Journal of the American Society of Cytopathology (2019) 8, 149e156

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.jascyto.org/

ORIGINAL ARTICLE

HPV status in women with high-grade dysplasia

on cervical biopsy and preceding negative HPV
tests
Yimin Ge, MDa,b,*, Roxanne R. Mody, MDc, Randall J. Olsen, MDa,b,
Haijun Zhou, MD, PhDa, Eric Luna, CT (ASCP)d,
Donna Armylagos, CT (ASCP)d, Natu Puntachart, MB (ASCP) CMa,
Heather Hendrickson, MBA, MB (ASCP) CMa, Mary R. Schwartz, MDa,
Dina R. Mody, MDa,b

a
Department of Pathology and Genomic Medicine, Houston Methodist Hospital, Houston, Texas
b
Weill Medical College of Cornell University, New York, New York
c
Department of Obstetrics and Gynecology, St. Joseph’s Hospital, Denver, Colorado
d
BioReference Laboratories, Houston, Texas

Received 28 November 2018; received in revised form 6 January 2019; accepted 8 January 2019

KEYWORDS Introduction A considerable number of patients with high-grade cervical lesions have undergone preced-
Human papillomavirus; ing human papillomavirus (HPV) tests with negative results. In the present study, we attempted to elucidate
HPV test; the factors potentially contributing to the findings by testing biopsied samples from these patients.
Papanicolaou test; Materials and methods Of the 1654 women with HPV testing and follow-up cervicovaginal biopsies from
HPV-cytology co-testing; March 1, 2013 to June 30, 2014, 21 of 252 women (8.3%) with biopsy-confirmed high-grade squamous in-
High-grade cervical lesion; traepithelial lesion (HSIL) or worse had had negative results from preceding high-risk (hr)HPV tests. The
Cervical cancer corresponding paraffin blocks were tested for HPV using the Cobas 4800 system, a DNA microarray against
40 HPV genotypes, and DNA sequencing.
Results HPV was detected in 20 (95%) of the 21 biopsies with HSIL or worse, including HPV16/18 in 4,
non-16/18 hrHPV in 10, and non-hrHPV in 6. HPV59 and HPV45 were 2.2 times more frequently detected
than HPV16/18 in these samples. One sample was negative for all 3 tests (5%).
Conclusions Our study has demonstrated that 8.3% of women with biopsy-confirmed HSIL or worse had pre-
ceding test results that were negative for hrHPV. The vast majority of the biopsied samples had detectable HPV,
primarily hrHPV genotypes (67%) with HPV59 and HPV45 predominance. This genotypic prevalence pattern

*Corresponding author: Yimin Ge, MD, Department of Pathology and College of Cornell University, 6565 Fannin Street, M227, Houston, TX
Genomic Medicine, Houston Methodist Hospital, Weill Cornell Medical 77030; Tel.: (713) 441-2283; Fax: (713) 793-1603.
E-mail address: yge@houstonmethodist.org (Y. Ge).

2213-2945/Ó 2019 The Authors. Published by Elsevier Inc. on behalf of American Society of Cytopathology. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
https://doi.org/10.1016/j.jasc.2019.01.001
150 Y. Ge et al.

was markedly different from those reported in the general population. Non-hrHPV genotypes contributed to 29%
of the cases, and HPV-negative cases were rare. In addition to the limited Cobas testing panel and rare possible
HPV-negative HSIL or worse, other possible contributing factors to the discrepancy include cytologic sampling,
interference material, technical errors, and reduced L1 gene expression in high-grade lesions.
Ó 2019 The Authors. Published by Elsevier Inc. on behalf of American Society of Cytopathology. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction database from March 1, 2013 to June 30, 2014 and identi-
fied 47,499 women who had undergone cytology and
Persistent infection with high-risk human papillomavirus hrHPV cotesting with the Cobas 4800 system (Roche Mo-
(hrHPV) causes most cervical cancers and precancerous le- lecular Diagnostics, Pleasanton, Calif). The Pap tests were
sions.1 In recent years, hrHPV testing has been increasingly performed using liquid-based methods with either the
used in clinical practice to triage equivocal cytology or as ThinPrep (Hologic, Madison, Wis) or SurePath (Becton
cotesting with cytology to maximize the detection rate of high- Dickinson, Franklin Lakes, NJ) platform. Our study was
grade cervical lesions in women aged 30 years.2 Recent conducted within a large general screening population in
studies conducted in Europe and Canada have demonstrated Texas with hrHPV prevalence and HSIL reporting rates
that primary hrHPV screening exhibits greater sensitivity in similar to the general population reported in other estab-
detecting precursor lesions of cervical cancer compared with lished trials and surveys performed in the United States.18
cytology alone in single and multiple rounds of screening.3-8 The cytology reporting performance in our laboratory
However, the efficiency of hrHPV testing alone as a primary benchmarked well with the national database reported by
screening method for cervical cancer has remained controver- the College of American Pathologists.
sial.9 After the Food and Drug Administration approved the
Cobas HPV test in April 2014 as an option for primary cervical
Biopsy confirmation
cancer screening in women aged 25 years, an interim clinical
guideline for the test was reported.10 Owing to the paucity of
The cases with follow-up biopsies performed within 1 year
data, this interim guideline was largely based on studies per-
of the cytologyeHPV cotesting were identified. Lesions
formed in Europe rather than prospective US-based studies.11,12
equal to or worse than HSIL were considered high-grade
Given that HPV cytology cotesting has been increasingly
lesions, including HSIL (cervical intraepithelial neoplasia
selected for women aged 30 years, cervical dysplasia and
[CIN]2, CIN3), adenocarcinoma in situ, and invasive car-
cancer with negative test results have been frequently
cinomas. Immunohistochemical staining for p16 was per-
encountered in clinical practice in the United States13 and
formed for all possible CIN2 lesions, and a diagnosis of
elsewhere.14 In 2015, Blatt et al.15 reported an extensive
CIN2 was rendered if a block-like immunostaining pattern
study involving 256,648 cases with cytologic and HPV
was observed. Endometrial lesions were excluded from the
cotesting. They demonstrated that 19% of women with cer-
present study.
vical cancer could have had a misdiagnosis if only primary
HPV cervical screening had been used.15 It has been reported
that 9% to 25% of patients with squamous cell carcinoma will HPV DNA extraction and genotyping
have had negative HPV tests in the preceding 1 to 5 years.16
A recent study suggested that more than one half of the Cobas 4800 system
cervical cancer cases with preceding negative HPV test re- Paraffin blocks from the biopsies with HSIL or worse in
sults could be truly hrHPV-negative carcinoma.17 In the women with negative hrHPV tests in the preceding cytology
present study, we attempted to elucidate the factors poten- specimens were obtained. Ten unstained paraffin sections (4
tially contributing to the findings by performing HPV tests on mm each) from each case were collected with extreme pre-
biopsied samples from women who had biopsy-confirmed caution. The microtome blade and water bath were changed
high-grade squamous intraepithelial lesions (HSILs) or between cases to avoid potential cross-contamination. The
worse and preceding HPV tests with negative results. sections were kept in closed containers before DNA
extraction. The lesional tissue was microdissected from the
paraffin sections and eluted with SurePath preservative fluid
Materials and methods (Becton Dickinson). The Cobas 4800 HPV test uses primers
and probes designed to detect 14 hrHPV genotypes. An
Patient population additional primer pair and probe were used to detect the
human b-globin gene as a control. HPV DNA was isolated
We retrospectively reviewed 130,648 Papanicolaou (Pap) on the Cobas 4800. Subsequently, a real-time polymerase
test results recorded in our Laboratory Information System chain reaction was performed on the Cobas 4800, using the
HPV detected in most of HSIL biopsies 151

Table 1 Human papillomavirus test results on biopsy samples from women with preceding negative test result.
Age (y) Previous cytology and HPV test results Diagnosis and HPV genotyping results from biopsy specimens
Cytology diagnosis Negative HPV tests Biopsy diagnosis Cobas 4800 DNA microarray DNA sequencing
32 ASC-H 1 CIN3 16 16, 18,67, 83 16, 18
23 ASC-US 2 CIN2 Non-16/18 39, 59. 66. 69 66
22 ASC-US 1 CIN2 Non-16/18 59, 66 66
22 LSIL 2 CIN2 Non-16/18 58, 59, 66, 91 66
28 ASC-US 1 CIN3 Non-16/18 45 45
43 ASC-H 1 CIN2 Non-16/18 53, 59, 90 53
37 ASC-H 1 CIN3 Non-16/18 11, 59, 91 59
36 NILM 3 CIN3 Non-16/18 6, 16(W) 16
25 ASC-H 1 CIN3 Non-16/18 6, 83, 84 84
21 HSIL 1 CIN3 Non-16/18 61 61
27 ASC-US 1 CIN2 Non-16/18 62 62
26 ASC-H 1 CIN3 Negative 11(W), 45 45
29 HSIL 3 CIN3 Negative 16(W), 45, 84 45
19 ASC-US 1 CIN2 Negative 68A(W), 69 69
44 HSIL 2 CIN3 Negative 81 81
49 LSIL 2 CIN2 Negative 55 55
35 NILM 1 CIN2 Negative Negative Negative
32 ASC-H 1 CIN2 Invalid 18 18
43 HSIL 1 CIN3 Invalid 11, 45, 59 11
46 HSIL 2 CIN3 Invalid 59, 69 69
25 ASC-H 1 CIN3 Invalid 83 83
Abbreviations: HPV, human papillomavirus; ASC-H, atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion; ASC-US, atypical
squamous cells of undetermined significance; CIN2, cervical intraepithelial neoplasia, grade 2; CIN3, cervical intraepithelial neoplasia, grade 3; LSIL, low-
grade squamous intraepithelial lesion; NILM, negative for intraepithelial lesion or malignancy; HSIL, high-grade squamous intraepithelial lesion; Non-16/
18, high-risk HPV excluding HPV16/18; W, weakly positive.

primer pairs and probes to detect and genotype HPV16/18 Categorization of HPV genotype and statistical
and other hrHPV genotypes. analysis

HPV DNA microarray (GoodGene 40 HPV genotyping The 2009 recommendations of the expert working group at
chip) the International Agency for Research on Cancer (IARC)
HPV DNA was extracted from the paraffin sections, as categorized HPV into 4 groups: carcinogenic (group 1),
previously reported,19 and the L1 region of the HPV probably carcinogenic (group 2A), possibly carcinogenic
genome was amplified by polymerase chain reaction. The (group 2B), and not classifiable (group 3).20 The HPV ge-
amplified HPV DNA was then labeled with Cy5 and hy- notypes in IARC groups 1 and 2A are commonly referred to
bridized with an HPV DNA microarray chip with 40 HPV as hrHPV. These genotypes include 16, 18, 31, 33, 35, 39,
genotypes, including 14 hrHPV and 26 non-hrHPV geno- 45, 51, 52, 56, 58, 59, and 68A/68B. In the present study,
types (GG HPV DNA Genotyping Chip Kit; GoodGene Inc, IARC groups 2B and 3 have been referred to as non-hrHPV.
Seoul, Korea). The signal was visualized using a GenePix All statistical analyses were performed with STATA version
4000B Microarray Scanner (Molecular Devices, Inc, Sun- 15 (StataCorp LP, College Station, TX). Significance was
nyvale, Calif). defined as two-tailed P < 0.05.

Direct HPV DNA sequencing Institutional review board approval

Conventional direct DNA sequencing methods were used to
further confirm the HPV DNA microarray data for the bi- The institutional review board of Houston Methodist Hos-
opsy samples from women with preceding negative HPV pital Research Institute approved the present study.
tests, as previously described.19 The sequence data obtained
by automated DNA sequencing were analyzed by the Basic Results
Local Alignment Search Tool search tool (available at:
http://www.ncbi.nlm.nih.gov/BLAST/) for HPV genotypes. Interpretable follow-up biopsy specimens were obtained
DNA sequencing identified the most dominant genotype in from 1654 women with a mean age of 37.9 years (range,
a given specimen and served as a confirmatory assay. 15.6-94.5). The corresponding Pap tests were performed on
152 Y. Ge et al.

12

10 Microarray
Cobas
8

Number of cases
All negaƟve

0
Non-16/18 Non-hrHPV HPV 16/18 NegaƟve
hrHPV
Figure 1 Human papillomavirus (HPV) testing results from bi-
opsy samples with high-grade squamous intraepithelial lesions
Figure 3 Detection of human papillomavirus (HPV) genotypes
(HSILs) from women with preceding negative HPV test results.
in biopsy samples using different testing methods. hrHPV, high-
hrHPV, high-risk HPV.
risk HPV.

either ThinPrep (n Z 867; 52.4%) or SurePath (n Z 787;

(ASC-H, n Z 7); and HSIL (n Z 5). The age of the HPV-
47.6%) platforms. Among the 1652 follow-up biopsy
negative group of women ranged from 19 to 49 years, with
specimens from women with cytologyeHPV cotesting, 252
an average age of 32 years, which was younger than the
were high-grade cervical lesions (HSIL or worse) by his-
average of 38 years for this screening population.
tologic evaluation. The preceding Pap tests had detected 230
The HPV test results from the biopsy samples from the 21
of the 252 cases, with a sensitivity of 91.3% (95% confi-
women in the HPV-negative group are summarized in
dence interval [CI], 86.7%-94.1%). In contrast, the HPV
Table 1 and Fig. 1. The Cobas HPV test results were positive
tests were positive in 231 of the 252 cases, with a sensitivity
in 11 biopsies (52%) with HPV16 in 1 and non-16/18 hrHPV
of 91.7% (95% CI, 87.1%-94.5%). No statistically signifi-
in 10 cases. Of the 11 cases with positive Cobas genotyping
cant difference (P Z 1.0) was found between the sensitiv-
results, 7, including 1 cases of HPV16 and 6 cases of non-6/
ities of the Pap and HPV tests in detecting biopsy-confirmed
18 hrHPV, were confirmed by DNA microarray or DNA
high-grade cervical lesions.
sequencing assays. Genotyping discrepancies were found in
Of the 252 women with biopsy-confirmed HSIL or
the 4 remaining cases with non-16/18 hrHPV found on Cobas
worse, 21 (8.3%; CIN2 in 9 and CIN3 in 12) had negative
but non-hrHPV genotypes (n Z 3) or HPV16 (n Z 1) found
Cobas HPV test results in the preceding cervical samples
on DNA microarray or sequencing assays.
(Table 1). The preceding Pap test results for the 21 women
Of the remaining 10 biopsies that were negative or inade-
included negative for intraepithelial lesions or malignancy
quate on Cobas testing, the DNA microarray test detected
(n Z 2); atypical squamous cells of undetermined signifi-
HPV16 or HPV18 in 2, non-16/18 hrHPV in 4, and non-
cance (ASC-US; n Z 5); low-grade squamous intra-
hrHPV in 3 cases, findings that were consistent with the results
epithelial lesion (LSIL; n Z 2); atypical squamous cells,
of the DNA sequencing assay. The concordance rate between
cannot rule out high-grade squamous intraepithelial lesion
Cobas and microarray/sequencing assays was 64% for those
positive on Cobas test. The sensitivity of the Cobas HPV test
8 for hrHPV in the biopsy tissue was 60% (9 of 15). Only 1
7 biopsy was negative for all 3 tests (1 of 21; 5%). A block-like
6 Red: hrHPV p16 immunostaining pattern was observed in all cases of CIN2.
Blue: non-hrHPV Overall, HPV genotypes were detected in 20 (95%) of 21
5
Frequency

4 biopsies with a diagnosis of HSIL or worse in women with

3 preceding negative Cobas HPV test results. Of the 20 biopsy
2 specimens with positive HPV tests, 14 (70%) had hrHPV
1 genotypes. Collectively, non-16/18 hrHPV genotypes (10 of
0
20; 50%) were more commonly identified than HPV16/18
(4 of 20; 20%) and non-hrHPV (6 of 20; 30%). Of the 14
59
45
16
66
69
83
18

84
91
39
68A
11
53
55
58
62
61
67
81
90
6

HPV genotypes
cases that were positive for hrHPV, HPV59 was the most
common genotype detected (n Z 7), followed by HPV45 (n
Figure 2 Human papillomavirus (HPV) genotype-specific prev- Z 4), HPV16 (n Z 3), HPV18 (n Z 2), HPV39 (n Z 1),
alence in biopsy-confirmed high-grade squamous intraepithelial le- HPV68A (n Z 1), and HPV58 (n Z 1; Fig. 2). HPV59/45
sions from women with preceding negative HPV test results. were 2.2 times more frequently detected than HPV16/18 in
hrHPV, high-risk HPV. these samples. Multiple genotypic HPV infections were
HPV detected in most of HSIL biopsies 153

common (13 of 20, 65%) in this cohort and primarily the biopsy samples with hrHPV genotypes were negative or
included non-16/18 hrHPV genotypes (Table 1). Of the 6 inadequate on Cobas testing, and many of those samples were
cases with non-hrHPV infection, single HPV genotypic also weakly positive on the DNA microarray assay. The tar-
infection was detected in 5 of the cases (83%), including geted viral DNA in these samples could have been present in
HPV genotypes 55, 61, 62, 81, and 83. One of the 6 cases amounts less than the threshold of the Cobas test.
had multiple non-hrHPV genotypes, including HPV6, In addition to possible inadequate sampling, high-grade
HPV83, and HPV84. cervical lesions usually occur in women with persistent
HPV infection characterized by overexpression of E6/E7
oncogenes after viral DNA integration into the host genome.
Discussion Compared with the early phase of productive infection, the
viron production and L1 gene expression might be signifi-
It is commonly perceived that hrHPV testing has greater cantly lower in women with persistent HPV infection when
sensitivity than cytomorphology in detecting cervicovaginal the incidence of high-grade lesions increases. In support of
dysplasia based on data from several large trials.3-8,11 this notion, recent studies have found that HPV E6/E7
However, our earlier study13 from our clinical quality mRNA testing has improved the specificity for CIN2-
assurance database demonstrated that the sensitivities of positive cervical lesions compared with HPV DNA as-
hrHPV and Pap tests in predicting high-grade lesions or any says.21-23 We recently studied the performance of Cobas
dysplastic lesions on follow-up biopsy samples were sta- HPV DNA test and Aptima E6/E7 mRNA HPV test among
tistically similar (91.3% versus 90.9% for high-grade le- 4562 women with cotesting and follow-up biopsies. We
sions, P Z 1.0; and 80.8% versus 81.2% for any dysplastic found that Aptima E6/E7 mRNA testing had significantly
lesions, P Z 0.86). In the present study, we found that 8.3% greater specificity and positive predictive values than Cobas
(21 of 252) of the biopsy-confirmed high-grade cervical testing for biopsy-confirmed HSILs or worse.24 Using
lesions had preceding negative hrHPV tests. This finding immunohistochemistry, Grapsa et al25 demonstrated that the
was not surprising because HPV-negative HSIL or cervical p16-positive/L1-negative pattern was significantly more
cancer cases have been reported in earlier studies. For common in HSILs than in LSILs. Given that a sufficient
example, a study involving 256,648 cases with cytologic amount of the HPV L1 gene is required for most of the HPV
and HPV cotesting demonstrated that 19% of the women DNA assays, including the Cobas test, reduced L1 expres-
with cervical cancer had preceding negative hrHPV test sion could contribute to false-negative HPV test results in a
results.15 Preceding negative HPV test results have also number of women with high-grade cervical lesions.
been reported in up to 25% of women with cervical The HPV16/18 genotypes were less commonly detected
cancer.16 than non-16/18 hrHPV genotypes among women with
We further analyzed the 21 biopsy samples with pre- positive hrHPV findings in the biopsy samples (4 of 14
ceding negative hrHPV test results using multiple sensitive [29%] versus 10 of 14 [71%]; Fig. 3). This pattern of
assays, including Cobas 4800 system, HPV DNA micro- genotypic prevalence markedly differed from that observed
array, and direct DNA sequencing. Our data showed that in women in the same study population with biopsy-
various HPV genotypes were detected by at least one of the confirmed HSILs or worse and previous positive HPV test
assays in most (95%) of the biopsy samples, despite pre- results. In the latter group, the prevalence for HPV16/18 and
vious negative HPV test results on cytology samples non-16/18 HPV was about equal. HPV59 and HPV45 were
(Table 1, Fig. 1). In the biopsy samples with hrHPV the 2 leading genotypes detected in the cohort with previous
infection, the Cobas test identified more than one half (57%) negative HPV test results and accounted for 90% of those
but missed the remaining cases (43%). Cobas testing was infected with non-16/18 hrHPV (Fig. 2). This is an unusual
also negative in 6 samples (29%) with non-hrHPV geno- pattern compared with the reported data on genotypic
types (Fig. 3). Only 1 sample was negative using all three composition in women with HSILs in the United States (in
HPV assays (5%). These findings have demonstrated that the decreasing prevalence of HPV16, HPV18, HPV31,
HPV infection is associated with the vast majority of HPV58, HPV35, HPV56, HPV52, HPV33, HPV66, and
biopsy-confirmed HSIL or worse lesions despite preceding HPV51).26-28 The genotypic prevalence pattern we detected
negative HPV test results, and hrHPV genotypes were the was also clearly inconsistent with the results from our earlier
most common causative agents (70%) of these high-grade study, which showed that HPV45 and HPV59 ranked 7th
cervical lesions. and 14th, respectively, in a local cohort of women with
The underlying reasons for the discrepancy between pos- cervical dysplasia.29 This suggests that HPV59 and HPV45
itive hrHPV test results in biopsy samples and previous might have lower expression levels of the L1 gene
negative hrHPV test results in cytology samples are not compared with other hrHPV genotypes, especially in high-
known. Several factors could potentially contribute to the grade lesions with HPV DNA integration. Another possi-
negative HPV test results in cytology samples, including a bility is that the Cobas test might have relatively lower
low viral load, inadequate sampling, and interference material sensitivity in detecting non-16/18 hrHPV genotypes, espe-
or technical errors. We found that approximately one third of cially HPV59 and HPV45.
154 Y. Ge et al.

Non-hrHPV genotypes were less commonly detected in ours,35,36 have indicated that the HPV genotypes might have
biopsy samples with HSIL or worse and preceding negative competitive interactions in women with multiple genotypic
HPV test results (6 of 21; 29%). Three of the 6 biopsy samples infections. However, it is unknown whether such in-
tested positive for non-16/18 hrHPV on the Cobas test; teractions among HPV genotypes are significant enough to
however, non-hrHPV genotypes were confirmed by DNA affect HPV detection. Further studies are necessary.
microarray or sequencing assays, suggesting possible cross- One of the 21 biopsied samples with HSIL or worse
reactions among the genotypes on the Cobas test. The non- tested negative with all 3 HPV assays, even with multiple
hrHPV genotypes detected in the 6 samples included HPV repeated attempts. It is unknown whether the negative HPV
genotypes 6, 55, 61, 62, 83, and 84 (Fig. 2). Understandably, assay findings represented a genuine HPV-negative HSIL or
the preceding Cobas HPV test results from the cytology false-negative test results. Although persistent HPV infec-
specimens were negative because the Cobas test panel is tion has been considered the cause of most cervical cancers,
limited to hrHPV. Abnormal cytomorphology was detected in HPV is not always detected before the cancer diagnosis.
previous Pap tests in all 6 cases and included ASC-US (n Z One study indicated that only 91% of cervical squamous cell
1), LSILs (n Z 1), ASC-H (n Z 2), and HSIL (n Z 2). Testing carcinomas were HPV positive using the linear array HPV
of follow-up biopsy specimens showed CIN2 in 2 cases and genotyping test.37 The data from the Kaiser Permanente
CIN3 in 4 cases. Of the 6 biopsy samples, 5 (83%) had a single study showed that 37% of women with cervical squamous
HPV infection with genotypes 55, 61, 62, 81, or 83. In the cell carcinoma were HPV-negative 5 years before the his-
absence of other HPV genotypes, these non-hrHPV geno- tologic diagnosis of cervical cancer.38 However, multiple
types were most likely the causative agents for high-grade factors should be considered before considering a case is
cervical lesions in these women. genuine HPV-negative HSIL or cancer. False-negative test
Using the IARC classification, these non-hrHPV geno- results can be caused by a low viral load, small lesion size,
types belong to group 3 and are commonly referred to as interfering material, technical errors, a limited testing panel,
“non-hrHPV” owing to insufficient data on their oncogenic or insufficient sensitivity of the assays.
potential. We believe that some of these genotypes might Several factors could have affected the interpretation of
not be as benign as commonly perceived and might pote- our results. First, ours was a retrospective study rather than a
tially cause cervical dysplasia or even cancer, especially in controlled trial. Although patient selection bias was un-
women with “negative” hrHPV test results. In an earlier avoidable, the present study more likely reflects real-world
study, we first reported that HPV90, one of the “non- clinical practice than would a controlled trial. Second, the
hrHPV,” was associated with cervical dysplasia, including study analyzed HPV tests performed on SurePath and
HSILs, in a North American population.30 In a Western blot ThinPrep samples together, and HPV-negative cases with
assay, Fu et al.31 demonstrated that the K16N mutation of biopsy-confirmed HSIL or worse lesions were detected in
HPV90 E6 enabled it to fully degrade p53 as effectively as both preparations. It has been suggested that some liquid-
HPV16 E6 in a single-transfected cell assay. In the present based preparations can interfere with HPV testing sensitivity
study, we identified the association of additional “non- and, thus, might have affected the interpretation of our data.
hrHPV” genotypes (HPV55, HPV61, HPV62, HPV81, and Third, the study cohort was relatively small and lacked
HPV83) with HSILs or worse as a single genotypic infec- patients with glandular lesions or carcinoma. In addition, the
tion. These nonconventional HPV genotypes have often lesions in the biopsy samples were generally small and
been overlooked because they are undetectable by most could have affected the HPV assays, despite careful tissue
HPV assays commonly used in practice. In addition, these dissection. Large prospective studies are necessary to further
genotypes have generally been perceived as low risk or even validate the findings.
nononcogenic. However, our data have indicated that at
least several of those “non-hrHPV” genotypes likely are the
genuine causative agents for high-grade cervical lesions and Conclusions
might play increasingly important roles in the development
of cervical cancer and precancerous lesions in the post-HPV Our study has demonstrated that 8.3% of women with
vaccination era. Further studies are needed to elucidate the biopsy-confirmed HSIL or worse had negative hrHPV test
oncogenic potential of these uncommon HPV genotypes. results in the preceding cytology samples. Despite the pre-
Our study also demonstrated a high rate of multiple vious negative HPV test results, a wide range of HPV ge-
genotypic HPV infection (62%) in women with biopsy- notypes were detected in the vast majority (95%) of biopsy
confirmed HSIL or worse and previous negative HPV test samples. These women were often infected by hrHPV
results. The rate was considerably higher than the 20% to (67%) with HPV59 and HPV45 predominance, a genotypic
40% in most previously reported studies, with variations prevalence pattern markedly different from that encountered
dependent on the age and severity of the cervical le- in the general population. Less commonly, non-hrHPV ge-
sions.29,32-34 In this general population, women with mixed notypes were associated with 29% of HSIL or worse lesions
HPV16/18 and non-16/18 hrHPV found on preceding HPV in the cohort, primarily with a single genotype infection. No
tests accounted for only 17%. Several studies, including detectable HPV was found in 1 of the biopsy samples after
HPV detected in most of HSIL biopsies 155

analysis using all 3 HPV assays. The latter results might be prevention and early detection of cervical cancer. Am J Clin Pathol.
attributable to a genuine noneHPV-driven lesion. In addi- 2012;137:516e542.
10. Huh WK, Ault KA, Chelmow D, et al. Use of primary high-risk human
tion to infection by non-hrHPV genotypes or possible papillomavirus testing for cervical cancer screening: interim clinical
noneHPV-related dysplasia, multiple other factors could guidance. Obstet Gynecol. 2015;125:330e337.
have contributed to the false-negative hrHPV test results in 11. Wright Jr TC, Stoler MH, Behrens CM, et al. The ATHENA human
preceding cytology samples, including inadequate sampling, papillomavirus study: design, methods, and baseline results. Am J
interfering material, technical errors, and reduced L1 gene Obstet Gynecol. 2012;206:46.e1e46.e11.
12. Austin RM, Zhao C. Is 58% sensitivity for detection of cervical intra-
expression in high-grade lesions. Additional studies are epithelial neoplasia 3 and invasive cervical cancer optimal for cervical
necessary to validate the findings and elucidate the screening? CytoJournal. 2014;11:14.
contributing factors and underlying mechanisms. 13. Zhou H, Mody RR, Luna E, et al. Clinical performance of the Food and
Drug Administration-approved high-risk HPV test for the detection of
high-grade cervicovaginal lesions. Cancer Cytopathol. 2016;124:
Acknowledgments 317e323.
14. Vassilakos P, Tran PL, Sahli R, Low N, Petignat P. HPV-negative
The authors thank Woo-Chul Moon, MD, PhD, for assisting CIN3 and cervical cancer in Switzerland: any evidence of impact on
the human papillomavirus genotyping tests, and Helen screening policies? Swiss Med Wkly. 2017;147:w14559.
Chifotides, PhD, for editing the manuscript. 15. Blatt AJ, Kennedy R, Luff RD, Austin RM, Rabin DS. Comparison of
cervical cancer screening results among 256,648 women in multiple
clinical practices. Cancer Cytopathol. 2015;123:282e288.
Funding sources 16. Zhao C, Li Z, Nayar R, et al. Prior high-risk human papillomavirus
testing and Papanicolaou test results of 70 invasive cervical carcinomas
diagnosed in 2012: results of a retrospective multicenter study. Arch
The present study was funded by a microgrant from the Pathol Lab Med. 2015;139:184e188.
Department of Pathology and Genomic Medicine, Houston 17. Tao X, Zheng B, Yin F, et al. Polymerase chain reaction human papil-
Methodist Hospital, Houston, Texas. lomavirus (HPV) detection and HPV genotyping in invasive cervical
cancers with prior negative HC2 test results. Am J Clin Pathol.
2017;147:477e483.
Conflict of interest disclosures 18. Mody DR, Krishnamurthy S, Anton R, Thrall M. Diagnostic Pathol-
ogy: Cytopathology. Philadelphia, PA: Lippincott, Williams & Wil-
The authors made no disclosures. kins; 2014.
19. Zhou H, Schwartz MR, Coffey D, et al. Should LSIL-H be a distinct
cytology category? A study on the frequency and distribution of 40 hu-
References man papillomavirus genotypes in 808 women. Cancer Cytopathol.
2012;120:373e379.
1. Schiffman M, Castle PE. The promise of global cervical-cancer preven- 20. International Agency for Research on Cancer. International Agency for
tion. N Engl J Med. 2005;353:2101e2104. Research on Cancer Monographs: Human Papillomaviruses. 100B.
2. Moyer VA. Screening for cervical cancer: U.S. Preventive Services Lyon, France: International Agency for Research on Cancer; 2009.
Task Force recommendation statement. Ann Intern Med. 2012;156: 21. Szarewski A, Mesher D, Cadman L, et al. Comparison of seven tests
880e891. for high-grade cervical intraepithelial neoplasia in women with
3. Gyllensten U, Gustavsson I, Lindell M, Wilander E. Primary high-risk abnormal smears: the Predictors 2 study. J Clin Microbiol. 2012;50:
HPV screening for cervical cancer in post-menopausal women. Gyne- 1867e1873.
col Oncol. 2012;125:343e345. 22. Castle PE, Eaton B, Reid J, Getman D, Dockter J. Comparison of hu-
4. Leinonen MK, Nieminen P, Lönnberg S, et al. Detection rates of pre- man papillomavirus detection by Aptima HPV and Cobas HPV tests in
cancerous and cancerous cervical lesions within one screening round of a population of women referred for colposcopy following detection of
primary human papillomavirus DNA testing: prospective randomised atypical squamous cells of undetermined significance by Pap cytology.
trial in Finland. BMJ. 2012;345:e7789. J Clin Microbiol. 2015;53:1277e1281.
5. Malila N, Leinonen M, Kotaniemi-Talonen L, et al. The HPV test has 23. Iftner T, Becker S, Neis KJ, et al. Head-to-head comparison of the
similar sensitivity but more overdiagnosis than the Pap testda rando- RNA-based Aptima human papillomavirus (HPV) assay and the
mised health services study on cervical cancer screening in Finland. Int DNA-based hybrid Capture 2 HPV test in a routine screening popula-
J Cancer. 2013;132:2141e2147. tion of women aged 30 to 60 years in Germany. J Clin Microbiol.
6. Ogilvie GS, Krajden M, van Niekirk DJ, et al. Primary cervical cancer 2015;53:2509e2516.
screening with HPV testing compared with liquid-based cytology: re- 24. Ge Y, Christensen P, Luna E, et al. Performance of Aptima and Cobas
sults of round 1 of a randomised controlled trialdthe HPV FOCAL HPV testing platforms in detecting high-grade cervical dysplasia and
study. Br J Cancer. 2012;107:1917e1924. cancer. Cancer Cytopathol. 2017;125:652e657.
7. Rijkaart DC, Berkhof J, van Kemenade FJ, et al. HPV DNA testing in 25. Grapsa D, Frangou-Plemenou M, Kondi-Pafiti A, et al. Immunocyto-
population-based cervical screening (VUSA-Screen study): results and chemical expression of P53, PTEN, FAS (CD95), P16INK4A and
implications. Br J Cancer. 2012;106:975e981. HPV L1 major capsid proteins in ThinPrep cervical samples with squa-
8. Ronco G, Dillner J, Elfström KM, et al. Efficacy of HPV-based mous intraepithelial lesions. Diagn Cytopathol. 2014;42:465e475.
screening for prevention of invasive cervical cancer: follow-up of 26. Castellsagué X, de Sanjosé S, Aguado T, et al. HPV and cervical
four European randomised controlled trials. Lancet. 2014;383: cancer in the world: 2007 report. Vaccine. 2007;25(suppl 3):
524e532. C1eC230.
9. Saslow D, Solomon D, Lawson HW, et al. American Cancer Society, 27. Clifford GM, Smith JS, Aguado T, Franceschi S. Comparison of HPV
American Society for Colposcopy and Cervical Pathology, and Amer- type distribution in high-grade cervical lesions and cervical cancer: a
ican Society for Clinical Pathology screening guidelines for the meta-analysis. Br J Cancer. 2003;89:101e105.
156 Y. Ge et al.

28. Smith JS, Lindsay L, Hoots B, et al. Human papillomavirus type dis- IARC HPV Prevalence Surveys. Cancer Epidemiol Biomarkers Prev.
tribution in invasive cervical cancer and high-grade cervical lesions: 2010;19:503e510.
a meta-analysis update. Int J Cancer. 2007;121:621e632. 34. Beca F, Pinheiro J, Rios E, Pontes P, Amendoeira I. Genotypes and
29. Zhou H, Mody D, Schwartz MR, Schwartz DM. Genotype-specific prevalence of HPV single and multiple concurrent infections in women
prevalence and distribution of human papillomavirus genotypes in un- with HSIL. Diagn Cytopathol. 2014;42:919e923.
derserved Latino women with abnormal Papanicolaou tests. J Am Soc 35. Salazar KL, Zhou HS, Xu J, et al. Multiple human papilloma virus in-
Cytopathol. 2014;3:42e48. fections and their impact on the development of high-risk cervical le-
30. Quiroga-Garza G, Zhou H, Mody DR, Schwartz MR, Ge Y. Unex- sions. Acta Cytol. 2015;59:391e398.
pected high prevalence of HPV 90 infection in an underserved popula- 36. Chaturvedi AK, Myers L, Hammons AF, et al. Prevalence and
tion: is it really a low-risk genotype? Arch Pathol Lab Med. 2013;137: clustering patterns of human papillomavirus genotypes in multiple
1569e1573. infections. Cancer Epidemiol Biomarkers Prev. 2005;14:
31. Fu L, Van Doorslaer K, Chen Z, et al. Degradation of p53 by human 2439e2445.
Alphapapillomavirus E6 proteins shows a stronger correlation with 37. Hopenhayn C, Christian A, Christian WJ, et al. Prevalence of human
phylogeny than oncogenicity. PLoS One. 2010;5. papillomavirus types in invasive cervical cancers from 7 US cancer
32. Cuschieri KS, Cubie HA, Whitley MW, et al. Multiple high risk registries before vaccine introduction. J Low Genit Tract Dis. 2014;
HPV infections are common in cervical neoplasia and young 18:182e189.
women in a cervical screening population. J Clin Pathol. 2004; 38. Katki HA, Kinney WK, Fetterman B, et al. Cervical cancer risk for
57:68e72. women undergoing concurrent testing for human papillomavirus and
33. Vaccarella S, Francheschi S, Snijders PJ, et al. Concurrent infection cervical cytology: a population-based study in routine clinical practice.
with multiple human papillomavirus types: pooled analysis of the Lancet Oncol. 2011;12:663e672.