Article Text


Host gene expression profiling of cervical smear is eligible for cancer risk evaluation
  1. Olga Bourmenskaya1,
  2. Ekaterina Shubina2,
  3. Dmitry Trofimov1,
  4. Denis Rebrikov1,
  5. Elina Sabdulaeva3,
  6. Oksana Nepsha1,
  7. Vladimir Bozhenko4,
  8. Svetlana Rogovskaya5,
  9. Gennady Sukhikh1
  1. 1Research Center for Obstetrics, Gynecology and Perinatology, Moscow, Russia
  2. 2Moscow Institute of Physics and Technology, Moscow, Russia
  3. 3I.M. Sechenov First Moscow State Medical University, Moscow, Russia
  4. 4The Russian Scientific Center of Roentgenology & Radiology, Moscow, Russia
  5. 5Russian Medical Academy of Postgraduate Education, Moscow, Russia
  1. Correspondence to Dr Denis Rebrikov, Research Center for Obstetrics, Gynecology and Perinatology, 4 Oparina Street, Moscow 117513, Russia; ncagip4{at}


Aims Uterine cervical carcinoma (CC) is known to be a delayed consequence of human papillomavirus (HPV) infection. Considering the reported influence of HPV on host genome activity, we conceived an approach to capture human gene expression profiles corresponding to increased risks of carcinogenesis.

Methods A sample set of 143 female participants included a ‘control’ group of 23, a ‘pathology’ group of 83 (cervical abnormalities of varied grade including 10 cases of CC), and a ‘HPV carrier’ group of 37 (infected but manifesting normal cytology). HPV detection, viral load measurements and gene expression profiling were performed by real-time PCR assays.

Results Gradual increase in expression of proliferation markers and a decrease in expression of proapoptotic genes, some receptors, PTEN and PTGS2 were demonstrated for progressive grades of cervical intraepithelial neoplasia leading to cancer. All reported trends were statistically significant, for instance, correlation of gene expression values for MKI67, CCNB1 and BIRC5.

A model was proposed that employed mRNA concentrations for genes MKI67, CDKN2A, PGR and BAX. Prompt distinction between the norm and the cancer, provided by initial calculation, suggested that positive values of the function could indicate the higher individual risks. Indeed, all patients assigned to high risk by calculation were HPV infected and showed elevated viral E6, E7 mRNA concentration known to be associated with CC onset.

Conclusions The research was concentrated on dynamical gene expression profiling upon pathological changes ultimately leading to CC. Differences of normalised mRNA concentrations were used for quantitative model design and its primary approbation.

Statistics from


Uterine cervical carcinoma (CC) is the third most common oncological condition in women after breast and colorectal cancers (9%). Hundreds of thousands of new CC cases, and hundreds of thousands of fatalities from this decease are reported every year (eg, 529 800 (new) and 275 100 (fatalities) in 2008). About 85% of fatalities from CC occur in developing countries. A decrease in CC case numbers in Europe and North America in recent decades is got at a cost of screening programmes introduced into national healthcare systems.1 

Papanicolaou cervical smear test (Pap test), a cytological assay used for CC prognosis, reveals the presence of abnormal cells. Some variants of cell morphology, for instance, so-called ‘atypical squamous cells of undetermined significance’, cannot be interpreted definitely. This inaccuracy adds to a fraction of CC cases which pass unnoticed by screening examination to emerge 3–5 years later.2

Almost all CC cases (99.7%) are associated with the human papillomavirus (HPV) infection.3 The use of HPV DNA tests as a complementary screening procedure considerably increases the reliability of prognosis. HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73, 82, and also possibly types 53, 26 and 66 are carcinogenic; these are so-called high-risk (hrHPV).4

Although more than 50% of women become infected with HPV at least once in a lifetime, most HPV infections clear on their own within 1–2 years following incident. In the remaining 10% cases, virus persistence may lead to carcinogenesis; full CC development from HPV infection takes about 10 years.2 ,5 It is still not possible to predict whether the infection will clear or persist. The attempts are underway to use other markers for CC development risk assessment. A number of candidate molecular markers already include viral E6, E7 and human MKI67 and CDKN2A mRNA species. Their proper use can significantly contribute to CC diagnosis, prognosis and treatment.

Cervical intraepithelial neoplasia and CC are considered sequential outcomes of HPV infection. They are thought to be related to increased expression of HPV proteins E6 and E7, accompanying integration of the viral DNA into host cell genome.6 ,7 The HPV proteins E5, E6 and E7 are known to activate cell growth and transformation: E6 interacts with the tumour suppressor p53 and the apoptosis activator Bak mediating their degradation, while E7 interacts with pRB; the latter interaction leads to E2F release, enhanced production of p16INK4A and activated cell proliferation.8

Taking into account the influence of viral E6 and E7 proteins on host cell genome activity, we conceived an approach to capture human gene expression profiles corresponding to increased risks of carcinogenesis. Here, we describe some distinct modes of gene expression related to observable grade of cervical intraepithelial neoplasia. We propose a quantitative model for CC prediction based on trends in molecular dynamics.


The patient management always corresponded to medical research participation standards, including obligate informed consent procedure. The material was collected from each patient in accordance with gynaecology practice standards. HPV viral load measurements and genotyping were performed by real-time PCR assays (DNA-Technology JSC, Russia).

Human MKI67, CCNB1, BIRC5, AURKA, MYBL2, CDKN2A, NDRG1, CTSL2, BCL2, BAX, BAG1, ESR1, PGR, CD68, PTEN, TERT, GSTM1, PTGS2, SCUBE mRNA concentrations were measured by quantitative real-time reverse transcription PCR assay (DNA-Technology JSC, Russia), and normalised by ΔΔCq transformation using four reference genes (HPRT1, TBP, GUSB, B2M) and the control group median value for zero.

The median value (Me) was taken for a calculated central tendency value for all quantitative measurements, with upper and lower quartiles taken as a CI and results represented as Me (0.25–0.75). Statistical significance of differences between groups was determined using Mann–Whitney test for quantitative data or χ2 test with Yates correction for small sample sets for qualitative data. Correlations were evaluated using Spearman rank correlation test.


Clinical characterisation of participants

According to Bethesda System for reporting cervical cytological diagnoses, low-grade squamous intraepithelial lesions (LSIL) and high-grade squamous intraepithelial lesions (HSIL) are two basic types of abnormal cervical cytology. LSIL generally correspond to histological diagnoses of flat condyloma, cervical intraepithelial neoplasia grade I or low-grade dysplasia; HSIL correspond to cervical intraepithelial neoplasia grade II–III, medium- or high-grade dysplasia.9 LSIL and HSIL indicate sequential steps of pathogenic process possibly leading to CC.

Of 143 female patients aged 18–55 years (31±7.6 years on average), 23 patients manifested normal cervical cytology clean of HPV; these constituted a ‘control’ group. A ‘pathology’ group consisted of 83 patients with cervical abnormalities, including 35 cases of LSIL, 38 cases of HSIL and 10 cases of CC. The rest of the 37 patients were assigned to an ‘HPV carrier’ group, for they had cervical HPV infections but manifested normal cervical cytology.

All participants were questioned on sexual activity, for instance, the number of sexual partners. The diverse’ sexual activity turned out to be associated with increased risks of cervical pathology, since the percentage of women who admitted having sex with more than five men ever was almost two times higher in the pathology group as compared with the control group. This calls up the well-known opinion on the nature of risk factors for specifically female conditions, such as cervical pathology, and reminds of special healthcare demands for some social groups and lifestyles.

Hpv genotyping and viral load measurements

The hrHPV types were detected in 23 LSIL probes (66%), 35 HSIL probes (92%) and six CC probes (60%). This corresponds to incidence of hrHPV in cervical pathology stated in reports based on meta-analysis of 8308 LSIL, 4338 HSIL and 8550 CC cases.10 ,11 The incidence of hrHPV in HSIL was significantly higher than in LSIL (p=0.013).

In 55.9% of the hrHPV-positive probes of the pathology group, HPV was represented by one type, in 27.5% by two and in 16.6% by three or more types. The most common hrHPV type was 16 followed by other types (table 1).

Table 1

HPV types in hrHPV-positive probes of the pathology group

HPV viral load values are given in table 2.

Table 2

HPV viral load

Identification of human gene expression profiles

All data for this section are presented in table 3 with p-values confirming significance of all reported trends.

Table 3

Human gene expression profiles

MKI67 expression was 2.6 times higher for the HPV carrier group than for the control group.

A 1.6 times increase in MK167 expression, a similar 3.3 times increase for CDKN2A, a 4.2 times increase for CTSL2, and a 1.4 times decrease for BCL2 were found in LSIL probes as compared with the control group.

More pronounced increases in MKI67 and CDKN2A expression (4.7 and 8.3, respectively), increases for other proliferation markers (four times for MYBL2, 1.5–2 times for CCNB1, BIRC5, AURKA, NDRG1) and decreases for BCL2 and PTEN (1.2 and 1.4 times, respectively), were found in HSIL probes as compared with the control group.

In CC probes, the increase in expression of proliferation markers was yet more pronounced (14 times for MKI67, nine times for CDKN2A, 3–4 times for MYBL2, CCNB1, BIRC5, AURKA) and paralleled with a drastic decrease of expression for oestrogen and progesterone receptors (34 for ESR1, 45 times for PGR), and also a decrease for BCL2, BAX, PTEN, CD68, PTGS2 (5.4, 2.5, 2.1, 3 and 8.2 times, respectively) as compared with the control group.

The telomerase gene TERT and the detoxication glutathione transferase gene GSTM1 showed the weakest expression. Their mRNA was detected for only 41 and 75 patients, respectively (29% and 52% of total sample). The apparent absence of TERT mRNA in many probes was probably due to the effect of weak expression exacerbated by small quantities of material. The absence of GSTM1 mRNA might be explained by 0/0 deletion naturally occurring with a frequency up to 50%.12

Gene expression values were found to be correlated for three proliferation markers, MKI67, CCNB1 and BIRC5 (rS=0.77–0.82, p<0.01), and also for ESR1 and PGR (rS=0.76, p<0.001).

Finally, a quantitative model was proposed to correlate gene expression data with individual risks of CC. The first step in its design was to obtain a characteristic distinction between the norm and the cancer.

Such a distinction was provided by the formulaEmbedded Image (square brackets indicate normalised mRNA concentrations) that is, negative values for all patients of the control group and positive values for all patients with CC. By assumption, positive values of the function could indicate the higher individual risks of CC.

The results of calculations are presented in figure 1. Only one patient of the HPV carrier group was assigned to ‘CC calculated high risk’. Her tissue probe contained seven hrHPV types including type 16 and also elevated E6, E7 mRNA concentration was shown for this probe upon total HPV viral load of two copies per cell.

Figure 1

Application of formula combining individual normalised mRNA concentrations. The diagram shows distribution of the function values calculated individually for all participants by formula (1). The data are presented in arrays of individual tags plotted against linear scale and arranged in parallel lines by current diagnoses.

Four patients of the LSIL subgroup (11%) assigned to ‘СС calculated high risk’ were hrHPV-positive; also elevated E6, E7 mRNA concentrations were shown for two of the probes.

Ten patients (26%) of the HSIL subgroup assigned to ‘СС calculated high risk’. were hrHPV-positive; also elevated E6, E7 mRNA concentrations were shown for nine of the probes.

Thus, positive values of the function were likely to correspond to the higher risks of CC development.


Of the given sample set, hrHPV most commonly emerged in HSIL probes. Of HPV phylogenetic groups, determined according to the Burk classification system,13 ,14 predominant were α9 (types 16, 31, 58, 52, 33 and 35) and α7 (types 45, 39, 18, 68, 59); of single HPV types predominant was 16. Also relatively high frequencies for Asian types 58 and 52 and relatively low frequency for type 18 are noteworthy.

The association of HPV loads with extent of neoplasia (estimated by grade of intraepithelial lesions) corresponds to the results of Gravitt et al.15 CI (0.25–0.75) for the viral load overlap in their upper parts. The apparent insufficiency of viral load as a single value for cancer risk assessment emphasises the demand for other relevant markers.

The results on gene expression implicate MKI67 and CDKN2A as candidate positive’ markers of carcinogenesis. The genes participate in proliferation and cell-cycle regulation. The MKI67 protein products are active in G1, S, G2 and M phases of cell cycle;16 CDKN2A encodes proteins that regulate p53 and RB1 cell-cycle regulatory pathways.17

The results indicate the association of carcinogenesis with a decrease of mRNA concentration for oestrogen receptor ESR1, progesterone receptor PGR and macrophage scavenger receptor CD68.

Tumours characterised by uncontrolled cell proliferation are also deficient in apoptotic activity. Decreases in expression demonstrated for proapoptotic BAX and antiapoptotic BCL2 and BAG1, as well as in CC probes, may seem controversial. Perhaps, activity of BCL2 is more complex, for BCL2 overexpression in osteoblasts was shown to induce apoptosis in vitro,18 and an increase of BCL2 expression in vivo was shown to activate proliferation improving net cell survival.19

The results also indicate association of the pathological processes with decreased expression of PTGS2 and PTEN. Cyclo-oxygenase PTGS2 expression is usually associated with inflammatory processes, de novo angiogenesis and metastases.20 It is considered antiapoptotic;21 it correlates with cervical intraepithelial neoplasia grade.22 Our results for PTGS2 provoke further investigation, for they do not match this concept. PTEN protein product is multifunctional: a cell-cycle regulator, it is also involved in apoptosis as a BCL2 inhibitor; it weakens cell adhesion, thus augmenting cell propagation and migration, and interferes with growth factor signalling. Decreases in PTEN activity are commonly associated with late tumours and metastases and, notably, with endometrial intraepithelial neoplasia progression to cancer.23 ,24


The research was concentrated on dynamical gene expression profiling upon pathological changes ultimately leading to CC. Differences of normalised mRNA concentrations were used for quantitative model design and its primary approbation. The results imply that aside from HPV genotyping, the measurement of host gene expression in cervical pathology assessment is also warranted. We believe that judicious use of this approach can reinforce diagnostics and management of cervical pathologies.

Take home messages

  • The results indicate association of cervical intraepithelial neoplasia with progressive changes in host gene expression. A formula combining normalised mRNA concentrations is approbated as a quantitative measure of cancer risks. The measurement of host gene expression in cervical pathology assessment is warranted.

What the paper adds

  • The results imply that aside from HPV genotyping, the measurement of host gene expression in cervical pathology assessment is also warranted. We believe that judicious use of this approach can reinforce diagnostics and management of cervical pathologies.

View Abstract


  • Contributors OB and ES carried out the real-time reverse transcription PCR studies, performed statistical analysis and wrote the manuscript; DT and DR directed the research, were responsible for its design and helped to draft the manuscript; ES and ON participated in HPV genotyping and sequence alignment; VB and SR arranged datasets and coordinated patient management; GS conceived of the study.

  • Funding This study was supported by Ministry of Education and Science of Russian Federation (State contract no 16.522.12.2009 from 29.09.2011).

  • Competing interests None.

  • Ethics approval Research Center for Obstetrics, Gynecology and Perinatology Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.