Genital Human Papillomavirus Infection

in Female University Students as

Determined by a PCR-Based Method
Heidi M. Bauer; Yi Ting, MS; Catherine E. Greer; Janet C. Chambers, MD; Cathy J. Tashiro, MPH; Joseph Chimera, PhD;
Arthur Reingold, MD; M. Michele Manos, PhD

The presence of genital human papillomavirus (HPV) was determined at cervical ficity. The accurate determination of
and vulvar sites using two methods, the Food and Drug Administration\p=m-\ap- HPV prevalence in different popula¬
proved ViraPap test and polymerase chain reaction (PCR) DNA amplification tions is key to studying the natural his¬
technology, in 467 women presenting to a university health service for a routine tory of HPV and its role in cervical
disease.
annual gynecologic examination. The PCR system afforded the sensitive detec-
tion of a broad spectrum of genital HPV types. Using PCR, we found that 46% of Many methods have been used to de¬
the study population was infected with HPV; the ViraPap test showed a preva-
tect HPV, including clinical observa¬
tion, cytology, electron microscopy, im-
lence of 11% infected. PCR analyses demonstrated that 69% of the HPV\x=req-\ munocytochemistry, and DNA hybridi¬
positive women were infected at both genital sites. Subsequent HPV-type zation.2 Although Southern blot hybrid¬
determination showed that 33% of the study population had HPV types 6, 11, 16, ization is currently considered the most
18,31,33,35,39,45,51,52, or other previously isolated types, and 13% had yet accurate, it requires large amounts of
unidentified types. Almost all (92%) of the women diagnosed by Papanicolaou DNA and is extremely laborious. Poly-
smear with condylomatous atypia or dysplasia (n 12) were HPV positive. The
= merase chain reaction (PCR) DNA am¬

PCR method proved to be an informative and rapid way to detect HPV in large plification,910 the most sensitive method
numbers of clinical samples. Our results demonstrate that genital HPV infection for nucleic acid detection currently
is common among sexually active young women. available, has yet to be widely used in
HPV studies, although its use in detect¬
(JAMA. 1991;265:472-477)
ing viral DNA in clinical specimens has
been demonstrated.11" The only com¬
mercially available Food and Drug Ad¬
RECOGNIZED as a sexually transmit¬ of infection in the progression to malig¬ ministration-approved genital HPV
ted agent, genital human papilloma¬ nancy has yet to be determined. Of the tests, ViraPap and ViraType, are re¬
virus (HPV) is also believed to be a con¬ more than 60 characterized HPV types, producible, relatively easy to perform,
tributing factor in cervical, vaginal, and about one third infect mucosal tissue. and have a sensitivity comparable to
vulvar intraepithelial neoplasia and car¬ Human papillomavirus types 6 and 11 Southern blot.15 This HPV DNA detec¬
cinoma.12 Human papillomavirus DNA are commonly associated with benign tion method is based on dot blot hybrid¬
has been detected in the majority of anogenital warts (condyloma acumina- ization and it detects seven HPV types.
primary and metastatic cervical carci¬ tum), while types 16 and 18 are found in Many HPV DNA detection systems
nomas and cervical intraepithelial neo- a majority of carcinomas. Other types are limited to the HPV types for which
plasias (CINs), although the exact role (eg, HPV types 31, 33, 35, 39, 45, 51, they include viral genomic probes or
and 52) also are reported to be associat¬ type-specific oligonucleotide probes,
ed with CIN or carcinoma.1 Although thus excluding detection of some known
From the Department of Infectious Diseases, Cetus and all unidentified HPV types. Recent¬
Corp, Emeryville, Calif (Mss Bauer, Ting, and Greer and the epidemiology of cervical cancer has
Dr Manos); the University Health Service (Dr Chambers
and Ms Tashiro) and the Program of Epidemiology (Dr
been studied extensively, the preva¬ ly we described a highly sensitive PCR-
lence of subclinical HPV infection is dif¬ based method that uses degenerate con¬
Reingold), University of California, Berkeley; and the
Department of Molecular Biology, Roche Biomed- ficult to determine from recent studies3"8 sensus primers to amplify many un¬
ical Laboratories, Research Triangle Park, NC (Dr because of the different detection sys¬ identified HPV types in addition to the
Chimera). tems employed and the variation in known types.12ie" These consensus
Reprint requests and correspondence to Cetus
Corp, 1400 53rd St, Emeryville, CA94608 (Dr Manos). their sensitivities and virus type-speci- primers target the highly conserved

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015

 late region 1 (LI), which encodes a viral tamination.22 Two hundred microliters temperature cycling (30 seconds at
capsid protein. This method was shown of each sample was removed for PCR 95°C, 30 seconds at 55°C, and 1 minute
to have a comparable type-spectrum analysis with a disposable pipette, pre¬ at 72°C). In order to assess possible con¬
but greater sensitivity than low-strin¬ cipitated at -20°C overnight with 2- tamination during PCR preparation,
gency Southern blot analysis.18 Subse¬ mol/L ammonium acetate and 70% etha¬ tubes containing the K562-negative
quently, other groups also have report¬ nol, dried, and resuspended in 20 |xL of control DNA or no added DNA were
ed the use of consensus primers for TE (10-mmol/L TRIS hydrochloride included in every set of reactions. Five
HPV DNA amplification.I9'20 [pH 7.5] and 1-mmol/L ethylenediamen- microliters of each completed reaction
In this study, we used an improved etetraacetic acid [EDTA]). Cervical and was electrophoresed on a 7% polyacryl-
PCR-based detection and typing sys¬ vulvar specimens were processed and amide gel, stained with ethidium bro¬
tem to determine the prevalence of cer¬ analyzed separately. ViraPap collection mide, and visualized with UV light. The
vical and vulvar HPV infection in a uni¬ tubes containing only known amounts of sufficiencies of internal control (ß-glo¬
versity population. We also compared purified human genomic DNA from the bin) amplifications were assessed from
the PCR method with ViraPap and K562 cell line (ATCC CCL243), which photographs of the gels.
ViraType. This work is part of an ongo¬ does not contain HPV DNA, were pro¬ To determine the analytical sensitiv¬
ing study of the epidemiology and asso¬ cessed with each batch of 20 samples to ity of the PCR-based technique, we ana¬
ciated risks of HPV infection in this pop¬ monitor possible contamination in sam¬ lyzed ViraPap tubes with various con¬
ulation of women. ple processing. These negative controls centrations of HeLa cells (ATCC
were interspersed and analyzed along CCL2), which contain 10 to 50 copies of
METHODS
with the clinical samples. HPV 18 per cell; or HPV 16 purified
Study Population Two hundred fifty microliters of sam¬ plasmid DNA in a background of 2 x 104
A cross-sectional survey was under¬ ple remaining in the collection tube (25% K562 cells. These tubes were pro¬
taken at the University Health Service of the original specimen) was analyzed cessed, amplified, and hybridized using
at the University of California, Berke¬ for the presence of HPV DNA using the the protocol presented herein.
ley. During the study interval (August ViraPap detection system (Life Tech¬ To determine the presence of HPV,
through October 1989), women who nologies Inc, Gaithersburg, Md), which we used a labeled generic HPV probe
called to make an appointment for a rou¬ uses a combination of radiolabeled RNA mix to hybridize dot blots containing
tine gynecologic examination were in¬ probes for HPV types 6, 11, 16, 18, 31, sample amplifications. The generic
formed about the study and invited to 33, and 35. If the sample demonstrated HPV probes were synthesized from the
participate. Participants were asked to an intensity equal to or above the low LI PCR fragments of HPV 16, HPV 18,
refrain from sexual intercourse and the positive control, it was analyzed by Vir- and two highly divergent yet unidenti¬
use of vaginal medications for 72 hours aType (Life Technologies Inc, Gaithers¬ fied HPV sequences (PAP88 and
prior to their examination. At the time burg, Md) to distinguish the HPV type PAP238B), previously isolated in the
of their appointments, women gave category (6/11, 16/18, and 31/33/35). Manos laboratory from clinical speci¬
written consent and completed a self- Manufacturer's instructions were fol¬ mens. (It was later determined that
administered questionnaire. The study lowed explicitly. All determinations PAP238B is virtually identical to HPV
was approved by the Committee for the were made without knowledge of other 31.) These four fragments used togeth¬
Protection of Human Subjects, Univer¬ data. er were found to detect a broad spec¬
sity of California, Berkeley. DNA Amplification
trum of HPV types. To synthesize ra¬
dioactive probes, amplifications were
Clinical Examination The PCR amplifications that target a performed separately with nested type-
Registered nurse practitioners col¬ portion of the HPV LI region (approxi¬ specific primers (Table 1) to each of the
lected both a vulvar and a cervical swab mately 450 bp [base pair]) were per¬ MY09/MYll-generated LI fragments
in separate ViraPap collection tubes formed on each sample preparation as using the same reaction and thermocy-
(Life Technologies Inc, Gaithersburg, previously described,16 with two modifi¬ cling profile previously described but
Md) for HPV analysis. Before the spec¬ cations: the addition of a simultaneous with 50-(j.mol/L each deoxynucleotide
ulum was inserted for the pelvic exami¬ human ß-globin gene amplification and triphosphate, 62.5 pmol (50 -xCi) of each
nation, a vulvar sample from the poste¬ an increase in cycle number to 35. Suc¬ a-phosphorus 32-labeled deoxynucleo¬
rior fourchette and inner aspects of the cessful amplification of the ß-globin tide triphosphate, and 20 pmol of each
labia minora was taken with a swab fragment (268 bp) indicated that the primer in a final volume of 50 (xL. The
moistened with saline. The cervical sample was adequate for HPV analysis mineral oil overlay was omitted and 25
swab for HPV testing was taken from and that no inhibitors were present. temperature cycles were run. To assess
the endocervical canal and the transfor¬ All reagent lots were kept constant the efficiency of the amplification, the
mation zone. A cervical swab for Papa- throughout the study. Each 100-|aL re¬ PCR products from identical concur¬
nicolaou smear analysis was collected action contained 200-|xmol/L each deox- rent unlabeled reactions were visual¬
last. Papanicolaou smears were evalu¬ ynucleotide triphosphate, 10-mmol/L ized using electrophoresis and ethidium
ated by pathologists at an independent TRIS hydrochloride (pH 8.5), 50- bromide staining. The resultant radio-
laboratory and abnormal results were mmol/L potassium chloride, and actively labeled fragments were puri¬
categorized according to the Bethesda 4.0-mmol/L magnesium chloride; 50 fied using a fractionating column (G50
System.21 pmol of each HPV LI consensus primer Sephadex). The four fragments were
(MY09 and MY11, Table 1); 5 pmol of denatured at 95°C for 10 minutes in the
Sample Preparation and each ß-globin primer (PC04 and GH20, presence of sheared salmon sperm
ViraPap Analysis Table 1), and 2.5 U of Taq polymerase DNA, then rapidly cooled in a dry
ViraPap collection tubes containing (Amplitaq, Perkin-Elmer Cetus, Nor- ice-alcohol bath immediately before ad¬
the swab in 1 mL of buffer were pro¬ walk, Conn). Aliquots of the clinical dition to the hybridization solution.
cessed according to the manufacturer's samples (2 (xL and 5 |xL, 2% and 5%, Membranes were hybridized with a
instructions. Samples were handled respectively, of the original swab sam¬ mixture of 1.0 to 1.5 xlO5 cpm/mL of
with extreme care to avoid cross-con- ple) were added immediately before each probe.

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015

 Table 1.—Oligonucleotide Probes and Primers Used in the PCR Method knowledge of Papanicolaou smear,
Name Sequence* Target_Purpose!_ ViraPap, or other data. All of the sam¬
MY09 CGTCCMARRGGAWACTGATC HPVL1_PCR consensus primer ( )"
ples that yielded HPV bands on ethi-
MY11 GCMCAGGGWCATAAYAATGG HPV L1
dium-stained gels also hybridized with
PCR primer ( + )12
-

consensus
MY12 6
the generic probe. Approximately half
CATCCGTAACTACATCTTCCA Type Type-specific probe" of the samples that hybridized with the
MY13 TCTGTGTCTAAATCTGCTACA Type 11 Type-specific probe'*
MY14 CATACACCTCCAGCACCTAA Type 16 Type-specific probe'2 generic probe on dot blots did not yield
WD74
HPV amplification products that were
GGATGCTGCACCGGCTGA Type 18 Type-specific probe'2 visible on gels.
WD126 CCAAAAGCCCAAGGAAGATC Type 31 Type-specific probe Two microliters of each generic-
WD128 TTGCAAACAGTGATACTACATT Type 31 Type-specific probe
M Y16 CACACAAGTAACTAGTG ACAG Type 33 Type-specific probe'2 probe-positive amplification was ap¬
MY59 AAAAACAGTACCTCCAAAGGA Type 33 Type-specific probe plied to replicate membranes and hy¬
bridized individually with probes for
MY115 CTGCTGTGTCTTCTAGTGACAG Type 35 Type-specific probe HPV types 6/11, 16, 18, 31, 33, 35, 39,
MY117 ATCATCTTTAGGTTTTGGTGC Type 35 Type-specific probe
MY89 TAGAGTCTTCCATACCTTCTAC Type 39_Type-specific probe_
45, 51, 52, W13A, PAP88, PAP155, and
PAP251 (Table 1). Probes for these "W"
MY90 CTGTAGCTCCTCCACCATCT Type 39 Type-specific probe and "PAP" specimens were designed
MY70 TAGTGGACACTACCCGCAG Type 45 Type-specific probe from the LI sequences amplified from
MY98 GCACAGGATTTTGTGTAGAGG Type 45 Type-specific probe
MY87 TATTAGCACTGCCACTGCTG Type 51 Type-specific probe previously isolated but yet unidentified
HPVs. The volume of amplification re¬
MY88 CCCAACATTTACTCCAAGTAAC Type 51 Type-specific probe action applied to the filters was doubled
MY81 CACTTCTACTGCTATAACTTGT Type 52 Type-specific probe for specimens that hybridized to the ge¬
MY82 ACACACCACCTAAAGGAAAGG Type 52 Type-specific probe neric probe with less than one-tenth the
MY101 CGCAACCACACAGTCTATGT W13A Clinical HPV probe
MY102 TTCTACCTTACTGGAAGACTGG W13A_Clinical HPV probe_ intensity of control HPV DNAs. Type-
MY103 GGAGGTCAÄTTTGCAAAAC W13A_Clinical HPV probe_ specific oligonucleotide probes were la¬
beled with phosphorus 32 using a stan¬
MY83 ATTAATGCAGCTAAAAGCACATT PAP88_Clinical HPV probe_ dard kinase reaction.23 Membranes
MY84 GATGCCCGTGAAATCAATCAA PAP88_Clinical HPV probe_ were prehybridized and hybridized as
MY86 TACTTGCAGTCTCGCGCCA PAP155 Clinical HPV probe
MY93 GCACTGAAGTAACTAAGGAAGG PAP251_Clinical HPV probe_ previously described, then washed at
57°C for the 6/11,31,33,35, and 52 type-
MY94 AGCACCCCCTAAAGAAAAGGA PAP251_Clinical HPV probe_ specific probes; 58°C for the 39, 51,
MY74 CATTTGTTGGGGTAACCAAC Type 16 Internal PCR primers
(+)
MY75 TAGGTCTGCAGAAAACTTTTC Type 16 for generic probe (
) W13A, PAP88, PAP155, and PAP251
MY76 TGTTTGCTGGCATAATCAAT Type 18 Internal PCR primers
(+)
-

probes; and at 59°C for the 16,18, and 45

MY77 TAAGTCTAAAGAAAACTTTTC Type 18 for generic probe (
) type-specific probes. Autoradiograph
CÄTATGCTGGGGTAATCAGG PAP88 Internal PCR primers
(+) exposures that gave an intensity com¬
-

MY47
MY48 CAGGTCTGCAGAAAAGCTGT PAP88 for generic probe (
)
MY49 TATTTGTTGGGGCAATCAG PAP238B (+)
Internal PCR primers
-
parable with the generic probe were an¬
MY50 CTAAATCTGCAGAAAACTTTT PAP238B_for generic probe ( ) alyzed using the same standards al¬
GH20 GAAGAGCCAAGGACAGGTAC ß-globin PCR primer ( + )
-

ready described.
PC04 CAACTTCATCCACGTTCACC ß-globin PCR primer ( )_ Samples hybridizing to a type-specif¬
PC03 ACACAACTGTGTTCACTAGC ß-globin Internal probe
-

ic probe were identified as having that

HPV type. Specimens hybridizing to
•Degenerate code indicates M A + C, R A + G, W A + T, and Y C +DNA
= = = = T. the generic probe, but not to any of the
t( + ) indicates primer for positive DNA strand; (-), primer for negative strand; HPV, human papillomavirus;
and PCR, polymerase chain reaction. type-specific probes, were identified as
"not typed" or "unidentified." Samples
scored as multiple infections included
only those that had more than one
known type.
Hybridization Analysis quate dot blot preparation. The generic
of PCR Products HPV probe was used to determine the Statistical Methods
presence of HPV in each sample. Filters
Dot blots of every specimen were pre¬ were prehybridized at 55°C for 30 min¬
Differences between population sub¬
pared after choosing the amplification utes in 6xSSC (lxSSC 0.15-mol/L = groups were assessed using x2 and t
(either 2- or 5-|xL reaction) with the sodium chloride and 0.015-mol/L triso- tests, as appropriate.
highest overall yield based on electro- dium citrate [pH 7.0]), 5xDenhart's, RESULTS
phoresis data. With about 80% of the 0.1% sodium dodecyl sulfate, and 100
specimens, the 2- and 5-(xL reactions mg/L of denatured salmon sperm DNA. Study Population
produced identical results; however, in Probes labeled with phosphorus 32 (1.0 Participation by those invited to take
about 20% of specimens, the 2-|xL reac¬ to 1.5 x 105 cpm/mL) were added and part in the study exceeded 95%. Alto¬
tion gave a higher yield. This phenome¬ allowed to hybridize at 55°C for 60 to 90 gether, 467 women were enrolled, most
non is probably the result of an inhibitor minutes. Blots were washed twice for 10 of whom were white (72%), young
in the sample or collection buffer. Two minutes in 2 x SSC and 0.1% sodium do¬ (mean age, 22.9 years; SD, 4.2), and
microliters of the reaction was mixed decyl sulfate at 56°C. Several autoradio- single (94%), had never been pregnant
with 100 |xL of denaturing solution graph exposures were obtained from 2 (81%), and were or had been involved in
(0.4N sodium hydroxide and 25-mmol/L to 48 hours after hybridization. Back¬ heterosexual relationships (97%). The
ethylenediamenetetraacetic acid) and ground was defined by negative con¬ median lifetime number of sexual part¬
trols, which included ß-globin amplifica¬ ners was four. Seventy percent had
applied to replicate nylon membranes.
One set of the blots was hybridized with tions of human genomic DNA. All used oral contraceptives and 41% were
a ß-globin probe (Table 1) to assess ade- determinations were made without current users. The majority of enrolled

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015

 Table 2.—Prevalence of HPV Infection at the Cervix Table 4. Prevalence of HP V Infection at the Cervix Compared With Cytological Evaluation of Papanicolaou
and Vulva (n 467 at Each Site)*
-
Smears Collected From 454 Women*
—

Genital Site Sampled, Condylomatous

No. (%) Normal Atyplaf Atypia Dysplasia
Test Result Cervix Vulva _(n 421)_(n=21)_(n 7)_(n=5) = =

PCR method, No. (%)_130 (31)_14 (67)_6

PCR (86)_4 (80)
Positive! 154 (33) 201 (43) ViraPap, No. (%) 30(7) 3(14) 1 (14) 1 (20)
Negative 298 (64) 255 (55)
Indeterminate 3 (1) 6 (1) *HPV indicates human papillomavirus; PCR, potymerase chain reaction.
Insufficient! 12 (3) 5 (1) tlncludes reactive changes and squamous cell abnormalities of undetermined significance.
ViraPap
Positive! 35 (7) 39 (8)
Negative 379 (81) 403 (86)
Indeterminate
Insufficient*
45
8
(10)
(2)
25
0
(5)
(0)
atypia that was metaplastic, reactive, 4). Of the five women with cervical dys-
or of undetermined significance. Sixty- plasia (CIN 1 or CIN 2), all were deter¬
•HPV indicates human one (13%) of the women reported having mined to be infected with HPV at one or
papillomavirus; PCR, poly-
merase chain reaction. had previous abnormal Papanicolaou both genital sites using the PCR meth¬
tPrevalence of HPV at either site was 46%.
smear results. od, although one patient tested positive
^Includes samples that were insufficient or not avail¬
able for analysis.
HPV Prevalence and at the vulva but not the cervix. One
§Prevalence of HPV at either site was 11%. patient was infected with HPV 31 at the
Type Distribution cervix, and the other patients were in¬
Of the total 467 women, 213 (46%) fected with HPV 51 or 52 or PAP88 or
Table 3.—Prevalence of HPV Types at Either
showed evidence of infection with HPV PAP155. Using ViraPap, we identified
Genital Site as Determined by PCR and
at one or both sites using the PCR meth¬ only the patient with HPV type 31 as
ViraPap
Methods* od, while only 51 (11%) were positive by positive. Of the cervical specimens from
ViraPap. We were unable to obtain cer¬ the women with condylomatous atypia
Prevalence, vical specimens from eight (2%) of the (n 7), six were HPV positive by PCR,
=
HPV Type n (% of Population)
467 participants. Of the PCR-positive while only one of these six was HPV
PCR method
6,11 16 (3) subjects, 146 (69%) of the women were positive by ViraPap. Seven (70%) of 10
16
18
40
24
(9) positive at both sites. Of the ViraPap- women with atypia of undetermined
(5) positive subjects, 23 (45%) ofthe women
31 22 (5) significance and seven (64%) of 11 wom¬
33 12 (3) were positive at both sites. Using the en with reactive atypia were infected
35 2 (<1) PCR system, we found 154 (33%) were with HPV at the cervix according to
39 15 (3)
45 5 (1) infected at the cervix and 201 (43%) PCR data. According to ViraPap, only
51 13 (3) were infected at the vulva (Table 2). All three of these women (all with reactive
52 21 (5)
W13A 22 (5) negative control specimens (see "Meth¬ atypia) were HPV positive at the cer¬
PAP88 22 (5) ods" section) were found to be HPV neg¬ vix. The HPV type distributions among
PAP 155 17 (4)
PAP251 11 (2)
ative. By ViraPap, only 35 (7%) of the the women with atypia were similar to
Not typed 61 (13) women were infected at the cervix and those found in the normal population
Any HPV infectiont 213 (46) 39 (8%) were infected at the vulva. For (data not shown). Of the five women
ViraPap both PCR and ViraPap, the distribu¬
6/11
16/18
11
30
(2)
tions of HPV types at each site were
presenting with anal or vulvar warts,
(6) four (80%) were found by PCR to be
31/33/35 23 (5)
Any HPV infectiont 51 (11) similar (data not shown). infected with HPV.
Using PCR, 152 (33%) of the 467 PCR-Based HPV Detection
*HPV indicates human papillomavirus; PCR, poly- women were found to be infected with
merase chain reaction.
tBy PCR, 39 women were infected by two identified types 6,11,16,18, 31,33,35,39,45,51, The improved generic HPV probe
viral types, 14 women by three, five women by four, and 52, or with previously isolated yet un¬ used in this study was significantly
two women by five viral types. identified HPV types W13A, PAP88, more effective in detecting amplified
tBy ViraPap, 13 women were infected by two viral
types from different groups. PAP155, or PAP251 (Table 3). Sixty- types than the generic oligonucleotide
one (13%) of the women were infected probe mixture previously described.16"
with other HPVs, which did not hybrid¬ Using plasmids and cell line DNA, we
ize with any of the type-specific probes. determined that this PCR system am¬
women were being seen only for a rou¬ The data reported in Table 3 represent plified and the improved generic probe
tine annual pelvic examination (87%), numbers of patients in whom a particu¬ detected the known HPV types 6, 11,
with the remainder being seen for con¬ lar HPV type was detected and are com¬ 16, 18, 30, 31, 33, 35, 39, 40, 42, 43, 45,
traception (8%), menstrual problems plicated by patient samples found to 51, 52, 53, 54, 55, 57, 58, 59 and at least
(3%), and associated miscellaneous contain more than one HPV type. Multi¬ 20 other unidentified types. The sensi¬
problems (2%) in addition to the ple HPV infections involving the known tivityof this PCR method, as deter¬
examination. and previously isolated types that we mined by analysis of ViraPap collection
Vulvar, anal, or vaginal warts were distinguished were found in 60 (13%) of tubes containing known amounts of
detected by gross visual inspection in the women. Using ViraPap, we found DNA, was less than 500 copies of cloned
five (1%) of the women in this study; 41 that 13 (3%) of the women were dually HPV 16 DNA or 10 HeLa cells (100 to
(9%) of the women had a history of geni¬ infected with viruses from two HPV- 500 copies of HPV 18) per collection
tal warts. Papanicolaou smears ob¬ type groups. tube.
tained from 454 (97%) of the women at
the time of enrollment in the study HPV Infection and Cytology PCR-ViraPap Comparison
showed that five (1%) of the women had To examine the relationship between Of the 51 women who were HPV posi¬
mild to moderate cervical dysplasia, abnormal cytology and HPV infection, tive by ViraPap, all were positive by
seven (1%) of the women had condylo- we compared Papanicolaou smear re¬ PCR. Of the women found positive by
matous atypia, and 21 (4%) of them had sults with HPV determinations (Table PCR but negative or indeterminant by

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015

 ViraPap (n 162), 44 (27%) were infect¬
= cervical HPV infection: 10% in upper- Although the ViraPap method was
ed with types included in the ViraPap- class women with normal cytology4; 16% easy to use, we found the PCR method
ViraType systems, while the remaining in university students and 18% in wom¬ superior in its analytical sensitivity and
118 (73%) women were infected with en attending a sexually transmitted dis¬ type detection. Using PCR, we were
types not tested for by ViraPap. In ease clinic in Seattle (Wash)3; 38% in able to detect 10 HeLa cells (100 to 500
cases found to be positiveby ViraPap, young black and Hispanic women7; and copies of HPV 18) per collection tube
type-determination data using the PCR up to 82% in Panamanian prostitutes.5 compared with 20 000 HeLa cells per
method correlated well with the Vira- Some researchers have reported even tube in the ViraPap "low positive con¬
Type data: 98% (40/41) agreement for higher prevalences using PCR meth¬ trol." Although the overall HPV preva¬
women identified as HPV type 6/11 or ods,24"26 but these findings have yet to be lence by PCR in this study is about four¬
HPV type 16/18; and 65% (15/23) agree¬ verified and in some cases have been fold higher than that of ViraPap, the
ment for women identified as HPV type retracted27 or refuted.28 prevalence for HPV types included in
31/33/35 by ViraType. This latter dis¬ It is clear from our results that HPV both systems is only about twofold high¬
cordance may be explained in part by infection is common in the healthy wom¬ er. The ViraPap method is limited in the
the cross-reactivity of ViraType HPV en we studied, who are likely to repre¬ number of HPV types detected. Over
31/33/35 probes with other HPV types. sent other young, sexually active wom¬ 70% of infections missed by ViraPap
To explore whether PCR-positive- en. It is possible that we are studying an contained types of HPV not included in
ViraPap-negative specimens were age-specific prevalence since infection the assay. Since the PCR method uti¬
more likely to represent improved sen¬ rates may decrease with increasing lizes consensus HPV primers and
sitivity rather than lower specificity or age.29,30 Although other genetic or envi¬ probes, it is able to detect many known
contamination of the PCR detection ronmental variables may be implicated as well as yet unidentified types.
method, we compared the sexual histor¬ in the progression to cervical disease, The PCR method revealed a high
ies of the 162 women with PCR-posi- infection with particular HPVs could prevalence of unidentified HPV types
tive-ViraPap-negative results with the still be predictive of disease in a subset and novel HPVs previously isolated
51 women who were ViraPap positive of women. Clearly, long-term prospec¬ from clinical samples (eg, W13A,
and the 254 women whose results were tive studies are needed to understand PAP88, PAP155, and PAP251). This
negative by both tests. We reasoned the implications of HPV infections in group of viruses is likely to be a mixture
that since genital HPV infection is sexu¬ asymptomatic women and the role of of novel benign HPVs, disease-associat¬
ally transmitted, the sexual histories of HPV in disease. ed HPVs, and possibly HPVs that are
women found accurately to be HPV pos¬ Although we expected a high correla¬ not sexually transmitted. These types
itive would differ significantly from the tion of HPV infection between the cervi¬ may also include some of the character¬
histories of women who were HPV neg¬ cal and vulvar sites, we detected HPV ized genital types numbered in the 30s,
ative. The distribution of the number of infection at both sites in only 69% ofthe 40s, 50s, and 60s31 for which we did not
lifetime sexual partners was similar infected women. It appears that HPV include type-specific probes. Further
among women with PCR-positive- can infect either the vulva or the cervix investigation of the nontyped speci¬
ViraPap-negative results and women (and possibly other genital sites) inde¬ mens (Table 3) using restriction diges¬
with ViraPap-positive results (mean, pendently. While site distribution could tion of the HPV amplification products16
8.2 vs 8.4; P- .91, t test; percent with reflect the tissue specificity of different indicates that this group of samples con¬
over five lifetime sexual partners, 48% HPV types, our typing data do not sug¬ tains at least 20 distinct secondary or
vs 57%; x2=l-2; P=.28). Compared gest this. The site distribution also novel HPV types. Although these con¬
with women with PCR-negative re¬ could be an indication of incomplete sensus primers amplify some dermal
sults, the women with PCR-positive- sample collection or inconsistent viral HPVs, none of the nontyped HPVs ana¬
ViraPap-negative results had more life¬ shedding, although the sensitivity of lyzed thus far have been identified as
time sexual partners (mean, 8.4 vs 5.2; PCR makes this possibility less likely. dermal types la, 5, or 8. Clearly, more
P= .0001, t test; percent with over five Clearly, more work is needed to deter¬ work needs to be done to obtain a more
lifetime sexual partners, 48% vs 24%; mine the factors that influence site dis¬ complete catalog of genital HPV types.
X2 25.89; P<.0001).
= tribution and whether infection at the It is interesting to note that two of the
vulva generally precedes and predis¬ women with cervical dysplasia and
COMMENT many of those with atypia harbored yet
poses to infection at the cervix.
We determined the prevalence of As expected, women diagnosed as unidentified types (W13A, PAP88,
HPV infection in a young, sexually ac¬ having condylomatous atypia or CIN PAP155, and PAP251), suggesting that
tive population of female university stu¬ had a higher prevalence of HPV infec¬ these secondary or novel genital HPVs
dents. Of the two DNA detection meth¬ tion. The rate of HPV infection in wom¬ may play a role in cervical disease. De¬
ods employed, the PCR-based method en with benign atypia was also high. A tection methods that recognize these vi¬
was found to be highly sensitive and recent study using ViraPap also found rus types clearly are informative re¬
useful in detecting a broad spectrum of similar prevalences of HPV in women search tools.
HPV types. with dysplasia and benign atypia (Ste¬ It is important to note that the preva¬
Our results using the ViraPap system ven Anderson, PhD, Joseph Chimera, lence of infections with multiple HPV
to detect HPV yielded a positivity rate PhD, written communication, June types was underestimated using either
of 7% at the cervix. Using our PCR- 1990). However, other researchers detection system. The ViraPap system
based system, we found a positivity rate have found infection rates in women does not distinguish double infections if
of33% at the cervix. Unfortunately, it is with atypia to be closer to those of cyto- the types are in the same probing group
difficult to compare these prevalence logically normal women.4 This discrep¬ (eg, a 16 and 18 double infection). The
data with previous studies because of ancy may indicate differences in inter¬ PCR method would not have detected
variation in populations and methods of pretation of Papanicolaou smears. double infections in which one HPV
analysis. For example, researchers us¬ However, we believe that in this popu¬ type was identified and the other was
ing DNA hybridization techniques have lation the presence of atypia may often unidentified, or if both types were un¬
reported a wide range in prevalence of indicate infection with HPV. identified. Currently, we are develop-

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015

 ing a system that utilizes restriction en- forhelp in planning and training; and Catherine tion. J Infect Dis. 1990;161:113-115.
donuclease digestion of amplification Ley for help in data analysis. We are also indebted 14. Shibata D, Arnheim N, Martin WJ. Detection
to Ira Friedman, MD, the medical director of the of human papillomavirus in paraffin-embedded tis-
products16 to detect multiple infections. University Health Service at the University of Cal¬ sue using the polymerase chain reaction. J Exp
The increased sensitivity of PCR may ifornia, Berkeley. We give special thanks to the Med. 1988;167:225-230.
have been critical for detecting HPV in staff at the University Health Service Women's 15. Kiviat NB, Koutsky LA, Critchlow CW, et al.
Clinic: nurse practitioners Mary Kenney, Sally Comparison of Southern transfer hybridization and
suboptimal samples and in identifying Stockton, and Irma Waldron for sample collection; dot filter hybridization for detection of cervical hu-
low-level infections. The only disadvan¬ Janis Fox-Davis, RN, and Lynanne Jacob, RN, for man papillomavirus infection with types 6, 11, 16,
tage of the extreme sensitivity of PCR patient consultation; Grace Smith, MD, for helpful 18, 31, 33, and 35. Am J Clin Pathol. 1990;94:561\x=req-\
is the potential for false-positive results discussions; and Keiko Kubo and Demeter Lyons 565.
for patient scheduling and contact. We thank Robin 16. Ting Y, Manos MM. Detection and typing of
arising from sample cross-contamina¬ Kurka and Donna Mooney for help in manuscript genital human papillomaviruses. In: Innis M, Gel-
tion or PCR product carryover. In this preparation. Many thanks to Mark Schiffman, MD, fand D, Sninsky J, White T, eds. PCR Protocols: A
study, extreme care was taken to pre¬ for critical review and extremely useful Guide to Methods and Applications. San Diego,
vent such contamination. Only dispos¬ discussions. Calif: Academic Press Inc; 1990:356-367.
able or positive displacement pipettes 17. Resnick RM, Cornelissen MTE, Wright DK, et
References al. Detection and typing of HPV in archival cervical
were used in sample handling and PCR cancer specimens using DNA amplification with
setup. In addition, we separated the 1. Koutsky LA, Galloway DA, Holmes KK. Epide- consensus primers. J Natl Cancer Inst. 1990;

pre- and post-PCR samples and re¬ miology of genital human papillomavirus infection. 82:1477-1484.
agents throughout all stages of the Epidemiol Rev. 1988;10:122-163. 18. Schiffman MH, Bauer HM, Lorincz AT, et al. A
2. Roman A, Fife KH. Human papillomaviruses: comparison of Southern blot hybridization and
work. The absence of any HPV-positive are we ready to type? Clin Microbiol Rev. polymerase chain reaction methods for the detec-
results from our numerous negative 1989;2:166-190. tion of human papillomavirus DNA. J Clin Micro-
controls suggested that contamination 3. Kiviat NB, Koutsky LA, Paavonen JA, et al. biol. In press.
did not occur at any point in sample Prevalence of genital papillomavirus infection 19. Gregoire L, Arella M, Campione-Piccardo J,
among women attending a college student health Lancaster WD. Amplification of human papilloma-
handling. We conclude that most, if not clinic or a sexually transmitted diseases clinic. J virus DNA sequences by using conserved primers.
all, of the positive results reflect the Infect Dis. 1989;159:293-302. J Clin Microbiol. 1989;27:2660-2665.
presence of HPV in the clinical speci¬ 4. Lorincz AT, Schiffman MH, Jaffurs WJ, Marlow 20. Snijders PJF, van de Brule AJC, Schrijne-
mens. Furthermore, the similarity of J, Quinn AP, Temple GF. Temporal associations of makers HFJ, Snow G, Meijer CJLM, Walboomers
human papillomavirus infection with cervical cyto- JMM. The use of general primers in the polymerase
women with PCR-positive-ViraPap-
logical abnormalities. Am J Obstet Gynecol. chain reaction permits the detection of a broad
negative results to women with Vira- 1990;162:645-651. spectrum of human papillomavirus genotypes. J
Pap-positive results with regard to sex¬ 5. Reeves WC, Arosemena JR, Garcia M, et al.
Genital human papillomavirus infection in Panama
Gen Virol. 1990;71:173-181.
21. National Cancer Institute Workshop. The 1988
ual histories suggests that the PCR
results were accurate. City prostitutes. J Infect Dis. 1989;160:599-603. Bethesda System for reporting cervical/vaginal cy-
6. Reeves WC, Brinton LA, Garcia M, et al. Hu- tologic diagnoses. JAMA. 1989;262:931-934.
The PCR method presented herein man papillomavirus infection and cervical cancer in 22. Kwok S, Higuchi R. Avoiding false positives
will undoubtedly play a role in future Latin America. N Engl J Med. 1989;320:1437-1441. with PCR. Nature. 1989;339:237-238.
7. Rosenfeld WD, Vermund SH, Wentz SJ, Burk 23. Maniatis T, Fritsch EF, Sambrook J. Molecu-
epidemiologic studies of HPV. The RD. High prevalence rate of human papillomavirus lar Cloning: A Laboratory Manual. New York,
broad range in type detection and high infection and association with abnormal Papanico- NY: Cold Spring Harbor Laboratory Press;
sensitivity will be important in assess¬ laou smears in sexually active adolescents. AJDC. 1982:122-123.
ing the incidence and prevalence of in¬ 1989;143:1443-1447.
8. Villa LL, Franco ELF. Epidemiologic corre-
24. Tidy JA, Vousden KH, Farrell PJ. Relation
between infection with a subtype of HPV16 and
fection. Further analyses were recently
lates of cervical neoplasia and risk of human papillo- cervical neoplasia. Lancet. 1989;1:1225-1227.
performed to determine the demo¬ mavirus infection in asymptomatic women in Bra- 25. Ward P, Parry GN, Yule R, Coleman DV,
graphic and behavioral risk factors for zil. J Natl Cancer Inst. 1989;81:332-340. Malcolm ADB. Human papillomavirus subtype
HPV infection in this population.30 This 9. Mullis KB, Faloona FA. Specific synthesis of 16a. Lancet. 1989;1:170.
work and other epidemiologic studies DNA in vitro via a polymerase-catalyzed chain re- 26. Young LS, Bevan IS, Johnson MA, et al. The
action. Methods Enzymol. 1987;155:335-350. polymerase chain reaction: a new epidemiological
employing these methods will contrib¬ 10. Saiki RK, Scharf S, Faloona F, et al. Enzymat- tool for investigating cervical human papilloma-
ute to our understanding of the natural ic amplification of \g=b\-globingenomic sequence and virus infection. BMJ. 1989;298:14-18.
history of HPV infection and the role of restriction site analysis for diagnosis of sickle cell
anemia. Science. 1985;230:1350-1354.
27. Tidy J, Farrell PJ. Retraction: human paillo-
mavirus subtype 16b. Lancet. 1989;2:1535.
HPVs in cervical disease. In addition,
11. Ferre F, Garduno F. Detection of human papil- 28. Manos M, Lee K, Greer C, et al. Looking for
prospective studies will determine the lomavirus types 6/11, 16, and 18 using the polymer- human papillomavirus type 16 by PCR. Lancet.
potential clinical applications of HPV ase chain reaction. Cancer Cells 7, Molecular Diag- 1990;335:734.
detection by PCR methods. nostics of Human Cancer. New York, NY: Cold 29. Kjaer SK, Engholm G, Teisen C, et al. Risk
Spring Harbor Laboratory Press; 1989:215-218. factors for cervical human papillomavirus and her-
12. Manos MM, Ting Y, Wright DK, Lewis AJ, pes simplex virus infections in Greenland and Den-
We are indebted to the following people whose Broker TR, Wolinsky SM. Use of polymerase chain mark: a population-based study. Am J Epidemiol.
effort and enthusiasm made this work possible: reaction amplification for the detection of genital 1990;131:669-682.
Anita O'Connor for patient contact and study ad¬ human papillomaviruses. Cancer Cells 7, Molecu- 30. Manos M. Prevalence of genital HPV infection
ministration; Kenneth Lee and Robert Resnick for lar Diagnostics of Human Cancer. New York, NY: in female university students determined by a
assistance in the Cetus laboratory; Joy Miller and Cold Spring Harbor Laboratory Press; 1989:209\x=req-\ PCR-based method. Presented at the Papilloma-
Harold Noell for conducting the ViraPap analyses 214. virus Workshop; May 16, 1990; Heidelberg,
at Roche Biomédical Laboratories; the DNA syn¬ 13. Pao CC, Lin C-Y, Maa J-S, Lai C-H, Wu S-Y, Germany.
thesis group at Cetus; John Sninsky, PhD, for his Soong Y-K. Detection of human papillomaviruses 31. de Villiers E-M. Heterogeneity of the human
support and encouragement; Jeff Waldman, MD, in cervicovaginal cells using polymerase chain reac- papillomavirus group. J Virol. 1989;63:4898-4903.

Downloaded From: http://jama.jamanetwork.com/ by a Columbia University User on 05/05/2015