Human papillomavirus DNA and antibodies to human papillomaviruses 16 E2, L2, and E7 peptides as...

Human Papillomavirus DNA and Antibodies to HumanPapillomaviruses 16 E2, L2, and E7 Peptides as

Predictors of Survival in Patients With Squamous CellCervical Cancer

By Pau Viladiu, Francesc-Xavier Bosch, Xavier Castellsagu6, Nubia Muiioz, Josep Maria Escrib6, Eva Hamikova,Vanda Hofmannov6, Eloisa Guerrero, Angel Izquierdo, Carmen Navarro, Pilar Moreo, Isabel Izarzugaza,

Nieves Ascunce, Miguel Gili, Maria Teresa Muhioz, Luis Tafur, Keerti V. Shah, and Vladimir Vonka

Purpose: To assess whether human papillomavirus(HPV) DNA detection in cervical cancer specimens, or an-tibodies to selected HPV 16 peptides are predictors oftumor recurrence and long-term survival in patients withsquamous cell invasive cervical cancer.

Subjects and Methods: Four hundred seventy-onecases included in two population-based case-controlstudies underwent follow-up evaluation. The survivaland cause of death were ascertained for 410 cases(87%), with a median follow-up time of 4.6 years afterdiagnosis. HPV DNA was assessed using an L1 polymer-ase chain reaction (PCR)-based system and Southern hy-bridization (SH) on scraped cytologic specimens or biop-sies. HPV 16 antibodies to E2, L2, and E7 peptides weredetected with enzyme-linked immunosorbent assay(ELISA).

Results: Clinical stage was the only independent

UMAN PAPILLOMAVIRUSES (HPV) have been

HX shown in epidemiologic studies to be the centralrisk factor for cervical cancer.'3 DNA hybridization andpolymerase chain reaction (PCR) methods have been cru-

From the Institut CatalI d'Oncologia, Servei d'Epidemiologia iRegistre del Cancer, Barcelona; Hospital de Santa Caterina, Gir-ona; Consejeria de Sanidad y Politica Social, Murcia; Servicio Pro-vincial de Sanidad, Zaragoza; Departamento de Sanidad y Consumo,Vitoria-Gasteiz; Programa de Cdncer de Mama, Pamplona; Facul-tad de Medicina, Universidad de Sevilla, Sevilla; Servicio Territorialde Sanidad y Bienestar Social, Salamanca, Spain; InternationalAgency for Research on Cancer, Lyon, France; Institute of Hematol-ogy and Blood Transfusion, Prague, Czech Republic; The JohnsHopkins University School of Hygiene and Public Health, Baltimore,MD; and Universidad del Valle, Cali, Colombia.

Submitted February 21, 1996; accepted July 12, 1996.Supported in part by grants from the International Agency for

Research on Cancer, France, the Fondo de Investigaciones Sanita-rias of the Spanish Government 9FIS 94/1728), Spain, and the Grant-ing Agency of Czech Ministry of Health (grant no. 1914-3), CzechRepublic. J.M.E. was supported by a fellowship from the Fundaci6August Pi i Sunyer at the Ciutat Sanitaria i Universitaria de Bell-vitge, Hospitalet de Llobregat, Spain.

Address reprint requests to F. Xavier Bosch, MD, Institut Catalad'Oncologia, Servei d'Epidemiologia i Registre del Cancer, GranVia s/n, km 2.7, E-08907 Hospitalet Llobregat, Barcelona, Spain;Email [email protected].

© 1997 by American Society of Clinical Oncology.0732-183X/97/1502-0021$3.00/0

prognostic factor for recurrence or survival. Although se-ropositivity to HPV 16 E7/3 peptide predicted a twofoldexcess risk of mortality (adjusted hazards ratio [HRa] =2.0; 95% confidence interval [CI], 1.2 to 3.3), the associa-tion was restricted to stage I (HRa = 6.6; 95% Cl, 1.2 to37.6) and II (HRa = 5.9; 95% Cl, 2.1 to 16.5) patients.The presence of HPV DNA (HRa = 0.9; 95% Cl, 0.5 to1.5), different estimates of the HPV viral load and theHPV type identified were not predictors of tumor recur-rence or survival.

Conclusion: The presence of antibodies to HPV 16 E7proteins is of prognostic value in early-stage cervicalcancer. Our results provide strong evidence that detec-tion and typing of HPV DNA in cervical cells or tissues isnot a prognostic factor for recurrence or survival.

J Clin Oncol 15:610-619. © 1997 by American So-ciety of Clinical Oncology.

cial in establishing this relationship. The reported preva-lence of HPV DNA in cervical cancer specimens in-creased from 30% to 50% in studies reported in the early1980s4 to 90% and greater in studies reported more re-

cently.5 Several investigators have explored the prognos-tic value of HPV DNA detection with inconsistent results.

Some reports indicated that patients with HPV DNA-negative cancers had worse survival than HPV-positivepatients,6 s whereas others reported no differences in sur-vival9 '0 or in stage at diagnosis." More specifically, somestudies related poor survival to the presence of HPV 1612or of HPV 1813 in cervical cancers. Only one study hasbeen reported that related survival to the presence orabsence of antibodies to HPV proteins.'4

We report here on the 3- and 5-year survival experienceof 471 incident cases of squamous cell invasive cancerin which the presence of HPV DNA was detected usingL1-based PCR or Southern hybridization (SH).15 Seropos-itivity to HPV 16 E2-, L2-, and E7-derived peptides wasdetermined using enzyme-linked immunosorbent assay(ELISA).'6

SUBJECTS AND METHODS

Between 1985 and 1988, the International Agency for Researchon Cancer (IARC) conducted population-based case-control studiesin Spain and Colombia to investigate the role of HPV in the etiologyof cervical intraepithelial neoplasia (CIN) III/carcinoma-in-situ andinvasive squamous cell cervical cancer. Each participant completed

Journal of Clinical Oncology, Vol 15, No 2 (February), 1997: pp 610-619610

128.255.6.125Information downloaded from jco.ascopubs.org and provided by at The University of Iowa Libraries on February 18, 2015 from

Copyright © 1997 American Society of Clinical Oncology. All rights reserved.

HPV AND CERVICAL CANCER SURVIVAL

a comprehensive questionnaire on lifetime exposure to the estab-lished and suspected risk factors for cervical cancer. In addition,

each participant contributed a 10-mL blood sample and two cervical

scrapes. Fresh-frozen tissue biopsies were available for many of the

patients for additional HPV DNA studies. A paraffin section from

each case was submitted to a panel of three expert cytopathologists

for diagnosis confirmation, cell type, and differentiation degree clas-

sification (for full description of the panel review, see Kato et al' 7).

The main results of the case-control studies have been reportedpreviously. z', 5

In September 1992, the vital status of the cases was assessed by

searches in the medical records of the hospitals in which the patients

had been treated. If no information was obtained, the names of thesubjects were searched in the local cancer registries, the local mortal-

ity files, and exceptionally by contacting the person or some family

member. From the records, and when necessary by personal inter-

view with the treating physician, information was obtained concern-

ing stage at diagnosis according to the Fdration Internationale de

Gyncologie et d'Obstetrique (FIGO) and to International Union

Against Cancer (UICC) tumor-node-metastasis (TNM) classifica-

tion, type of treatment (surgery, radiotherapy, chemotherapy, andother), time to recurrence, and date and cause of death if applicable.

HPV Testing

HPV DNA detection was performed as previously described.'5 In

brief, cell samples were divided into aliquots and tested sequentially

by several methods depending on their availability. A first aliquot

was tested with Virapap (VP; Digene Diagnostics Inc, Silver Spring,

MD), as well as by SH. All specimens from these groups for which

a second aliquot was available were also examined by PCR. Subjectswho tested HPV-negative by PCR in the cell pellet were further

analyzed using newly cut fragments of the frozen biopsy specimens.

For PCR, HPV Ll consensus primers that amplify a 450-bp fragment

were used. The amplification products were hybridized to oligomer

type-specific probes of HPV 6, 11, 16, 18, 31, 33, and 35 and with

an HPV generic probe. 18 Primers for a fragment of B-globin gene

served as an internal control to assess the sufficiency of each speci-

men for amplification. All specimens were prepared and analyzedin one laboratory (K.V. S.), following standard procedures. None of

the negative controls (one human kidney tissue fragment per 25

tumors analyzed) showed HPV DNA, which suggests that laboratory

contamination was not common. Quality-control (internal reproduc-

ibility) specimens (one masked, repeated tumor specimen per 25

tumors analyzed) showed perfect agreement, which suggests both

reliability and lack of contamination. All utensils were disposable

and were discarded immediately after each use (single use for eachspecimen).

HPV DNA prevalence was estimated by combining the results of

the PCR assay in exfoliated cells and biopsies as follows. Specimens

positive for HPV DNA included those that tested positive by PCR

(n = 172) and those that tested positive by SH (n = 32) if the

PCR assay was inadequate (negative for both B-globin and HPV

amplification by PCR, n = 15) or not performed (n = 139). Speci-mens that tested negative by SH and lacked a PCR result remained

in the category "inadequate or not done" because of the high speci-

ficity and low sensitivity of the SH assay to detect HPV DNA.Three estimates of the HPV viral load were made: first, by grading

the intensity of the VP signal into five categories (1 to 5); second,

by estimating the HPV copy number per cell in the specimens that

were tested by SH; and third, by combining the results of the VP,

SH, and PCR into three categories (negative by all three methods;positive by PCR and by either VP or SH; and positive by PCR only).

Seroreactivity to five peptides derived from HPV 16 open-reading

frames (ORF) E2 (E2-245), E7 (E7/1, E7/2 and E7/3), and L2 (L2)

was determined by an ELISA at the Department of ExperimentalVirology in Prague, Czech Republic.' 6 The peptide derived from E2

protein (aa 328-345) and designated as E2-245 by Dillner et al 9

was modified by addition of one cysteine residue at the C-terminus

and its dimer obtained by oxidation was used as antigen. Three E7

peptides corresponded to the N-terminal part of the HPV 16 E7

protein (E7/1 aa 1-20; E7/2 aa 11-30; E7/3 aa 21-40). The L2 peptide

was derived from capsid protein L2 (aa 197-216). Cutoff values for

seropositivity were determined as the mean absorbance value plus

3 SD in a group of sera obtained from children 3 to 11 years old.

These were considered to be HPV 16-negative. All determinations

of antibodies to peptides were made blindly as to the clinical status

of the patient. Results were given as positive/negative, as well as in

a three-grade intensity scale (weak, equal or slightly above cutoff

value; moderate, well above but less than double cutoff value; andstrong, above the double of the cutoff value).

Serologic assays were used as previously described" to assess

exposure to Gonococcus, Treponema pallidum, Chlamydia tracho-

matis, herpes simplex virus type II (HSV2), and hepatitis B virus(HBV).

Statistical Methods

The probability of survival was calculated using Kaplan-Meierestimates. Survival probability distributions were compared with the

log-rank test (Mantel-Cox). Crude and adjusted hazards ratios (HRa)

were calculated using Cox's regression models. Age and countrywere systematically added in the adjusted analyses.

As prognostic factors, the following were evaluated: clinical stage;

treatment modality; HPV DNA presence (as detected by any of the

three methods); HPV type; HPV viral load; seropositivity to peptides

derived from HPV 16 E2, E7, (1-3), and L2; lifetime exposure to

oral contraceptives (OC); smoking; seropositivity to any of the four

sexually transmitted diseases (STDs) tested; and history of genitalcondylomas. Other variables evaluated were those related to sexual

and reproductive behavior, education level, histology subtype, and

previous Papincolaou (Pap) smear history. All analyses producedcomparable results in Spain and Colombia; therefore, the combined

results are presented. Similarly, the prognostic factors for tumorrecurrence were the same as those that predicted survival. Therefore,only the survival analyses are presented in detail.

The proportional hazards assumption for each univariate Coxmodel was assessed by visual inspection of the plot of the log minuslog of the cumulative survival function versus time for each category

of each potential prognostic factor. All Cox models appeared to

appropriately fit the data. Statistical significance was established at

the 0.05 alpha value and accordingly 95% confidence intervals (CIs)around hazards ratios are presented. All P values are two-tailed.

Ethics Clearance

The study protocols were cleared by the IARC Ethical Committee,as well as by the ethics reviewers of the local institutions in thefield.

RESULTS

The four studies conducted in Spain and Colombia

included 399 cases of invasive cervical cancer and 525

611



VILADIU ET AL

cases of CIN III. After review by the panel of cytopathol-ogists, three cases originally diagnosed as invasive weredowngraded to CIN III and 72 cases originally diagnosedas CIN III were upgraded to invasive. These 75 reclassi-fied cases for which at least one pathologist diagnosedinvasive cervical cancer were included in the follow-upanalysis, but analyzed throughout as a separate categoryidentified as CIN III/stage I. The total number of eligiblewomen was thus 471, of which successful follow-up eval-uation was achieved in 410 (87.0%), including 61 of the75 reclassified cases. There was a significant difference inthe follow-up rate between Spain (96.8%) and Colombia(75.7%) (P < .001). The mean follow-up time of the 410cases was 45.0 months and the median, 55.0 months.

The cumulative survival proportions at 3 and 5 yearswere 72.2% and 66.9%, respectively. The median survivaltime was not achieved, since only 30% (122 of 410) ofthe patients had died at the end of the study. If the 61patients classified as CIN III/stage 1 and who completedfollow-up evaluation are excluded, the 3- and 5-year cu-mulative survival proportions were 68.9% and 63.1%,respectively .

Table 1 lists the distribution of cases by relevant vari-ables, crude estimates of the hazards ratio of death (HRc),and HRa estimates with the corresponding 95% CIs. Fac-tors significantly related to survival were clinical stage (asclassified by either TNM or FIGO systems) and treatmentmodality; therefore, all subsequent analyses were adjustedby these two variables. In addition, age (in three catego-ries) and country were also systematically added to theadjustment. Age was included because of a moderate as-sociation with survival in the crude analyses, and countrybecause of study design and the likelihood of unmeasuredrelevant differences between countries. Small-cell cancershad better survival than the reference group.

The univariate analyses showed associations betweenthe use of OCs and the record of previous Pap smearwith survival (HRc in Table 1). These associations weresignificantly reduced after adjusting for stage and treat-ment modality.

Treatment modality showed an association with sur-vival that was statistically significant after adjustment forstage (HRa, Table 1). In our series, there was a strongcorrelation between clinical stage and treatment modality.Stage I patients were largely treated with surgery and thesurvival time was long (ie, 85% alive at 7 years). Incontrast, stage IV patients were exclusively treated withradiotherapy, mostly as a palliative treatment, and theirsurvival was poor (ie, 44.4% at 1 year). Therefore, theindependence of treatment modality as a prognostic factorwas assessed in stages II and III in which the effect of

treatment choice could be better assessed (ie, treatmentsfor stages II and III were surgery 19%, radiotherapy 60%,and combined 30%). In the analysis restricted to stagesII and III (n = 209), there were no differences in survivalfor patients in relation to treatment modality. As com-pared with surgically treated patients, the HRa for patientstreated with radiotherapy was 1.5 (95% CI, 0.6 to 3.9),and for combinations of surgery and radiotherapy, theHRa was 1.3 (95% CI, 0.5 to 3.4).

No prognostic value was observed for any of the riskfactors for cervical cancer such as country (Spain v Co-lombia), number of sexual partners, age at first sexualintercourse, histologic subtype (except for small-cell car-cinomas), cell differentiation, level of education, age atfirst birth, parity, smoking, or history of genital warts.Similarly, there was no association between survival andserologic evidence of previous exposure to commonSTDs: Gonoccocus, HBV, Chlamydia trachomatis,HSV2, or Treponema pallidum.

HPV DNA positivity was identified in 204 of 256 spec-imens (79.7%) adequately tested. Fifty-two cases (20.3%)were considered HPV-negative and 154 specimens wereconsidered to be HPV-inadequate (n = 15) or not done(n = 139; see Methods for details). HPV type distributionwas as follows: HPV 16, 136 of 204 (66.7%); HPV 18,12 of 204 (5.9%); and HPV 31, 33, and 35, 24 of 204(11.8%). Twenty-five cases (12.3%) were HPV-positiveby uncharacterized types. There were no significant dif-ferences between the two countries with respect to HPVpositivity and type-specific distributions.

Table 2 lists the association of several HPV-relatedvariables with survival. There was no effect of HPV DNAdetection or any of the estimates of HPV viral load. HPV-positive cases seemed to have a slightly better survivalthan HPV 18-positive or than HPV-positive cases by thegeneric probe only, but the differences were not statisti-cally significant. The same results were obtained whenthe analyses were restricted to individuals who had anadequate PCR result (ie, after excluding the 32 specimensthat were SH-positive and inadequate for PCR testing).There was no association between HPV type or estimatesof HPV viral load and clinical stage (data not shown).

Table 3 lists the results of the survival analyses inrelation to HPV 16 E7 antibodies. Women who wereseropositive to HPV 16 E7/3 had a twofold increasedmortality rate as compared with women who were sero-negative. The analyses restricted to the cases that wereHPV 16-positive (n = 136) confirmed the result withslightly higher point estimates (the corresponding HRaestimates were as follows: E7/1, 1.7 [95% CI, 0.6 to 4.8];E7/2, 2.5 [95% CI, 1.1 to 5.6], E7/3, 2.5 [95% CI, 1.1 to

612



HPV AND CERVICAL CANCER SURVIVAL 613

Table 1. Risk of Death of Cervical Cancer in Relation to Sociodemographic and Clinical Characteristics

PatientsNo. of

Characteristic No. % Deaths HRc HRo 95% CI

Age, years39

40-5455

CountrySpainColombia

FIGO stageCIN III/I

I

III

IV-M

UnknownP for trend test

TreatmentSurgerySurgery and radiotherapyRadiotherapyOther

Histologic subtypeLarge-cell nonkeratLarge-cell keratSmall-cell

Squamous NOSIn situOther and unknown

SmokingNeverEver

Oral contraceptivesNeverEver

Papanicolau testst

97162151

245165

61101

116931326

13799

15519

1065618

157

3934

266144

282128

23.739.536.8

59.840.2

14.924.628.322.7

3.26.3

33.424.137.8

4.6

25.913.7

4.438.3

9.58.3

64.935.1

68.831.2

11.1

1.8'

1

1.0

1

3.17.9*

18.9*52.6*

8.2*< .001

1

3.2*5.4*

13.9*

1

0.80.50.6*0.08*0.2*

1

0.8

10.6*

1

0.70.9

0.4-1.30.5-1.6

10.9 0.5-1.4

1

2.84.9

11.116.7

5.3< .001

1

1.72.35.2

1

1.00.20.60.60.3

1

0.9

11.0

0.6-12.41.1-22.12.4-50.93.1-91.41.0-26.8

0.8-3.41.1-5.01.9-14.0

0.6-1.70.1-0.60.4-1.00.1-6.10.1-0.9

0.6-1.3

0.5-1.6

Never 251 61.2 93 1 1

Ever 159 38.8 29 0.5 * 0.7 0.4-1.2

NOTE. Bold figures indicate statistical significance.

Abbreviations: HRc, crude hazards ratio; HRa, hazard ratio adjusted for age, country, stage, and treatment; CI, confidence interval; FIGO, Fed6rationInternationalle de Gyncologie et d'Obstetrique; kerat, keratinizing; NOS, not otherwise specified; CIN, cervical introepithelial neoplasia.

*P< .05

tMore than 2 years before diagnosis.

6.1]). When the association was further investigated bystage, the prognostic value of HPV 16 E7/3 seropositivitywas restricted to stages I (HRa = 6.6 [95% CI, 1.2 to37.6]) and II (HRa = 5.9 [95% CI, 2.1 to 16.5]).

HPV 16 E7/2 was also related to survival in stage IIpatients (HRa = 3.8 [95% CI, 1.6 to -9.1]). Seropositivityto combinations of the three E7-derived peptides showeda clear dose-response relationship for stage II cases (testfor trend, P = .003). The 61 cases classified as CIN III/stage I showed HPV 16 E7/3 seropositivity rates closeto stage I cases (13% v 11%, respectively). Two deaths

occurred in this group, but no serologic results were avail-able for them.

The prognostic value of HPV 16 E7/2 and E7/3 antibodieswas investigated by level of seroreactivity. The associationswith survival were restricted to the cases that displayed aweak seroreaction. For HPV 16 E7/2, HRa for weak reactionwas 3.4 (95% CI, 1.7 to 6.7) and for HPV 16 E7/3, 6.7(95% CI, 2.8 to 16.2). When the analyses were stratified bystage, the strong prognostic value for women who showeda weak seroreaction was again restricted to stages I and II(HRa for E7/2 = 8.5 [95% CI, 2.9 to 25.4]; HRa for E7/3



614 VILADIU ET AL

Table 2. Risk of Death of Cervical Cancer in Relation to HPV DNA Detection and to Three Estimates of HPV Viral Load

PatientsNo. of

Variable No. % Deaths HRc HRa 95% CI

HPV DNA (PCR and SH)Negative 52 (12.7) 16 1 1Positive 204 (49.8) 55 0.9 0.9 0.5-1.5Inadequate or not done 154 (37.6) 51 1.2 1.2 0.7-2.2

HPV type (PCR and SH)Negative 52 (12.7) 16 1 1Generic 25 (6.1) 12 1.7 1.1 0.5-2.416 136 (33.2) 32 0.7 0.7 0.4-1.418 12 (2.9) 4 1.1 1.1 0.4-3.431, 33, 35 24 (5.9) 7 1.1 1.1 0.5-2.8Inadequate or not done' 161 (39.3) 51 1.0 1.2 0.7-2.2

Virapap signal intensityLightly positive (1) 31 (7.6) 4 1 1Positive (2-3) 60 (14.6) 16 2.5 2.4 0.8-7.1Strongly positive (4-5) 57 (13.9) 12 1.6 1.5 0.5-4.6Negative or inadequate 262 (63.9) 90

Viral load estimatePCR SH VP

+ - - 83 (20.2) 30 1.0 1.0+ and (+ or +) 88 (21.5) 21 0.6 0.9 0.5-1.7- - _ 44 (10.7) 15 1.0 1.2 0.6-2.2Other 195 (47.6) 56 - -

No. of HPV DNA copies per cell (SH)< 0.17 21 (5.1) 4 1 10.17-1.99 21 (5.1) 4 1.0 1.2 0.3-4.82.00-6.99 19 (4.6) 7 1.6 1.6 0.5-5.57.00-39.99 18 (4.4) 1 0.2 0.1 0.01-1.22 40 17 (4.1) 2 0.5 0.4 0.08-2.3Negative or not done 314 (76.6) 104

*Seven cases that were HPV(+) by SH low-stringency only are included in this category.

= 14.6 [95% CI, 4.9 to 43.4]), and not for stages III andabove (HRa for E7/2 = 1.5 [95% CI, 0.5 to 4.4]; HRa forE7/3 = 1.5 [95% CI, 0.2 to 12.2]). HPV 16 E2 and L2antibodies did not show any significant association withsurvival; however, the point estimates of the hazards ratewere all higher for women who showed a weak reaction ascompared with women who showed higher antibody levels(data not shown).

Although no association was found between HPV DNAstatus and any of the serologic markers of HPV, seroposi-tivity prevalences were found to increase with increasingHPV viral load levels, in particular for E7/1, E7/2, andE7/3 peptides. These associations were however of bor-derline statistical significance (data not shown).

Figures 1 and 2 show the Kaplan-Meier survival esti-mates at 7 years in the 410 cases by HPV DNA statusand HPV type, respectively. Figure 3 shows the survivalexperience in relation to the presence of E7/3 antibodiesrestricted to patients with stage I and II tumors.

DISCUSSION

In this series of 410 cases of squamous cell cervicalcancer, HPV DNA detection and HPV type or HPV viralload estimates did not show any significant associationwith survival. A moderate association was observed withseroreactivity to HPV 16 E7/2 and E7/3 peptides. How-ever, when the analysis was restricted to early stages,seroreactivity to HPV 16 E7/3 and E7/2 was stronglyrelated to survival.

This is a large series of incident cases of squamouscell cervical cancer with a follow-up rate of 87% for amedian interval of 4.6 years. The differences in the fol-low-up rates obtained in Spain (96.8%) and Colombia(75.5%) were largely attributed to higher hospital mobil-ity of the population in Cali. The 53 cases lost to follow-up evaluation in Cali were compared with the 165 thatwere successfully traced and no differences were ob-served concerning age, level of education, or any otherepidemiologic variable described in the questionnaire. We



HPV AND CERVICAL CANCER SURVIVAL 615

Table 3. Prognostic Value of Seropositivity to HPV 16 E7-Derived Peptides by Stage (FIGO)

Variable Stage

III III & IV Allt

No. of No. of No. of No. of No. of No. of No. of No. ofAg Ab Cases Deaths HR 95% Cl' Cases Deaths HR 95%CI C Cases Deaths HR 95% CIt Cases Deaths HR 95% CIt

E7/1 Negative 60 8 1.0 62 18 1.0 61 37 1.0 239 67 1.0Positive 11 2 1.1 0.2-5.4 15 4 1.5 0.5-4.4 15 7 0.6 0.2-1.4 49 16 0.9 0.5-1.6

E7/2 Negative 53 6 1.0 56 12 1.0 50 29 1.0 210 52 1.0Positive 18 4 1.7 0.5-6.4 21 10 3.8 1.6-9.1 26 15 0.8 0.4-1.6 78 31 1.4 0.9-2.2

E7/3 Negative 63 7 1.0 60 15 1.0 62 35 1.0 240 62 1.0Positive 8 3 6.6 1.2-37.6 17 7 5.9 2.1-16.5 14 9 1.1 0.5-2.3 48 21 2.01.2-3.3

E7/1-3 summary

E7/1 E7/2 E7/3

49 6 1.0 48 11 1.0 44 26 1.0 185 46 1.01 or 2 positive 17 2 0.7 0.1-3.8 20 7 2.17 0.8-5.8 26 14 0.8 0.4-1.6 79 26 1.2 0.8-2.0+ + + 5 2 4.6 0.6-32.8 9 4 5.2 1.5-18.1 6 4 1.0 0.3-3.1 24 11 1.9 1.0-3.8P for trend 0.8 0.0035 .75 .08

NOTE. Bold figures indicate statistically significant hazards ratios.Abbreviations: Ag, antigen; Ab, antibody.*Hazards ratio adjusted for age group, country, and treatment.tHazards ratio adjusted for age group, country, treatment, and stage.Includes all cases with serology: stage I-IV (n = 224), CIN III/stage I (n

therefore assumed that their absence from the main analy-sis did not introduce a significant selection of the series.

All cases were recruited in the context of population-based case-control studies. At recruitment, the informa-tion was obtained through a comprehensive questionnaireadministered by specially trained interviewers. HPVDNA status was assessed using a well-characterized PCRsystem and SH in a single research laboratory.

= 46), and unknown stage In = 18).

In our study, we could verify that cervical cancer wasthe cause of death in 97 of 101 patients (96.0%) who haddied in the interval. Therefore, the relative survival rate(survival estimates corrected for other causes of death)is close to the observed. The 5-year survival proportionobserved in our study was 63.1% (after excluding CINIII/stage I cases). This is slightly better than the survivalobserved in population-based series and closer to the fig-

1.04

Positive

~-,.__.~____ NegativeUnknown

z

r

a,

.8~

.7

.61

.5

.4

.3

.2

.1

.012 24 36 48 60 72 84 96

Months

16

ID 12 24 36 48 60 72 84 96

Months

Fig 1. Survival distribution in 410 cases of invasive squamous cellcervical carcinoma according to PCR-detected HPV DNA status. Log-rank P = .38 (see Table 2 for multivariate assessment of survivaldistributions).

Fig 2. Survival distribution in 410 cases of invasive squamous cellcervical carcinoma according to PCR-detected HPV DNA type-specificstatus. Log-rank P = .22 (see Table 2 for multivariate assessment ofsurvival distributions).

z

coAWe

1.0

.9

.8

.7'

.61

.5,

.4

.3'

.2

.1

00

-

n

r



VILADIU ET AL

zz

Months

Fig 3. Survival distribution in 148 cases of invasive squamouscell cervical carcinoma, stage I-I, according to anti-E7/3 (aa 21-40)serologic response. Log-rank P = .001 (see Table 3 for multivariateassessment of survival distributions).

ures reported by hospital-based series. In a population-based cancer registry in Girona, Spain (one of the areas

participating in the study), the 5-year relative survivalproportion was 55%.22 The 5-year relative survival pro-

portions reported by the summary of European registriesfrom 12 countries (based on 24,000 expected cases peryear) was 59%.23 In Sweden, a cohort of approximately

6,000 cervical cancer cases treated at a single large hospi-tal reported a 5-year relative survival proportion of 60.8%

for patients diagnosed from 1980 to 1990.24 The age and

stage distribution of the cases in our series closelymatches the distribution reported by population-basedregistries. The increased survival estimates reported heremay partially reflect the inclusion criteria (all cases were

squamous cell carcinomas, and the upper age limit was75 years) and do not suggest that the series is biasedin the sense of preferentially including cases with more

favorable prognostic.The HPV DNA prevalence of approximately 80% in

our study was obtained by PCR and in 32 cases by SH.There was no difference in the survival experience of

patients by HPV DNA status. Cases in the category ofHPV inadequate or not done had the same survival asthat of the rest of the series. Likewise, restriction of theanalyses to cases that had an adequate PCR showed nodifferences in survival. Considering the size of the studyand the follow-up duration, our results provide strongevidence that detection of HPV DNA is not a prognosticfactor for recurrence or survival. This conclusion is inagreement with the most recent evidence, which indicated

that HPV-negative cervical cancer cases may be more ofa detection artifact than a discrete clinical entity. 5' "2 5

Similarly, we could not find any association betweenestimates of the viral load and survival. Given the resultslisted in Table 2 and the systematic univariate and multi-

variate assessment of many other variables related to cer-vical cancer and/or survival (data not shown), it is un-

likely that a major effect due to negative confounding

was missed inadvertently.Table 4 lists the results of other studies that have ad-

dressed the prognostic value of HPV DNA detection.Moderate increased risk of death among HPV DNA-nega-tive cases were found in three studies,6 8 one of which

used the occurrence of relapse/metastases as a surrogatemeasurement of survival.7 In four studies, there were no

differences in survival by HPV status.9 -' 11 '3Misclassification (largely underdetection) of HPV sta-

tus and/or clinical stage is a possible explanation for thecontroversy. Although the HPV DNA prevalence in cervi-

cal cancers obtained in our study is high, it still falls short

of prevalences reported in more recent studies 5 l' or ofthose using other well-established PCR systems." 25

Three of the studies that reported a significant prognosticvalue of HPV detection (Table 4) also reported an HPVprevalence of 80% or greater. If HPV underdetection is

the explanation of the conflicting results, some associa-tion must exist between poor survival and HPV detect-ability. Technical problems and/or quality of the specimencould be proposed as possible causes and more detailed

analyses of these data sets would be useful. Misclassifi-cation of clinical stage is discussed later.

Previous studies have suggested that HPV 18-positivecases were more often found in advanced stages and dis-played a more rapid progression rate than HPV 16cases.26 27 This hypothesis has not been confirmed by onelarge study that did not find any differences in clinicalstage at diagnosis by HPV type.5 We found a lower sur-vival proportion with HPV 18 cases as compared withHPV 16 cases, but this observation is of limited value as

it is based on only 12 cases of HPV 18. Hetcht et a128

have recently described an HPV 18 subtype of decreasedoncogenic potential. Likewise, variants in the E6 proteinof some of the patients included in our study showed

significant differences in their immortalizing activity.2 9

Further characterization of the HPV types and subtypeswith respect to their pathogenic potential and geographicdistribution may clarify these apparent discrepancies.

The relationship between seropositivity to HPV 16 E7/2 and E7/3 and poor survival in patients with stage I orII may reflect advanced disease and increased exposureof the immune system to the viral proteins. It has been

616

is




Table 4. Studies Evaluating HPV DNA as a Prognostic Factor in Cases of Invasive Cervical Cancer

No. of -HVP+First Author, Year Cases Stages HPV Test and Types No. % Survival Comparisons by HPV Status

King, 19899 85 I-111b In situ hybridization in 38 45 HPV 16 MST: 72.9 months

formalin-fixed specimensHPV 16 and 18

HPV 18 MST: 80.1 monthsHPVI-) MST: 83.4 months (NS)

Walker, 1989'3 100 I-IV SH 62 63 HPV(-) ORc = 1.0HPV 6, 11, 16, 18, and 31 HPV(+) ORe = 2.6 (0.7-10.1)'

Riou, 19907 106 Ib-llb PCR and SH 89 84 HPV(+) HRa = 1.0HPV 6, 11, 16, 18, 31, 33, HPV(-) HRa relapse = 2.6; P

35, 39, and 42 < .05HRa metastases = 4.5;

P < .01HPV types NS

Higgins, 19916 212 I-IV RNA; in situ hybridization in 170 80.2 HPV(-) HRa = 1.0

formalin-fixed specimensHPV 16, 18, 31 and 33 HPV(+) HRa = 0.34 (0.19-

0.64)HPV types NS

Jarrell, 1992 t ° 127 lb, Ila PCR, dot blot 67 53 HPV 16(+)HPV 16 and 18 HPV 16(-) P = .09

DeBritton, 19938 196 I-IV PCR, SH, slot-blot 144 81 HPV(+) HRa = 1.0HPV 16 and 33 HPV(-) HRa = 1.9 (1.1-3.3)

Van Bommel, 64 Ib-lla PCR 64 100 Presence of metastatic nodes199311

HPV 6, 11, 16, 18, 31,33 Nodes(+) HPV prevalence: 100%and X

Nodes(-) HPV prevalence: 100%HPV types (NS)

NOTE. Bold figures indicate statistically significant risk estimates.Abbreviations: MST, mean survival time; NS, not statistically significant.'Odds ratios for death (calculated from published data).

shown that seropositivity to E7 is related to neoplasticinvasion. In several studies, HPV DNA 16-positive CINIII cases rarely developed antibodies to linear epitopes ofHPV 16, whereas - 50% of HPV 16 invasive cancersshowed seroreactivity.3 ""3 2 De Sanjos6 et al33 showed thatHPV 16 E7 seropositivity increases with the presence ofneoplastic invasion (ie, from CIN III to stage I or above).Seroreactivity to HPV 16 E7/1-3 in the present series wasas follows: 26.1% in CIN III/stage I cases, 31.0% in stageI cases, 37.7% in stage II cases, and 42.1% in stage IIIand IV cases (P for trend = .05).

Three previous reports that related HPV 16 E7 antibod-ies to stage and prognosis described results consistentwith our findings. Onda et a134 observed that seroreactivityto HPV 16 anti-E4 and -E7 was more often found amongHPV 16 invasive cancer cases if lymph node or distantmetastases was found. Baay et al'4 reported that seroreac-tivity to HPV 16 E7 protein (aa 6-35) was correlated totumor size, lymph node involvement, and a worse progno-sis. These investigators further observed in a small num-

ber of cases that antibody titers decreased following treat-ment and increased with tumor recurrence.35

In our study, 148 patients were clinically classified asstage I and II, of whom 16 died during the interval follow-ing regional recurrence (n = 5) or distant metastasis (n= 4). In the entire group, those who developed HPV 16E7/2-3 antibodies had a sixfold to sevenfold increasedrisk of death as compared with those who did not developantibodies. Under the hypothesis that seroreactivity devel-ops after lymph node (or other) metastasis, our resultssuggest that in perhaps 16 cases (10.8%) there had beenclinical understaging that could have been identified withthe E7/3 serologic test. Understaging of early cases ofcervical cancer has been extensively recognized in theliterature. In a review of more than 3,000 cases of cervicalcancer, the presence of paraaortic lymph node metastasisin stages I and II has been documented in a range from4.0% (stage Ib) to 20.0% (stage IIb).3 6 This is of clinicalrelevance, since the standard radiation fields for stages Iand II do not cover the paraaortic nodes. If these findings

617



VILADIU ET AL

are confirmed, antibodies to HPV proteins may be ofinterest in predicting the development of recurrence ormetastasis in patients who would otherwise be consideredto have a good prognosis.

Finally, our analyses of the strength of the immuneresponse suggest that in the presence of invasive disease,a weak immunologic reaction to HPV 16 E7/2-3 is amarker of poor prognosis, whereas cases with high orvery high antibody titers do not show any differences insurvival as compared with seronegative cases. We donot have a satisfactory explanation for this association.However, it could be speculated that high antibody titersreflect an effective immunologic protection against pro-gression of the neoplasia. Because we cannot entirelyrule out statistical limitations due to small numbers or an

artifact due to the overlapping structure of the peptidesused, these findings should be interpreted cautiously andneed to be confirmed in other studies.

In summary, clinical stage is the central independentprognostic factor for survival in invasive squamous cellcarcinoma. Seropositivity to HPV 16 E7/3 predicts a six-fold to sevenfold increase risk of death in early stage (Iand II) cases. The presence/absence of HPV DNA, HPVtype, estimates of HPV viral load, and none of the estab-lished risk factors for cervical cancer were found to bepredictors of tumor recurrence or survival.

ACKNOWLEDGMENT

We recognize with gratitude the efforts of the many gynecologists,pathologists, and oncologists who provided the clinical follow-upevaluations of the patients.

REFERENCES

1. Bosch FX, Mufioz N, Shah K, et al: Second InternationalWorkshop on the Epidemiology of Cervical Cancer and HumanPapillomavirus. Int J Cancer 52:171-173, 1992

2. Mufioz N, Bosch FX, de Sanjos6 S, et al: The causal linkbetween human papillomavirus and invasive cervical cancer: A pop-ulation-based case-control study in Colombia and Spain. Int J Cancer52:743-749, 1992

3. Bosch FX, Mufioz N, de Sanjos6 S, et al: Human papillomavi-rus and cervical intraepithelial neoplasia grade III/carcinoma in situ:A case-control study in Spain and Colombia. Cancer EpidemiolBiomarkers Prev 2:415-422, 1993

4. Bosch FX, Mufioz N: Human papillomavirus and cervical neo-plasia: A critical review of the epidemiological evidence, in MufozN, Bosch FX, Jensen OM (eds): Human Papillomavirus and CervicalCancer. Lyon, France, IARC Scientific, Publications No. 94, 1989,pp 135-151

5. Bosch FX, Manos MM, Mufioz N, et al: Prevalence of humanpapillomavirus in cervical cancer: A worldwide perspective. J NatlCancer Inst 87:796-802, 1995

6. Higgins GD, Davy M, Roder D, et al: Increased age and mortal-ity associated with cervical carcinomas negative for human papillo-mavirus RNA. Lancet 338:910-913, 1991

7. Riou G, Favre M, Jeannel D, et al: Association between poorprognosis in early-stage invasive cervical carcinomas and non-detec-tion of HPV-DNA. Lancet 335:1171-1174, 1990

8. DeBritton RC, Hildesheim A, De Lao SL, et al: Human papillo-maviruses and other influences on survival from cervical cancer inPanama. Obstet Gynecol 81:19-24, 1993

9. King LA, Tase T, Twiggs LB, et al: Prognostic significanceof the presence of human papillomavirus DNA in patients withinvasive carcinoma of the cervix. Cancer 63:897-900, 1989

10. Jarrell MA, Heitz N, Howard P, et al: Squamous cell carci-noma of the cervix: HPV 16 and DNA ploidy as predictors of sur-vival. Gynecol Oncol 46:361-366, 1992

11. Van Bommel PFJ, Van den Brule AJC, Helmerhorst TJM, etal: HPV DNA presence and HPV genotypes as prognostic factorsin low-stage squamous cell cervical cancer. Gynecol Oncol 48:333-337, 1993

12. Girardi F, Fuchs P, Haas J: Prognostic importance of humanpapillomavirus type 16 DNA in cervical cancer. Cancer 69:2502-2504, 1992

13. Walker J, Bloss JD, Liao S-Y, et al: Human papillomavirusgenotype as a prognostic indicator in carcinoma of the uterine cervix.Obstet Gynecol 74:781-785, 1989

14. Baay MFD, Duk JM, Burger MPM, et al: Antibodies to hu-man papillomavirus type 16 E7 related to clinicopathological datain patients with cervical carcinoma. J Clin Pathol 48:410-414, 1995

15. Guerrero E, Daniel RW, Bosch FX, et al: Comparison ofVirapap, Southern hybridization and polymerase chain reactionmethods for human papillomavirus identification in an epidemiologi-cal investigation of cervical cancer. J Clin Microbiol 30:2951-2959,1992

16. Hamsikova E, Novak J, Hofmannova V, et al: Presence ofantibody to seven HPV16 derived peptides in cervical cancer patientsand healthy control subjects. J Infect Dis 170:1424-1431, 1994

17. Kato I, Santamaria M, Alonso de Ruiz P, et al: Inter-observervariation in cytological and histological diagnoses for cervical neo-plasia and associated factors. J Clin Epidemiol 48:1167-1174, 1995

18. Bauer HM, Ting Y, Greer CE, et al: Genital human papillo-mavirus infection in female university students as determined by aPCR-based method. J Am Med Assoc 265:472-477, 1991

19. Dillner J, Dillner L, Robb J, et al: A synthetic peptide definesa serologic IgA response to a human papillomavirus-encoded nuclearantigen expressed in virus-carrying cervical neoplasia. Proc NatlAcad Sci USA 86:3838-3841, 1989

20. de Sanjos6 S, Mufioz N, Bosch FX, et al: Sexually transmittedagents and cervical neoplasia in Colombia and Spain. Int J Cancer56:358-363, 1994

21. Mufioz N, Kato I, Bosch FX, et al: Cervical cancer and herpessimplex virus type 2: Case-control studies with special reference toimmunoglobin G subclasses. Int J Cancer 60:438-442, 1995

22. Viladiu P, Izquierdo A, Beltran M, et al: Epidemiologia delcancer ginecol6gico y de mama en el area sanitaria de Girona. Gir-ona, Spain, Registre del Cncer de Girona, Asociaci6n EspafiolaContra El Cancer, 1995, pp 1-251 (in Spanish)

23. Berrino F, Sant M, Verdecchia A, et al (Eds): Survival ofCancer Patients in Europe. The EUROCARE Study. Lyon, France,IARC Scientific, Publications No. 132, 1995

24. Spar6n P, Gustafsson L, Friberg L-G, et al: Improved controlof invasive cervical cancer in Sweden over six decades by earlierclinical detection and better treatment. J Clin Oncol 13:715-725, 1995

25. Meijer CJLM, van der Brule AJC, Snijders PJF, et al: Detection

618




of human papillomavirus in cervical scrapes by the polymerase chainreaction in relation to cytology: possible implications for cervical cancerscreening, in Mufioz N, Bosch FX, Shah KV, et al (eds): The Epidemiol-ogy of Cervical Cancer and Human Papillomavirus. Lyon, France,IARC Scientific, Publications No. 119, 1992, pp 271-281

26. Lorincz AT, Temple GF, Kurman RJ, et al: Oncogenic associ-ation of specific human papillomavirus types with cervical neoplasia.J Natl Cancer Inst 79:671-676, 1987

27. Arends MJ, Donaldson YK, Duvall E, et al: Human papillo-mavirus type 18 associates with more advanced cervical neoplasiathan human papillomavirus type 16. Hum Pathol 24:432-437, 1993

28. Hetcht JL, Kadish AS, Jiang G, et al: Genetic characterizationof the human papillomavirus (HPV) 18 E2 gene in clinical specimenssuggests the presence of a subtype with decreased oncogenic poten-tial. Int J Cancer 60:369-376, 1995

29. Conrad-St6ppler M, Ching K, St6ppler H, et al: Natural vari-ants of the HPV 16 E6 protein differ in immortalizing activity andability to degrade p53. Fourteenth International Papillomavirus Con-ference. Quebec City, Canada, Universit6 Laval Facult6 de M6de-cine, 1995 (abstr 96)

30. MUller M, Viscidi RP, Sun Y, et al: Antibodies to HPV-16 E6and E7 proteins as markers for HPV-16-associated invasive cervicalcancer. Virology 187:508-514, 1992

31. Jochmus-Kudielka I, Schneider A, Braun R, et al: Antibodiesagainst the human papillomavirus type 16 early proteins in humansera: Correlation of anti-E7 reactivity with cervical cancer. J NatlCancer Inst 81:1698-1704, 1989

32. Mann VM, de Lao SL, Brenes M, et al: Occurrence of IgAand IgG antibodies to select peptides representing human papillo-mavirus type 16 among cervical cancer cases and controls. CancerRes 50:7815-7819, 1990

33. de Sanjos6 S, HamSikova E, Mufioz N, et al: Serologicalresponse to HPV 16 in CIN-III and cervical-cancer patients. Case-control studies in Spain and Colombia. Int J Cancer 66:70-74, 1996

34. Onda T, Kanda T, Zanma S, et al: Association of the antibod-ies against human papillomavirus 16 E4 and E7 proteins with cervi-cal cancer positive for human papillomavirus DNA. Int J Cancer54:624-628, 1993

35. Baay MFD, Duk JM, Burger MPM, et al: Follow-up of anti-body responses to human papillomavirus type 16 E7 in patientstreated for cervical carcinoma. J Med Virol 45:342-347, 1995

36. Van Nagell JR, Higgins RV: Clinical invasive carcinoma ofthe cervix: Clinical features, diagnosis, staging and pretreatmentevaluation, in Coppleson M, Monaghan JM, Morrow CP, et al (eds):Gynecologic Oncology, vol 1 (ed 2). New York, NY, ChurchillLivingstone, 1992, pp 663-671

619



Human papillomavirus DNA and antibodies to human papillomaviruses 16 E2, L2, and E7 peptides as...

Documents

Transcript of Human papillomavirus DNA and antibodies to human papillomaviruses 16 E2, L2, and E7 peptides as...