spiral.imperial.ac.uk · web viewquantitative measurement of l1 hpv16 methylation for the...
Post on 09-Aug-2020
4 Views
Preview:
TRANSCRIPT
Quantitative measurement of L1 HPV16 methylation for the prediction of pre-
invasive and invasive cervical disease
Christine Kottaridi1, Maria Kyrgiou2,3, Abraham Pouliakis1, Maria Magkana1, Evangelia
Aga1, Aris Spathis1, Anita Mitra2,3, George Makris4, Charalampos Chrelias4, Vassiliki
Mpakou5, Evangelos Paraskevaidis6, John G. Panayiotides7, Petros Karakitsos1
1Department of Cytopathology, National and Kapodistrian University of Athens,
Medical School, “ATTIKON” University Hospital, 1 Rimini, Haidari, 12462, Athens,
Greece.
2Department of Surgery and Cancer, IRDB, Faculty of Medicine, Imperial College,
London W12 0NN, UK
3West London Gynaecological Cancer Center, Queen Charlotte’s and Chelsea,
Hammersmith Hospital, Imperial Healthcare NHS Trust
43rd Department of Obstetrics and Gynecology, “ATTIKON” University Hospital, 1
Rimini, Haidari, 12462, Athens, Greece
5Second Dept. of Internal Medicine and Research Institute, University of Athens,
Medical School, “ATTIKON” University Hospital, 1 Rimini, Haidari, 12462, Athens,
Greece
6Department of Obstetrics and Gynecology, University Hospital of Ioannina, 45500
Ioannina, Greece
72nd Department of Pathology, University of Athens, Medical School, “ATTIKON”
University Hospital, 1 Rimini, Haidari, 12462, Athens, Greece
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Running title: HPV16 L1 methylation as a biomarker
Word count (excluding references): 3497, Abstract word count: 197
Number of figures: 4, Number of tables: 2
Supplementary data: 2 tables, 5 figures
Declaration of competing interests: The authors report no conflict of interests. All
the authors declare: no support from any organization for the submitted work; no
financial relationships with any organizations that might have an interest in the
submitted work in the previous three years; no other relationships or activities that
could appear to have influenced the submitted work.
Ethical approval: National Research Ethics Service Committee London – Fulham
(Approval number 13/LO/0126). Bioethics committee of “ATTIKON” University
Hospital (Approval number 5/14-06-2013)
Role of the funding source: This work was supported by the British Society of
Colposcopy Cervical Pathology Jordan/Singer Award (P47773)(MK); the Imperial
College Healthcare Charity (P47907) (AM, MK); Genesis Research Trust (P55549)
(MK); and the Imperial Healthcare NHS Trust NIHR Biomedical Research Centre
(P45272) (MK). None of the funders have had any influence on the study design; in
the collection, analysis, and interpretation of data; in the writing of the report; and in
the decision to submit the article for publication. This study was partially funded by
the Greek Ministry of Development (General Secretariat for Research and
2
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Technology-GSRT) and European Union. Project acronym: HPV-Guard (Cooperation
2011-2013, code: 11ΣΥΝ_10_250 http://HPVGuard.org).
Corresponding author:
Dr Maria Kyrgiou, MSc, PhD, MRCOG, 3rd Floor, Institute of Reproductive and
Developmental Biology, Department of Surgery and Cancer, Imperial College,
Hammersmith Campus, Du Cane Road, W12 0NN, London, UK. Tel: +44 (0)20 7594
2177; Email: m.kyrgiou@imperial.ac.uk
3
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
ABSTRACT
Background: Methylation of the HPV DNA has been proposed as a novel biomarker.
Here, we correlated the mean methylation level of 12 CpG sites within L1 gene, to
the histological grade of cervical precancer and cancer. We assessed whether HPV L1
gene methylation can predict the presence of high-grade disease at histology in
women testing positive for HPV 16 genotype.
Methods: Pyrosequencing was used for DNA methylation quantification and 145
women were recruited.
Results: We found that the L1 HPV16 mean methylation (+/-SD) significantly
increased with disease severity [CIN3=17.9%(±7.2) vs CIN2=11.6%(±6.5), p<0.001 or
vs CIN1 =9.0%(±3.5), p<0.001). Mean methylation was a good predictor of CIN3+
cases; the Area Under the Curve (AUC) was higher for sites 5611 in the prediction of
CIN2+ and higher for position 7145 for CIN3+. The evaluation of different
methylation thresholds for the prediction of CIN3+, showed that the optimal balance
of sensitivity and specificity (75.7% and 77.5%, respectively), PPV and NPV (74.7%
and 78.5%, respectively) was achieved for a methylation of 14.0% with overall
accuracy of 76.7%.
Conclusion: Elevated methylation level is associated with increased disease severity
and has good ability to discriminate HPV16 positive women that have high-grade
disease or worse.
Keywords: cervical intraepithelial neoplasia; CIN; HPV L1 gene methylation;
pyrosequencing; Human Papillomavirus
4
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
INTRODUCTION
The introduction of systematic call and recall screening programmes has resulted in a
profound decrease in the incidence and mortality from invasive cervical cancer as
pre-invasive lesions (cervical intra-epithelial neoplasia;CIN) can be treated
appropriately [1, 2]. Despite the high efficacy of screening in preventing cervical
cancer, this has traditionally relied on cytology that is known to have limitations.
Establishing that persistent infection with high-risk human papilloma-virus (HPV) is
causally associated with cervical cancer has led to major advances in primary and
secondary prevention. It is now recognized that the use of the HPV DNA test in
primary screening is likely to offer 60-70% greater protection against invasive cancer
compared with cytology-based screening [3], while the best policy in further triaging
women with positive results at HPV-based screening is unclear.
New biomarkers, exploring the viral genome and life cycle together with its
interactions with the host, have the potential to permit a more comprehensive
understanding of the disease process. These tests may enable a more accurate
detection of clinically significant pre-invasive lesions and a more personalised
identification and management of lesions with true progressive carcinogenic
potential [4-6]. Some target genes in the HPV genome play pivotal roles in the viral
life cycle and ongogenesis [7, 8]. Methylation of the viral DNA has been recently
proposed as a novel biomarker with encouraging results. The quantification of the
percentage of cytosines with a covalently added methyl-group at individual CpG
dinucleotides reflects the degree of epigenetic changes of the viral genome.
5
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
Research in the quantification of methylation at different CpG sites and genes
produced contradictory results, while the methylation threshold varied across
studies [9-11]. Although two small studies suggested that increased methylation was
associated with high-grade intra-epithelial lesions (HSIL) [9, 11], others examining
CpG sites within the upstream regulatory region (URR) documented that it is the lack
of methylation that associates with disease progression [12, 13]. Two further studies
on the Guanacaste Costa Rica cohort reported that high methylation at L1, L2, E2-4
CpG sites increased the risk of CIN3 by 50 times when compared to low methylation
(Odds Ratio(OR)=52, 95%CI:4.0–670). Increased methylation levels predicted both
the presence of HSIL in women with positive HPV16 infections as well as the risk for
development of future CIN2+ [10, 14].
This prospective study aims to assess whether pyrosequencing of the HPV16 L1 gene
and quantification of the methylation level of specific CpGs could predict the
presence of high-grade disease at histology in women testing positive for HPV16
genotype. More specifically, we aim to correlate the mean methylation level to the
histological grade and to determine its accuracy in predicting the disease severity by
establishing optimum methylation cut-offs.
METHODS
Study population
We conducted a multicentric prospective study that recruited in three University
Hospitals (Ioannina/Athens, Greece and Imperial College London, UK). Ethical
6
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
approval was obtained and all patients gave consent (Bioethics Committee, Greece
and National Research Ethics Service Committee London–Fulham (13/LO/0126).
We included non-pregnant women, 21–67 years of age who attended the
gynaecology clinics (May 2013 to 2015). All cervical samples that tested positive for
HPV16 DNA typing and had available histology (punch biopsy or cone - gold
standard) were included. If histology was available from both biopsies and cones, the
most severe lesion was used. The histological diagnoses included the following
groups: normal, CIN1, CIN2, CIN3, squamous cell(SCC) and adenocarcinoma (Adeno-
ca).
Women were included irrespective of their ethnicity and smoking habits.. Women
who were HIV or hepatitis B/C positive, with autoimmune disorders, or had a
previous history of cone were excluded..
Sample collection and methods
We prospectively collected patient characteristics and recorded the cytological,
colposcopic and histological findings. We obtained a liquid-based cytology sample
(LBC, ThinPrep® Pap Test, Hologic) that was prepared on a TP2000 Processor and was
reported by trained cytopathologists according to the Bethesda 2001 system
(TBS2001) [15]. The residual LBC sample was aliqouted and stored at 4°C until
further use.
7
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
The HPV DNA typing was performed with CLART® HPV2 kit (Genomica, Spain) which
is validated to genotype 20 HR HPV types (16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53,
56, 58, 59, 66, 68, 70, 73, 82 and 85) and 15 LR HPV types (6, 11, 40, 42, 43, 44, 54,
61, 62, 71, 72, 81, 83, 84 and 89). The extracted DNA concentration’s were measured
with Quant-iT™ PicoGreen® dsDNA Assay Kit (Thermo Fisher Scientific Inc., Waltham,
MA, USA).
DNAs were then bisulphite converted using the EpiTect Bisulfite Kit (Qiagen, Hilden,
Germany), according to the manufacturer’s instructions and stored at -80°C. Biotin-
labelled primer sets and PCR conditions [14] were used as previously described to
amplify HPV16 L1 region. DNA from the cervical cancer cell line SiHa that contains
integrated HPV16 genome [16] was extracted and bisulfite converted. The
methylation quantification was performed by Pyrosequencing technology (PyroMark
Q24 Qiagen, Hilden, Germany), which provides a site-specific quantification of
methylation at individual CpG sites. The analysis performed in the present study
included 12 CpGs mapped in the HPV16 L1 ORF and specifically 5611, 5726, 5927,
6367, 6389, 6457, 6581, 6650, 6796, 7091, 7136, 7145 of clinical samples’ and SiHa
DNA. The pyrograms were analyzed using the CpG mode of the PyroMark Q24
software, to determine the percentage methylation at each site as well as the overall
mean methylation.
Statistical Analysis
For each included sample, we calculated the mean methylation level (i.e. the mean
value of the measured methylation percentages) by dividing the sum of the
8
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
percentage methylation levels for all positions (P01M-P12M) by the number of
positions (i.e.12). We produced box and whisker plots of the mean methylation, the
standard deviation (SD) and range for each histological group. Each histological
category was given a number starting from 1 (NEGATIVE) and ending to 6 (SCC). We
performed a regression analysis [17] to identify and describe the linear mathematical
equation governing the correlation between mean methylation percentage and
histological grade that includes the intercept and slope of the line. We used t-test to
determine significance. This mathematical formula was developed with the aim to
predict the histological grade from the mean methylation measured. We further
produced a fit plot of the mean methylation and the histological grades to describe
the linear model, the confidence limits of the line and the limits of the prediction
accuracy. We used unadjusted and adjusted R-square to describe the performance of
the regression analysis and the fit of the line to the data with a value near to 100%
describing the perfect fit. We calculated the modified adjusted R-squared that
adjusts for several predictors of the model and has been shown to be a better
performance metric [17]. These analyses were performed for each of the 12 CpG
sites. We further performed a linear regression analysis to explore whether the
mean methylation level is affected by age.
We further assessed the discriminative capability for each of the twelve methylation
positions (P01M–P12M) in predicting the disease at four different histological cut-
offs (CIN1+, CIN2+, CIN3+, Cancer). We produced Receiver Operating Characteristics
(ROC) curves [18] and calculated the area under the curve (AUC), the standard error
(SE) and the 95% confidence intervals (CI). The largest AUC (near to 100%) indicated
9
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
methylation positions that had the highest prediction of the histological outcome for
the chosen cut-off.
We calculated different accuracy parameters for the ability of the mean methylation
to detect the presence of disease for 4 histological cut-offs: CIN1+, CIN2+, CIN3+ or
cancer. These included the sensitivity (S), specificity (Sp), positive (PPV) and negative
predictive value (NPV), the false positive (FP) and false negative (FN) rate, the overall
accuracy (OA) and the positive (PLR) and negative likelihood ratio (NLR). We applied
various threshold values starting from 0% and increasing up to 100% using an
increment step of 0.1%, as described in our previous studies [19-21]. We further
described the accuracy parameters for different methylation thresholds at each
histological cut-offs, and calculated the mean methylation thresholds that optimize
the overall accuracy and/or the balance between sensitivity and specificity. We
calculated the odds ratio (OR) to describe the likelihood of the histological diagnosis
if the mean methylation was above or below a chosen threshold. We subsequently
tested the produced algorithm in a small validation set to confirm reproducibility of
the results.
We used SAS9.4 for the statistical analysis (SAS Institute Inc. NC, USA) [23, 24] and
the algorithms for the determination of the optimum threshold values were
developed in-house within the MATLAB software environment and programming
language (The MathWorks, Inc. Natick, Massachusetts, U.S.A.).
RESULTS
10
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
We identified a total of 151 women that tested positive for HPV16 type and had
histology; 1 was excluded as the histology revealed VAIN2 (Table 1). The mean age
differed amongst histological groups; women with invasive disease tended to be
older (P-value<0.0001). All women with normal cytology had normal findings at
histology. None of the women presented with ASCUS cytology had HSIL in histology,
although one in four women presenting with LSIL (12/20, 60%) had CIN2+ at
histology. One in ten women (11/105, 10.4%) that presented with HSIL cytology had
CIN1 or normal histology. All cases of invasive cancer presented with HSIL or possible
invasion on cytology. Approximately half of the population (46.7%) had HPV16 only
at genotyping, while the remaining women were also infected with other high-risk
(44.6) or low-risk (8.7%) subtypes.
We found that the mean L1 HPV16 methylation increased with increasing disease
severity. The mean methylation (SD, range) was 8.1 (3.7, 5.1-15.7) for normal, 9.0
(3.4, 4.5-15.8) for CIN1, 11.6 (6.5, 4.2-34.5) for CIN2, 17.9 (7.2, 6-41.5) for CIN3, 38.1
(10.8, 27-48.6) for adenocarcinoma and 58.1 (17.3, 42.5-86.2) for SCC (Figure 1A).
The HPV16 L1 mean methylation levels (SD) of CIN3+ cases (17.9%±7.2%) were
significantly higher compared to CIN2 (11.6%±6.5%) or CIN1 (9.0%±3.5%) histological
groups (t-test for CIN3+ vs CIN2: t=-4.6, p<0.001; CIN3+ vs CIN1: t=-8.9, p<0.001).
The regression analysis produced a formula to predict the correlation between the
mean methylation and the histological grade (Mean Methylation=-
7.65+7.24*Histological grade (expressed as number: starting from 1 (NEGATIVE) to 6
(SCC))(Figure 1B). The fitted line showed no tolerance as the intercept (-7.65) and
slope (7.24) was statistically significant (t-test:p<0.05 for both). The positive slope of
11
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
this line (7.65) confirmed that the mean methylation increases with increasing
disease grade. Similar trends were found for the disease severity and the mean
methylation in each of the individual CpG sites (results not shown). We found that
for the 3 subgroups of women with CIN1, 2 and 3, there was no evidence that the
mean methylation level was affected by the age (adjusted R-square: -0.047, 0.036
and -0.016, respectively). There was a trend towards lower methylation levels in
older women with CIN2, but the results were not statistically significant.
We quantified the methylation of CpG sites 5611, 5726, 5927, 6367, 6389, 6457,
6581, 6650, 6796, 7091, 7136 and 7145 in the HPV16 L1 gene. We constructed ROC
curves to assess the diagnostic utility of the methylation levels of each individual CpG
site (assigned P01M-P12M, respectively) and of the mean methylation in predicting
the histological diagnosis for the different cut-offs described (CIN1+, CIN2+, CIN3+,
Cancer)(Figure 2A and B, Supplementary Figures 1-5, Supplementary Table 1). We
found the ability to predict the histological diagnosis increased with increased cut-off
used and this was consistent for different methylation positions. The AUCs for CIN2+
lesions ranged from 0.66 to 0.82. The lowest AUC value (AUC:0.66,95%CI%:0.57-
0.76) observed in site: 6367, the highest (AUC:0.82,95%CI:0.75-0.90) observed in site
5611 for CIN2+ lesions, while the mean methylation for CIN2+ was 0.81 (95%CI:0.74-
0.88) (Figure 2A, Supplementary Figure 3). For CIN3+ lesions, the AUCs ranged from
0.71 to 0.86 with the highest value assigned to site 7145 (AUC:0.86,,95%CI:0.80-
0.92), the lowest to site 5726 (AUC:0.7195%CI:0.73-0.87) with the one for the mean
methylation at 0.85 (95%CI:0.79-0.91)(Figure 2B, Supplementary Figure 4). The
highest performance was for the histological cut-off of cancer as expected (mean
12
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
methylation AUC:0.99,95%CI:0.98-1.00), with limited though clinical use
(Supplementary Figure 1, 5).
We further determined accuracy parameters for different histological cut-offs and
described the mean methylation thresholds that optimized overall accuracy and/or
the balance between sensitivity and specificity (Table 2, Figure 3, Supplementary
Figure 6). We found that for a histological cut-off of CIN2+ (Figure 3A), a mean
methylation of 5.6% maximised the OA (82.7%) and achieved optimal sensitivity
(97.4%) at the loss of specificity (34.3%). We therefore applied a different mean
methylation threshold (10.8%) that optimized the balance between sensitivity
(67.8%) and specificity (74.3%) with an overall accuracy of 69.3%. For the prediction
of CIN3+, a methylation level of 15.9% optimized the overall accuracy 79.3%,
although the sensitivity dropped to 67.1% with a specificity being high at 90.0%
(Figure 3B). The optimal balance of sensitivity and specificity (75.7% and 77.5%,
respectively), PPV and NPV (74.7% and 78.5%, respectively) was achieved for a
methylation of 14.0% and OA of 76.7%. Higher methylation levels were used for the
detection of cancer cases, where a threshold of 37.9% achieved an OA of 98.6% with
a sensitivity of 88.9% and a specificity of 99.3%. The use of a mean methylation
threshold of 27.0% optimized sensitivity (100%) at a small loss in specificity (94.3%)
and OA (94.7%). With the exception of the histological cut-off of invasive cancer, the
odds ratio was higher (OR:18.39, 95%CI:7.59-44.54) for a mean methylation
threshold of 15.9% for the prediction of CIN3+ (ie. odds for CIN3+ are 18.4 times
higher if the mean methylation is above as opposed to less than 15.9%).
13
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
We evaluated the algorithm for a validation set of 22 (normal:5; CIN1:3; CIN2:4;
CIN3:8 and SCC:2)(data not shown). The results were consistent with the training set,
although the number of patients was small. For all thresholds, the sensitivity was
excellent, although the specificity varied. For the cut-off of CIN2+, the sensitivity was
100% and the specificity was 75.0%, compared to 67.8% and 74.3% in the training set (best
balance at 10.8%). All 14 CIN2+ cases were correctly assigned, while 2 out of 8 normal/CIN1
cases were misclassified as CIN2+. For the CIN3+ cut-off (maximum accuracy threshold at
15.9%) the sensitivity was 100% and the specificity 66.67% (training sets 67.1% and 90.0%,
respectively). The mean methylation enabled the correct classification of the 10 CIN3+ cases,
while misclassified as high-grade 4 out of 12 CIN2- cases.
DISCUSSION
Main findings and Interpretation in light of other evidence
The knowledge that oncogenic HPV types are causally associated with invasive
cervical cancer has initiated research into viral genome’s alterations during the viral
lifecycle and their interactions with the host genome [25-27]. The identification of
women screened that have CIN2+ has been one of the major challenges of concerted
efforts over the last decade, particularly as we are moving towards the use of HPV
test at primary screening. The second major challenge was to identify biomarkers
and molecular determinants that could distinguish the rare HPV infections that have
a true oncogenic malignant potential from those common infections from the same
types that resolve spontaneously without leading to disease.
14
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
Our study suggests that quantification of the methylation levels of 12 CpG sites in
the L1 gene with pyrosequencing has the potential to serve as a molecular marker in
the prediction of the severity of cervical disease. We studied 150 HPV16 positive
cases with histologically confirmed diagnosis. We found that mean methylation level
increased with increasing disease severity, while the methylation patterns of each
CpG studied did not differ substantially, as previously reported [14]. The mean
methylation levels for CIN3+ cases were significantly higher than CIN1 or CIN2 cases,
in agreement with previously reported results [14, 28]. Furthermore, we calculated
accuracy parameters for various methylation thresholds and defined those that
would optimize the diagnostic accuracy of the test.
Various methods of evaluation of viral DNA methylation have been published [9-11,
14, 28-31]. Some researchers used restriction endonucleases in non-bisulfite based
studies [13, 32, 33], whilst others employed bisulfite-based assays where cytosine
residues from single-stranded DNA are deaminated by sodium bisulfate and
converted to uracils with the exception of 5-methyl cytosines that are protected
from conversion [34-36]. In the present study we used bisulfite-treated DNA and the
methylation patterns were analysed by pyrosequencing which is a method with
reproducible and accurate measures of the degree of methylation at several CpGs in
the same amplicon [37].
There have been considerable efforts to study epigenetic factors affecting viral genes
and their associations to cervical cancer and precancer. Researchers have studied
CpG sites along HPV genes L1, L2, E2-E4, E5 and URR [9-11, 14, 29, 30, 38]. The
15
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
results of methylation status vary between studies. Cross-sectional studies in
diagnostic samples have correlated elevated methylation levels of various genes such
as L1, L2 and E2-E4 to the disease severity, while those from enhancer and promoter
region of the URR region being less reliable, inconsistent and heterogenous often
indicating higher methylation frequencies in normal as opposed to HSIL or invasive
samples [28].
Mirabello et al. [14] demonstrated that methylation of several CpG sites within L1, L2
and E2-E4 predicted the presence of CIN2 or worse in women with HPV16 infection
as well as the risk for future HSIL disease in women testing positive for HPV16. The
association between increased DNA methylation in the L1 regions and CIN3+ was
also confirmed by Sun et al. [30]. Another study that explored 13 CpG sites within
the L1 gene reported that higher methylation levels at certain L1 sites was associated
with HPV persistence and cervical precancerous progression [38]. Bryant et al. [39],
however, stated that although HPV DNA methylation may be a promising biomarker
in the triage of HPV-positive cytology samples, its value was less pronounced in
young women. Our results did not confirm a correlation between HPV methylation
and age, although the analysis was limited due to the small sample size in each
group.
We quantified the methylation of 12 CpGs within L1 gene and the highest accuracy
was found for sites 5611, for the prediction of CIN2+, and for site 7145 for the
prediction of CIN3+, which is consistent with other reports. Louvanto et al. [40]
reported high accuracy for sites CpG 6367 and 6389 of the L1 HPV16 region in the
16
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
prediction of CIN2+ (AUC>0.7). Mirabello et al. [14] reported an AUC=0.82 for CpG
6457, while for Niyazi et al. [38] the position 6650 appeared to be the most robust
CpG site (AUC>0.82). In our study, the estimation of the mean methylation from
12CpG sites did not achieve better accuracy in the prediction of the disease status
when compared to loci 5611 and 7145, suggesting that the selective use of few sites
that optimize the accuracy consistently across different studies may be the most
cost-effective option for the future. The reasons why different L1 sites performed
better for CIN2+ and CIN3+ are unclear, although this may be due to the modest
sample size.
Strengths and limitations
This is a large cohort of patients assessing the methylation levels in the L1 gene in
women with histologically confirmed diagnosis of the disease status that also
provides data on the methylation thresholds that maximize accuracy of the
molecular marker. All samples were analysed in a single laboratory to minimize bias.
The findings add valuable information towards the better understanding of the
biological behavior of HPV16 and the ‘cross-talk’ between the viral and host genome
leading to increased methylated-CpG content in the L1 gene and disease
progression. Although this was a sizeable cohort of patients, there were only a
limited number of samples for healthy controls and for some of the histological
groups making interpretation of the results for these groups difficult. Correlation of
methylation levels to age was limited due to the limited number of samples in the
different histological subgroups. We tested the algorithm in a small validation
cohort. Although the results have shown similar trends as the training set, the
17
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
number of patients was small and we were not able to assess whether there were
statistically significant differences. Future studies should assess the reproducibility of
these results in large validation sets. Future studies should also analyze serial
samples from larger cohorts to further assess the value of methylation as a
predictive and diagnostic molecular determinant. The accuracy of HPV methylation
should be also compared to that of other currently used tests and biomarkers such
as cytology and HPV mRNA.
Conclusions
Elevated methylation levels within the L1 region of the viral DNA are associated with
increased disease severity. This molecular determinant may have a role in the triage
of screen-detected HPV positive women that warrant colposcopic investigation, but
has the potential to further distinguish the infections and pre-invasive lesions most
likely to progress allowing more advanced prognostic risk stratification. This will
allow the identification of women that would benefit from colposcopic assessment
with or without treatment with a reduction in the unnecessary visits and the
reproductive morbidity associated with interventions and treatment [41-45]. Further
serial samples from well-established biobanks should assess the impact of the viral
methylation levels to the infection/disease progression at various stages of the
natural history of the disease. Studies should assess its diagnostic accuracy for
different clinical groups and applications.
18
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
Acknowledgements: We thank all the participants of the study.
Authors’ contribution: The study was conceived and designed by CK, AP, MK and PK.
The samples and data was acquired and collated by MK, AM, CC, GM, VM, EP and
analyzed by CK, AP, MM, EA, AS, GJP and MK. The manuscript was drafted and
revised critically for important intellectual content by all authors (CK, MK, AP, MM,
EA, AS, AM, VM, EP, JGP, PK). CK and MK are joint first authors. All authors gave final
approval of the version to be published and have contributed to the manuscript.
19
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
References
1. Saslow D, Solomon D, Lawson HW, et al. American Cancer Society, American Society for
Colposcopy and Cervical Pathology, and American Society for Clinical Pathology screening
guidelines for the prevention and early detection of cervical cancer. Am J Clin Pathol 2012;
137:516-42.
2. Quinn M, Babb P, Jones J, Allen E. Effect of screening on incidence of and mortality from
cancer of cervix in England: evaluation based on routinely collected statistics. BMJ 1999;
318:904-8.
3. Koliopoulos G, Sotiriadis A, Kyrgiou M, Martin-Hirsch P, Makrydimas G, Paraskevaidis E.
Conservative surgical methods for FIGO stage IA2 squamous cervical carcinoma and their
role in preserving women's fertility. Gynecol Oncol 2004; 93:469-73.
4. Kyrgiou M, Pouliakis A, Panayiotides JG, et al. Personalised management of women with
cervical abnormalities using a clinical decision support scoring system. Gynecol Oncol 2016;
141:29-35.
5. Nasioutziki M, Daniilidis A, Dinas K, et al. The evaluation of p16INK4a
immunoexpression/immunostaining and human papillomavirus DNA test in cervical liquid-
based cytological samples. Int J Gynecol Cancer 2011; 21:79-85.
6. Tsoumpou I, Valasoulis G, Founta C, et al. High-risk human papillomavirus DNA test and
p16(INK4a) in the triage of LSIL: a prospective diagnostic study. Gynecol Oncol 2011; 121:49-
53.
7. Gustinucci D, Giorgi Rossi P, Cesarini E, et al. Use of Cytology, E6/E7 mRNA, and p16INK4a-
Ki-67 to Define the Management of Human Papillomavirus (HPV)-Positive Women in Cervical
Cancer Screening. Am J Clin Pathol 2016; 145:35-45.
8. Gage JC, Hunt WC, Schiffman M, et al. Risk Stratification Using Human Papillomavirus
Testing among Women with Equivocally Abnormal Cytology: Results from a State-Wide
Surveillance Program. Cancer Epidemiol Biomarkers Prev 2016; 25:36-42.
20
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
9. Fernandez AF, Rosales C, Lopez-Nieva P, et al. The dynamic DNA methylomes of double-
stranded DNA viruses associated with human cancer. Genome Res 2009; 19:438-51.
10. Mirabello L, Sun C, Ghosh A, et al. Methylation of human papillomavirus type 16 genome
and risk of cervical precancer in a Costa Rican population. Journal of the National Cancer
Institute 2012; 104:556-65.
11. Brandsma JL, Sun Y, Lizardi PM, et al. Distinct human papillomavirus type 16 methylomes
in cervical cells at different stages of premalignancy. Virology 2009; 389:100-7.
12. Piyathilake CJ, Macaluso M, Alvarez RD, et al. A higher degree of methylation of the HPV
16 E6 gene is associated with a lower likelihood of being diagnosed with cervical
intraepithelial neoplasia. Cancer 2011; 117:957-63.
13. Hublarova P, Hrstka R, Rotterova P, et al. Prediction of human papillomavirus 16 e6 gene
expression and cervical intraepithelial neoplasia progression by methylation status. Int J
Gynecol Cancer 2009; 19:321-5.
14. Mirabello L, Schiffman M, Ghosh A, et al. Elevated methylation of HPV16 DNA is
associated with the development of high grade cervical intraepithelial neoplasia.
International journal of cancer Journal international du cancer 2013; 132:1412-22.
15. Solomon D, Davey D, Kurman R, et al. The 2001 Bethesda System: terminology for
reporting results of cervical cytology. JAMA 2002; 287:2114-9.
16. el Awady MK, Kaplan JB, O'Brien SJ, Burk RD. Molecular analysis of integrated human
papillomavirus 16 sequences in the cervical cancer cell line SiHa. Virology 1987; 159:389-98.
17. Huang N. Paper 1760-2014 Scenarios Where Utilizing a Spline Model in Developing a
Regression Model Is Appropriate. Available at:
http://support.sas.com/resources/papers/proceedings14/1760-2014.pdf. Accessed 28 June
2016.
21
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
18. Shin S, Coulter B. Paper SP09-2009 ROC analysis for the evaluation of continuous
biomarkers: Existing tools and new features in SAS® 9.2 Available at:
http://pharmasug.org/download/papers/SP09.pdf. Accessed 28 June 2016.
19. Varlatzidou A, Pouliakis A, Stamataki M, et al. Cascaded learning vector quantizer neural
networks for the discrimination of thyroid lesions. Analytical and quantitative cytology and
histology / the International Academy of Cytology [and] American Society of Cytology 2011;
33:323-34.
20. Pouliakis A, Margari C, Margari N, et al. Using classification and regression trees, liquid-
based cytology and nuclear morphometry for the discrimination of endometrial lesions.
Diagn Cytopathol 2014; 42:582-91.
21. Zygouris D, Pouliakis A, Margari N, et al. Classification of endometrial lesions by nuclear
morphometry features extracted from liquid-based cytology samples: a system based on
logistic regression model. Anal Quant Cytopathol Histpathol 2014; 36:189-98.
22. Pouliakis A, Margari C, Margari N, et al. Using classification and regression trees, liquid-
based cytology and nuclear morphometry for the discrimination of endometrial lesions.
Diagnostic cytopathology 2013:In press (early view available).
23. DiMaggio C. SAS for epidemiologists : applications and methods. New York: Springer,
2013.
24. SAS Institute. SAS Home Page. Available at: http://www.sas.com.
25. Xu L, Verdoodt F, Wentzensen N, Bergeron C, Arbyn M. Triage of ASC-H: A meta-analysis
of the accuracy of high-risk HPV testing and other markers to detect cervical precancer.
Cancer Cytopathol 2016; 124:261-72.
26. Arbyn M, Roelens J, Simoens C, et al. Human papillomavirus testing versus repeat
cytology for triage of minor cytological cervical lesions. Cochrane Database Syst Rev 2013;
3:CD008054.
22
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
27. Lynge E, Antilla A, Arbyn M, Segnan N, Ronco G. What's next? Perspectives and future
needs of cervical screening in Europe in the era of molecular testing and vaccination. Eur J
Cancer 2009; 45:2714-21.
28. Clarke MA, Wentzensen N, Mirabello L, et al. Human papillomavirus DNA methylation as
a potential biomarker for cervical cancer. Cancer Epidemiol Biomarkers Prev 2012; 21:2125-
37.
29. Kalantari M, Garcia-Carranca A, Morales-Vazquez CD, et al. Laser capture microdissection
of cervical human papillomavirus infections: copy number of the virus in cancerous and
normal tissue and heterogeneous DNA methylation. Virology 2009; 390:261-7.
30. Sun C, Reimers LL, Burk RD. Methylation of HPV16 genome CpG sites is associated with
cervix precancer and cancer. Gynecol Oncol 2011; 121:59-63.
31. Sun C, McAndrew T, Smith BC, Chen Z, Frimer M, Burk RD. Characterization of HPV DNA
methylation of contiguous CpG sites by bisulfite treatment and massively parallel
sequencing-the FRAGMENT approach. Front Genet 2014; 5:150.
32. Badal V, Chuang LS, Tan EH, et al. CpG methylation of human papillomavirus type 16 DNA
in cervical cancer cell lines and in clinical specimens: genomic hypomethylation correlates
with carcinogenic progression. J Virol 2003; 77:6227-34.
33. Badal S, Badal V, Calleja-Macias IE, et al. The human papillomavirus-18 genome is
efficiently targeted by cellular DNA methylation. Virology 2004; 324:483-92.
34. Eads CA, Danenberg KD, Kawakami K, et al. MethyLight: a high-throughput assay to
measure DNA methylation. Nucleic Acids Res 2000; 28:E32.
35. Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. Methylation-specific PCR: a novel
PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A 1996; 93:9821-6.
36. Fraga MF, Esteller M. DNA methylation: a profile of methods and applications.
Biotechniques 2002; 33:632, 4, 6-49.
37. Tost J, Gut IG. DNA methylation analysis by pyrosequencing. Nat Protoc 2007; 2:2265-75.
23
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
38. Niyazi M, Sui S, Zhu K, Wang L, Jiao Z, Lu P. Correlation between Methylation of Human
Papillomavirus-16 L1 Gene and Cervical Carcinoma in Uyghur Women. Gynecol Obstet Invest
2016.
39. Bryant D, Hibbitts S, Almonte M, Tristram A, Fiander A, Powell N. Human papillomavirus
type 16 L1/L2 DNA methylation shows weak association with cervical disease grade in young
women. J Clin Virol 2015; 66:66-71.
40. Louvanto K, Franco EL, Ramanakumar AV, et al. Methylation of viral and host genes and
severity of cervical lesions associated with human papillomavirus type 16. Int J Cancer 2015;
136:E638-45.
41. Kyrgiou M, Athanasiou A, Paraskevaidi M, et al. Adverse obstetric outcomes after local
treatment for cervical preinvasive and early invasive disease according to cone depth:
systematic review and meta-analysis. BMJ 2016; 354:i3633.
42. Kyrgiou M, Mitra A, Arbyn M, et al. Fertility and early pregnancy outcomes after
treatment for cervical intraepithelial neoplasia: systematic review and meta-analysis. BMJ
2014; 349:g6192.
43. Kyrgiou M, Arbyn M, Martin-Hirsch P, Paraskevaidis E. Increased risk of preterm birth
after treatment for CIN. BMJ 2012; 345:e5847.
44. Kyrgiou M, Valasoulis G, Stasinou SM, et al. Proportion of cervical excision for cervical
intraepithelial neoplasia as a predictor of pregnancy outcomes. Int J Gynaecol Obstet 2015;
128:141-7.
45. Founta C, Arbyn M, Valasoulis G, et al. Proportion of excision and cervical healing after
large loop excision of the transformation zone for cervical intraepithelial neoplasia. BJOG
2010; 117:1468-74.
24
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
Figures legends
Figure 1. A: Box and whicker plot of the mean methylation for the studied
histological categories. The mean methylation values increase with increasing
disease severity. Each box contains the range from 25% to 75%, dashes inside the
boxes indicate the median value and diamonds the mean value. Whisker limits
indicate the minimum and maximum values, circles out of the whisker ranges
indicate outliers. B: Fit plot of the mean methylation and the histological categories.
The solid line represents the graph of the equation that predicts mean methylation
from the histological category. The shaded area represents the 95 confidence
interval limits of the fitted line and the dashed lines the range of the prediction
limits. The fit performance is represented by R-square=44.33% (Adjusted R-
square=43.96%).
Figure 2: Receiver Operating Characteristic (ROC) curves of the twelve methylation
positions (P01M-P12M) each one corresponding to CpG position (5611, 5726, 5927,
6367, 6389, 6457, 6581, 6650, 6796, 7091, 7136 and 7145) and the mean
methylation (bold lines) for the discrimination of: A) CIN2+ and B) CIN3+ cases. All
methylation positions exhibit similar behavior. The Area Under Curve (AUC) increases
as the cut-off severity increased from CIN2+ to CIN3+, for the majority of the
positions and the mean methylation. Mean methylation exhibits better performance
than most individual positions for both CIN2+ and CIN3+ thresholds.
25
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
Figure 3: Overall accuracy (bold lines), sensitivity (dashed lines) and specificity (solid
lines) for different mean methylation threshold values. Performance characteristics
are presented for: A) CIN2+ B) CIN3+. The horizontal axis represents different
threshold values for the mean methylation. Higher methylation thresholds increase
specificity, but at the cost of sensitivity.
26
603
604
605
606
607
top related