia: correlation of antibodies to E6 with cervical cancer.nt. J. Cancer, 53: 591-596, 1993.

1 1 . Eluf-Neto, J., Booth, M., Mu#{241}oz,N., Bosch, F. x., Meijer, C. J. L. M., andWalboomers, J. M. M. Human papillomavirus and invasive cervical cancerin Brazil. Br. J. Cancer, 69: 114-119, 1994.




12. Van den Brule, A. J. C., Meijer, C. J. L. M., Bakels, V., Kenemans, P., andWalboomers, J. M. M. Rapid human papillomavirus detection in cervical

scrapes by combined general primer-mediated and type-specific polymerase

chain reaction. J. Clin. Microbiol., 28: 2739-2742, 1990.

13. Snijders, P. J. F., Van den Brule A. J. C., Schrijnemakers, H. F. I., Snow,C., Meijer, C. J. L. M., and Walboomers, J. M. M. The use of general primersin the polymerases chain reaction permits the detection of a broad spectrum

of human papillomavirus genotypes. J. Gen. Virol., 71: 173-181, 1990.

14. Walboomers, J. M. M., Melkert, P. W. J., Van den Brule, A. J. C., Snijders,P. J. F., and Meijer, C. J. L. M. The polymerase chain reaction for humanpapillomavirus screening in diagnostic cytopathology of the cervix. In: C. S.Herrington and 0. 0. McGee (eds.), Diagnostic Molecular Pathology, Vol. 2,pp. 153-172. Oxford: RI Press, 1992.

15. Viscidi, R. P., Sun, Y., Tsuzaki, B., Bosch, F. X., Mu#{241}oz,N., and Shah,K. V. Serologic response to HPV-associated invasive cervical cancer. Int. J.

Cancer, 55: 780-784, 1994.

16. Dyson, N., Howley, P. M., MUnger, K., and Harlow, E. The humanpapilloma virus-i 6 E7 oncoprotein is able to bind to the retinoblastoma gene

product. Science (Washington DC), 243: 934-937, 1989.

17. Werness, B. A., Levine, A. 1., and Howley, P. M. Association of humanpapillomavirus type 1 6 and 1 8 E6 proteins with p53. Science (Washington

DC), 248: 76-79, 1990.

18. Kirkwood, B. R. Essentials of Medical Statistics. Oxford: Blackwell

Scientific Publications, Ltd., 1988.

19. Breslow, N. E., and Day, N. E. Statistical methods in cancer research: theanalysis of case-control studies. In: N. E. Breslow and N. E. Day (eds.),Unconditional Logistic Regression for Large Strata, IARC Scientific Pub. No.32, pp. 1 92-246. Lyon, France: International Agency for Research on can-

cer, 1980.

20. Franco, E. L. The sexually transmitted disease model for cervical cancer:incoherent epidemiologic findings and the role of misclassification of human

papillomavirus infection. Epidemiology, 2: 98-1 06, 1991.

21 . Smotkin, D., and Wettstein, F. 0. Transcription of human papillomavirus type 16 early genes in a cervical cancer and a cancer-derived cellline and identification of the E7 protein. Proc. NatI. Acad. Sci. USA, 83:4680-4684, 1986.

22. Baker, c. C., Phelps, W. C., Lindgern, V., Braun, M. J., Gonda, M. A.,

and Howley, P. M. Structural and transcriptional analysis of human papil-

lomavirus type-16 sequences in cervical carcinoma cell lines. J. Virol., 61:

962-971, 1987.

23. Shirasawa, H., Tomita, Y., Kubota, K., Kasai, T., Sekiya, S., Takmizawa,H., and Simizu, B. Transcriptional differences of the human papillomavirustype-i 6 genome between precancerous lesions and invasive carcinomas.J. Virol., 62: 1022-1027, 1988.

24. MOnger, K., Werness, B. A., Dyson, N., Phelps, W. C., Harlow, E., andHowley, P. M. Complex formation of human papillomavirus E7 proteins with

the retinoblastoma tumor suppressor gene product. EMBO J., 8: 4099-4105,

1989.

25. Hiraiwa, A., Kiyono, T., Segawa, K., Utsumi, K. R., Ohashi, M., and

Ishibashi, M. Comparative study on E6 and E7 genes of some cutaneous and

genital papillomaviruses of human origin for their ability to transform 3Y1

cells. Virology, 192: 102-111, 1993.

26. Huang, P. 5., Patrick, D. R., Edwards, G., Goodhart, P. J., Huber, H. E.,

Mile, L., Garsky, V. M., Oliff, A., and Heimbrook, D. C. Protein domains

governing interaction between E2F, the retinoblastoma gene product, andhuman papillomavirus type 16 E7 protein. Mol. Cell. Biol., 13: 953-960,1993.

27. Sasagawa, T., Inoue, M., Tanizawa, 0., Yutsudo, M., and Hakura, A.

Identification of antibodies against human papillomavirus type 1 6 E6 and E7

proteins in sera of patients with cervical neoplasia. Jpn. J. Cancer Res., 83:705-713, 1992.

28. Stacey, S. N., Bartholomew, J. S., Ghosh, A., Stern, P. L., Mackett, M.,

and Arrand, J. R. Expression of human papillomavirus type 1 6 E6 protein byrecombinant baculovirus and use for detection of anti-E6 antibodies inhuman sera. J. Gen. Virol., 73: 2337-2345, 1992.

29. Kirnbauer, R., Taub, J., Greenstone, H., Roden, R., Durst, M., Gissmann,

L., Lowy, D. R., and Schiller, J. T. Efficient self-assembly of human papillo-mavirus type 16 Li and Li plus L2 into virus-like particles. J. Virol., 67:

6926-6936, 1993.



1994;3:341-347. Cancer Epidemiol Biomarkers Prev Y Sun, J Eluf-Neto, F X Bosch, et al. cervical carcinoma in Brazil.Human papillomavirus-related serological markers of invasive

Updated version

http://cebp.aacrjournals.org/content/3/4/341

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://cebp.aacrjournals.org/content/3/4/341To request permission to re-use all or part of this article, use this link



http://cebp.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Human Papillomavirus-related Serological Markers of ...cebp.aacrjournals.org/content/cebp/3/4/341.full.pdf · Human Papillomavirus-related Serological Markers of ... cervical carcinoma