Gynecologic Oncology 120 (2011) 89–93

Contents lists available at ScienceDirect

Gynecologic Oncology
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o

Quantification of intracellular HPV E6/E7 mRNA expression increases the specificity

and positive predictive value of cervical cancer screening compared to HPV DNA
Gerald Coquillard a, Bibiana Palao b, Bruce K. Patterson a,⁎
a
Department of Pathology and Medicine, Division of Infectious Diseases and Geographic Medicine, Stanford University School of Medicine, Stanford, CA 94304, USA
b
Labec Pharma, Maria de Molina 14, Madrid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Objectives. Current methods for HPV screening rely on the detection of L1 DNA from high risk genotypes
Received 9 July 2010 (HRHPV). These assays have very high negative predictive values (~99%), however, the specificity and
Available online 14 October 2010 positive predictive value of HPV DNA tests for pre-cancerous and cancerous lesions (CIN 2+) is less than 50%.
The purpose of this study was to compare HPV DNA with intracellular HPV E6, E7 mRNA quantification in an
Keywords: effort to improve the performance of cervical cancer screening.
HPV
Methods. Liquid-based cervical cytology specimens collected in either PreservCyt or SurePath were
E6
E7
processed for routing cytology, HPV HRDNA detection by Hybrid Capture 2 and HPV E6, E7 mRNA
mRNA quantification in cells using the same sample. We analyzed a total of 2049 samples including 73 with CIN 2,
HPV OncoTect CIN 3, or squamous cell carcinoma by biopsy and 1694 samples from women with normal cytology.
Hybrid Capture 2 Results. The positive predictive value of HPV E6, E7 mRNA quantification in cells for CIN2+ was 78% which
CIN was greater than HPV DNA alone (43%). The specificity of HPV E6, E7 mRNA quantification was 96% based on
normal cytology compared to 82% for HC2 while the specificity of HPV E6, E7 mRNA quantification based on
CIN 2− histology was 85% compared to 35% for HC2.
Conclusions. With similar sensitivity and greater specificity/positive predictive value, HPV E6, E7 mRNA
quantification in cells is an improvement over HPV DNA for cervical cancer screening.
© 2010 Published by Elsevier Inc.

Introduction Since the overexpression of the HPV-encoded oncogenes E6 and E7

Effective cervical cancer screening programs like those implemented
in the United States have reduced the incidence and the mortality of
cervical cancer [1]. The Papanicolaou smear (Pap) has been used for over
fifty years to screen for cervical cancer and over the last ten years, HPV
DNA testing as a reflex test for ASCUS (atypical squamous cells of
unknown significance) has also been employed [2–5]. Based on current
recommendations, HPV DNA testing is used as an adjunctive test for all
cervical cytology samples (co-testing) in women over 30 years old [6,7].
The Pap smear is limited by low sensitivity for high grade lesions (CIN 2,
CIN 3, or squamous cell carcinoma) while the sensitivity of HPV DNA
testing for high grade lesions is greater than 90% [6–9]. HPV DNA testing
alone, however, has lower specificity for high grade lesions than does
Pap testing [6,9]. Because of the poor specificity of HPV DNA testing, the
positive predictive value of HPV DNA for high grade lesions by biopsy is
15%–25% which can lead to unnecessary colposcopy and biopsy
examinations in women with abnormal Pap smears who are positive
for high risk HPV DNA [10–14].
⁎ Corresponding author. Department of Pathology, Stanford University Medical
School, 3373 Hillview Ave, Palo Alto, CA 94304, USA. Fax: + 1 650 587 1528.
E-mail address: brucep@stanford.edu (B.K. Patterson).
plays a significant albeit not solitary role in cervical carcinogenesis
[15,16], quantification of these oncogenes could improve the specificity
and positive predictive value of cervical cancer screening with the goal
of detecting CIN 2+ lesions with high specificity [17–19].
The combination of the Pap smear and HPV DNA has proven to be a
useful, though a costly and labor intensive combination to effectively
screen for cervical cancer in areas where highly skilled cytotechnolo-
gists, cytopathologists, and virologists can perform this combination of
tests. The majority of the world, where greater than 85% of cervical
cancer occurs, lacks the resources to implement such a rigorous
screening program at the recommended intervals [14]. Therefore in
the present study, we compared a new technology that quantifies the
HPV-encoded genes (E6, E7 mRNA) that cause cervical cancer in
individual cells [17–19]. Upregulation of E6, E7 mRNA in individual cells
is the molecular trigger that begins the cycle of cellular transformation,
p53 and Rb degradation, and cell cycle dysregulation that ultimately
leads to cervical cancer [15,16]. Other molecular tests such as the Aptima
HPV Assay and the HPV Proofer assay detect the presence of E6, E7
mRNA from high risk HPV types but do not quantify the overexpression
of E6, E7 mRNA on a cell-by-cell basis, the hallmark of cellular
transformation [10]. Here, we demonstrate that quantification of E6,
E7 mRNA at the cellular level and quantification of the percentage of
cells expressing elevated levels of HPV E6, E7 mRNA have equivalent

0090-8258/$ – see front matter © 2010 Published by Elsevier Inc.

doi:10.1016/j.ygyno.2010.09.013
90 G. Coquillard et al. / Gynecologic Oncology 120 (2011) 89–93
sensitivity and far superior specificity/PPV than the FDA-approved
hybrid capture 2 assay for CIN2+ lesions detected by histology. Further,
we demonstrate that quantification of cells expressing high levels of E6,
E7 mRNA correlates with the severity of the cervical lesion on biopsy.
Materials and methods
Subjects
Cervical cytology samples from multiple sites including both low-
and high risk populations were used with ethical approval that
consisted of a waiver of consent for pre-existing samples. Cervical
cytology specimens were collected using a cytobrush and preserved
using either PreserCyt (Cytyc, Inc, Boxborough, MA) or SurePath
(TriPath Imaging, Burlington, NC) liquid-based cytology fixatives. The
smears were classified using The Bethesda System 2001 criteria.
Cell lines
HeLa cells were used as a positive cell control for the E6, E7 mRNA
assay. These cells were obtained from the American Type Culture
Collection (ATCC) and grown according to the instructions. Normal
human ectocervical cells (Clonetics, Inc) that were used as negative
control cells for the E6, E7 mRNA assay were grown in media supplied by
the company. In addition, positive and negative HPV control cells
commonly used for HPV DNA controls (Accurun, Seracare Life Sciences,
Milford MA) were used without undergoing lysis and nucleic acid
extraction that is necessary for HPV DNA analysis.
Simultaneous ultrasensitive signal-enhancing hybridization in situ (SUSHI)
A 0.5 mL aliquot was removed from the liquid-based cervical
cytology specimen. The cells were pelleted by centrifugation at 400 ×g
and washed once in phosphate buffered saline (PBS), pH 7.4 with 2%
fetal calf serum. Following washing, the cells were fixed and
permeabilized in IncellFP (InCellDx, Menlo Park CA) for 1 h at ambient
temperature. Following fixation and permeabilization, the cells were
washed once in PBS, pH 7.4, pelleted by centrifugation at 400 ×g,
washed again in 2× SSC and pelleted by centrifugation. SUSHI for E6, E7
mRNA was performed by resuspending the cells in a hybridization mix
and cocktail of signal-enhancing oligonucleotide probes specific for E6,
E7 mRNA from all high risk HPV genotypes (InCellDx, Menlo Park CA).
Hybridization was performed at 43 °C. for 30 min and was followed by
successive stringency washes. The cells were resuspended in PBS, pH 7.4
with 2% fetal calf serum for flow cytometric analysis.
Hybrid capture 2
Hybrid Capture 2 was performed on either ThinPrep preserved liquid
cytology samples or CytoRich preserved liquid cytology samples. Four
mL of residual sample was used for the assay. Nucleic acid extraction
was performed and the rapid capture robot was used to perform the
assay according to the manufacturer's instructions using the high risk
panel of probes (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68).
Flow cytometry
Flow cytometry was performed using forward and orthogonal (side)
light scatter to unequivocally identify the ectocervical population of cells
in the liquid-based cervical cytology sample [20–22]. Single color
analysis for HPV E6, E7 mRNA was performed on a FC500 flow cytometer
(Beckman-Coulter, Fullerton CA) or a dedicated BlueFISH™ Analyzer
(IncellDx, Menlo Park, CA). Analysis was restricted to ectocervical cells
defined by forward and orthogonal light scatter characteristics. The
cursor position defining HPV E6, E7 mRNA negative cells from HPV E6,
E7 mRNA positive cells was determined using negative control cell lines
(InCellDx, Menlo Park CA) and confirmed by running normal cervical
cytology samples fixed in ThinPre or SurePath liquids.
Statistical analysis
Statistical analyses were performed using either a t-test or a Mann–
Whitney Rank Sum test. P values b0.05 were considered statistically
significant.
Results
Liquid-based cytology specimen stability for E6/E7 mRNA quantification
Because stability of RNA is an issue in clinical diagnostic testing, we
determined the stability of HPV E6, E7 mRNA in clinical samples
preserved in LBC preservative over a period of three months.
Commercially available positive and negative control cells were stored
at room temperature in PreservCyt fixative over a period of three
months. The mean fluorescence intensity of the positive control cells
(Seracare, Life Sciences, Milford, MA) and the mean fluorescence
intensity (MFI) of the negative control cells (Seracare Life Sciences,
Milford, MA) were determined following the E6, E7 mRNA assay at
desired time intervals (Table 1, Fig. 1). The signal to noise (SNR) ratio as
determined by the MFI of the positive control cell peak minus the MFI of
the negative control cell peak was assessed at each time interval. In
addition to the control cells, residual samples from five cytologically
normal samples and five samples with ASCUS or LSIL cytology that were
positive for cells overexpressing E6, E7 mRNA were assayed at baseline,
three days, one week, one month and three months. The SNR of the
control cells varied by less than 20% over the course of three months and
the patient samples demonstrated little variability over three months
with no positive samples becoming negative because of the age of
sample (data not shown).
Interrun reproducibility of the in cell E6, E7 mRNA assay
To determine reproducibility, we ran in cell E6, E7 mRNA on thirty
paired 0.5 mL aliquots of the same sample. These samples were then run
on the same instrument using the same settings and analysis protocol.
As shown in Fig. 2, the correlation between the samples from the first
run compared to the samples from the second run yields a r2 = 0.98. Of
the thirty samples, no result would have changed the diagnosis from
positive to negative or from negative to positive for overexpression of
E6, E7 mRNA.
Liquid-based cervical cytology specimen adequacy
Prior to performing HPV E6, E7 analysis, the adequacy of LBCs was
assessed directly out of the Pap vial using the same molecular hybrid
platform used for E6, E7 analysis. As demonstrated previously [20,22],
LBC specimen adequacy as defined by the Bethesda System 2001 was
performed by placing the LBC vial into the BlueFISH™ Analyzer and
determining the number of ectocervical cells, endocervical cells, and
Table 1
Liquid-based cervical cytology specimen stability as determined by signal to noise ratio
(SNR) degradation over time at 25 °C storage using positive and negative control cells
preserved in SurePath liquid.
Time Mean fluorescence intensity (MFI)
Positive control cells Negative control cells
Day 0 11.4 0.7
Day 7 13.1 0.8
Day 14 10.9 0.7
Day 21 12 0.9
Month 1 14.2 1
Month 3 9.8 1.6
 G. Coquillard et al. / Gynecologic Oncology 120 (2011) 89–93 91

Fig. 1. Representative plots of positive and negative control cells demonstrating the expression of E6, E7 mRNA in individual cells. The dual parameter (FSC, SSC) dot plots represent
the light scatter characteristics of the control cells. The histograms represent the expression of E6, E7 mRNA in positive and negative control cells. The overlay of the positive and
negative control cells illustrates the difference in E6, E7 mRNA expression in the negative cells (left peak) and the positive cells (right peak). The signal to noise SNR) ratios (Table 1)
are calculated by dividing the mean fluorescence intensity (MFI, dotted lines) of the positive control peak by the MFI of the negative control peak.

inflammatory cells per microliter of LBC preservative. Using a cut-off of
500 ectocervical cells/microliter which corresponds to N5000 squamous
cells per slide [22], we determined that 2049 samples of 2094 total
samples were adequate (2.2% unsatisfactory samples) for cytologic
evaluation and HPV testing with both HC2 and E6, E7 mRNA (data not
shown).
Comparison of cytology, HC2, and E6, E7 mRNA with histology
Split sample testing was performed on all 2049 samples by removing a
0.5 mL aliquot from liquid-based cervical cytology specimens for HPV E6,
E7 mRNA quantification (Fig. 3) prior to performing Hybrid Capture 2 on
4 mL of the remaining sample. As shown in Table 2, the sensitivity of
Hybrid Capture 2 for CIN 2+ was 89% and the sensitivity of E6, E7 mRNA
quantification was approximately 84%. The sensitivity for both HC 2 and of
E6, E7 mRNA quantification was 91% based on 22 CIN 3 biopsies. Hybrid
Capture 2 was negative in one squamous cell carcinoma that was positive
by E6, E7 mRNA quantification. Hybrid Capture 2 indiscriminately
detected dysplasia of widely variable severity including 82% of CIN 1
lesions (compared to 27% for E6, E7 mRNA quantification) which typically
regress and most importantly do not require treatment. Conversely,
quantification of cells overexpressing E6, E7 mRNA was more specific for
“downstream” lesions CIN 2, CIN 3, and squamous cell carcinoma as
demonstrated in Table 2. The positive predictive value of HPV E6, E7
mRNA quantification in cells for CIN2+ was 78% which was greater than
HPV DNA alone (43%). The specificity of HPV E6, E7 mRNA quantification
was 96% based on normal cytology compared to 78% for HC2 (data not
shown) while the specificity of HPV E6, E7 mRNA quantification based on
CIN 2− histology was 85% compared to 35% for HC2.

Discussion

Currently, the combined use of cervical cytology and HPV DNA

Fig. 2. Interrun reproducibility using split liquid-based cervical cytology samples
demonstrating strong correlation.
testing is the mainstay of cervical cancer screening programs. Studies
comparing the performance of cervical cytology with high risk HPV DNA
detection for the detection of pre-cervical cancer or cervical cancer
lesions (CIN 2+) have shown that the Pap smear lacks sensitivity to
detect all women with pre-cervical cancer or cervical cancer lesions
although the specificity of the Pap smear is greater than 90% [6–9].
Conversely, the sensitivity of high risk HPV DNA is superior to that of
cytology for the detection of pre-cervical cancer and cervical cancer
92 G. Coquillard et al. / Gynecologic Oncology 120 (2011) 89–93

Fig. 3. Representative histograms of liquid-based cervical cytology specimens from women with normal cytology (left column) or from women with biopsies diagnosed as cervical
intraepithelial neoplasia 3 (CIN 3, right column). Women with N 2.0% of cells expressing high levels of E6, E7 mRNA correlated with high grade cervical lesions (CIN 2+).

lesions, however, the specificity of HPV DNA is less than cytology leading value of HPV DNA alone [13]. Similarly, opposition to the use of HPV
to a low positive predictive value for CIN 2+. The performance profile of DNA testing for primary cervical cancer screening has also focused on
these two common tests for cervical cancer screening necessitates their the low specificity and low positive predictive value of HPV DNA for CIN
combined use in order to detect almost all CIN 2+ lesions without 2+ which would send an unacceptable number of women for
sending an inordinate number of women for colposcopy and biopsy unnecessary cervical biopsies [14]. Further the combined use of cervical
because of the relatively low specificity and low positive predictive cytology and HPV DNA is costly and labor intensive; factors magnified in
resource poor settings. Here, we report the performance of a new RNA
Table 2 approach to cervical cancer screening which greatly improves the
Performance comparison of HC2 and E6/E7 mRNA quantification for the detection of specificity and PPV for CIN2+. This assay can be performed on any flow
cervical abnormalities. cytometry platform including those already installed throughout the
Biopsy diagnosis HPV DNA (HC2) Oncotect E6, E7 mRNA Total developed and developing world to monitor CD4 counts in individuals
infected by HIV.
Benign 57 (52%) 12 (11%) 109
CIN 1 64 (82%) 21 (26.9%) 78
Schiffman and colleagues describe the dynamic lifecycle of HPV
CIN 2 38 (88.3%) 36 (83.7%) 43 following infection [12]. Infection of women by high risk genotypes of HPV
CIN 3 20 (91%) 20 (91%) 22 including HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59,and 68 has been
SCC 7 7 (87.5%) 8 (100%) 8 associated with cervical cancer but b5% of these infections result in cancer
Specificity (incl CIN 2−) 35% 85%
if left untreated [13]. The oncogenic potential of these high risk genotypes
 G. Coquillard et al. / Gynecologic Oncology 120 (2011) 89–93
resides in their ability to express the HPV-encoded oncogenes E6 and E7
[15,16]. Previous studies demonstrated E6 transcripts in all CINs (CIN 1,
CIN 2, CIN 3, SCC) [23]. Similarly, E7 expression which may not be
necessary for the development of CIN was suggested to be a prerequisite
for the progression of CIN to cervical carcinoma [24]. We designed the
current assay to quantitatively assess the level of E6, E7 mRNA expression
on a cell-by-cell basis using a high-throughput cell analyzer such as a flow
cytometer [17]. Using standard fluorescent beads, we were able to
determine that abnormal squamous cells express hundreds of copies of
E6, E7 mRNA following transformation [17]. These data are consistent
with PCR data demonstrating the detection of E6/E7 transcripts in 58% of
CIN 1, 77% of CIN 2, and 84% of CIN 3 in one study [23] and 100% of cervical
cancers in another study [24]. In the latter study by Nakagawa et al., E6/E7
expression in cervical cancer was ubiquitous and irrespective of high risk
HPV type. Last, we show that the percentage of cells overexpressing E6, E7
mRNA in a liquid-based cervical cytology samples correlates with the
severity of the lesion. This finding reflects the fact that transforming cells
overexpressing E6, E7 mRNA are distributed differently in pre-cervical
cancer lesions depending on the severity of the lesion. As lesions progress
in severity, E6, E7 mRNA overexpressing cells become distributed
throughout the epithelium rather than in the basal layer as would be
expected in low grade (CIN 1) lesions [15,25,26]. Thus, the likelihood of
detecting these cells in a cervical cytology specimen would necessarily
increase in more severe lesions. Our findings await confirmation in
additional studies and a potential caveat of this technology could exist in
very small lesions where the percentage of E6, E7 mRNA may be
diminished by a large percentage of normal cells. Using this method-
ology, however, these cells would still be identified by the high levels of
E6, E7 mRNA. Additional studies may identify other gating parameters
to ensure detection of small numbers of cells expressing very high levels
of E6, E7 mRNA within a cell.
In summary, we present a novel, quantitative technology for the cell-
by-cell quantification of E6, E7 mRNA expression in liquid-based cervical
cytology. This triage technology increases specificity and positive
predictive value of cervical cancer screening with little if any loss of
sensitivity. This technology also holds the promise of distinguishing
between pre-squamous cell carcinoma and pre-adenocarcinoma based on
distinct ectocervical or endocervical gating on the same sample.
Conflict of interest statement
The authors have received no funding nor have any conflicts of interest regarding this
manuscript.
References
[1] Horner MJ, Ries LAG, Krapcho M. SEER Cancer Statistics Review 1975–2006.
National Cancer Institute; 2009.
[2] Runowicz CD, Garozzo S. Human papillomavirus testing for primary cervical cancer
screening. Curr Oncol Rep 2008;10:533–7.
[3] Wright Jr TC, Lorincz A, Ferris DG, et al. Reflex human papillomavirus
deoxyribonucleic acid testing in women with abnormal Papanicolaou smears.
Am J Obstet Gynecol 1998;178:962–6.
93
[4] Wright TC, Sun XW, Koulos J. Comparison of management algorithms for the
evaluation of women with low-grade cytologic abnormalities. Obstet Gynecol
1995;85:202–10.
[5] Solomon D, Schiffman M, Tarone R. Comparison of three management strategies
for patients with atypical squamous cells of undetermined significance: baseline
results from a randomized trial. J Natl Cancer Inst 2001;93:293–9.
[6] Mayrand MH, Duarte-Franco E, Rodrigues I, et al. Human papillomavirus DNA
versus Papanicolaou screening tests for cervical cancer. N Engl J Med 2007;357:
1579–88.
[7] Cox JT. History of the use of HPV testing in cervical screening and in the management
of abnormal cervical screening results. J Clin Virol 2009;45(Suppl 1):S3–S12.
[8] Nanda K, McCrory DC, Myers ER. Accuracy of the Papanicolaou test in screening for
and follow-up of cervical cytologic abnormalities: a systematic review. Ann Int Med
2000;132:810–9.
[9] Koliopolous G, Arbyn M, Martin-Hirsch P, et al. Diagnostic accuracy of human
papillomavirus testing in primary cervical screening: a systematic review and
meta-analysis of non-randomized studies. Gynecol Oncol 2007;104:232–46.
[10] Szarewski A, Ambroisine L, Cadman L, et al. Comparison of predictors for high-grade
cervical intraepithelial neoplasia in women with abnormal smears. Cancer Epidemiol
Biomarkers Prev 2008;17:3033–42.
[11] Naucler P, Ryd W, Tornberg S, et al. Human papillomavirus and Papanicolaou tests
to screen for cervical cancer. N Engl J Med 2007;357:1589–97.
[12] Schiffman M, Solomon D. Screening and prevention methods for cervical cancer.
JAMA 2009;302:1809–10.
[13] Sherman ME, Lorincz AT, Scott DR, et al. Baseline cytology, human papillomavirus
testing, and risk for cervical neoplasia: a 10-year cohort analysis. J Natl Cancer Inst
2003;95:46–52.
[14] Ronco G, Giorgi-Rossi F, Carozzi F, et al. Results at recruitment from a randomized
controlled trial comparing human papillomavirus testing alone with conventional
cytology as the primary cervical cancer screening test. J Natl Cancer Inst 2008;100:
492–501.
[15] Durst M, Glitz D, Schneider A. zur HH. Human papillomavirus type 16 (HPV 16)
gene expression and DNA replication in cervical neoplasia: analysis by in situ
hybridization. Virology 1992;189:132–40.
[16] Munger K, Phelps WC, Bubb V, Howley PM, Schlegel R. The E6 and E7 genes of the
human papillomavirus type 16 together are necessary and sufficient for
transformation of primary human keratinocytes. J Virol 1989;63:4417–21.
[17] Narimatsu R, Patterson BK. High-throughput cervical cancer screening using
intracellular human papillomavirus E6 and E7 mRNA quantification by flow
cytometry. Am J Clin Pathol 2005;123:716–23.
[18] Silvia P, González SP, Henríquez A, Sainz de la Cuesta R. Ginecología y Obstetricia
Clínica 2008;9:181–6.
[19] Kottaridi C, Tsiodras S, Georgoulakis J. A flow cytometric method for quantification
of HPV E6, E7 mRNA as a novel biomarker for cervical cancer screening. Eurogin
2008, Nice, France.
[20] Grundhoefer D, Patterson BK. Determination of liquid-based cervical cytology
specimen adequacy using cellular light scatter and flow cytometry. Cytometry
2001;46:340–4.
[21] Kottaridi C, Georgoulakis J, Kassanos D, et al. Use of flow cytometry as a quality
control device for liquid-based cervical cytology specimens. Cytometry B Clin
Cytom 2010;78:37–40.
[22] Polina R, Sturgis C, Patterson J, Patterson BK. Rapid, high throughput determina-
tion of cervical cytology specimen adequacy using a capillary-based cytometer.
Cytometry B Clin Cytom 2008;74:133–6.
[23] Sotlar K, Stubner A, Diemer D, et al. Detection of high-risk human papillomavirus
E6 and E7 oncogene transcripts in cervical scrapes by nested RT-polymerase chain
reaction. J Med Virol 2004;74:107–16.
[24] Nakagawa S, Yoshikawa H, Yasugi T, et al. Ubiquitous presence of E6 and E7
transcripts in human papillomavirus-positive cervical carcinomas regardless of its
type. J Med Virol 2000;62:251–8.
[25] Sun Q, Tang SC, Pater MM, Pater A. Different HPV16 E6/E7 oncogene expression
patterns in epithelia reconstructed from HPV16-immortalized human endocervical
cells and genital keratinocytes. Oncogene 1997;15:2399–408.
[26] Stoler MH, Rhodes CR, Whitbeck A, Wolinsky SM, Chow LT, Broker TR. Human
papillomavirus type 16 and 18 gene expression in cervical neoplasias. Hum Pathol
1992;23:117–28.