MiR-9, miR-21, and miR-155 as potential biomarkers for HPV positive and negative cervical cancer

A total of 52 FFPE cervical cancer tissue samples and 50 FFPE normal cervical tissue samples were used from the Department of Pathology, Yonsei University Wonju Severance Christian Hospital, Wonju, Republic of Korea, between January 2010 and December 2014 (Table 1). Institutional Ethics Committee at Yonsei University Wonju College of Medicine approved the study protocol (approval no. YWMR-12-4-010) and all subjects provided written informed consent. Cases with tissue biopsies available were reviewed by two pathologists. The 52 cervical cancer samples consisted of tissue samples from 50 squamous cell carcinomas and 2 adenocarcinomas.

Three to four 10-μm thick sections of FFPE cervical tissue were used for total RNA extraction. To remove paraffin from FFPE tissue, 160 μL of Deparaffinization solution (Qiagen, Hilden, Germany) was added and vortexed, followed by incubation for 3 min at 56 °C. RNA extraction was performed using the Qiagen RNeasy FFPE kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Total RNA purity and concentration were determined by measuring the ratio of the absorbance at 260 and 280 nm using an Infinite 200 spectrophotometer (Tecan, Salzburg, Austria). All preparation and handling procedures were conducted under RNase-free conditions. Isolated total RNA was stored at −70 °C until used.

Complementary DNA (cDNA) was synthesized using a TaqMan microRNA Reverse Transcriptase kit (Applied Biosystems by Life Technologies, Foster City, CA, USA) according to manufacturer’s instructions. Briefly, 5 to 10 ng of total RNA was used for cDNA synthesis. The reverse transcriptase (RT) reaction mixture contained 0.15 μL of 100 mM dNTP mix (100 mM each dATP, dGTP, dCTP, and dTTP at a neutral pH), 1 μL of 50 U/μL reverse transcriptase, 1.5 μL of 10× reverse transcriptase buffer, 0.19 μL of 20 U/μL RNase inhibitor, and adjusted the total reaction volume to 15 μL with nuclease free water. The cDNA synthesis reaction was performed as follows: 16 °C for 30 min followed by 42 °C for 30 min, and 85 °C for 5 min.

MiRNA expression was quantified by determining the cycle threshold (C_T) which is the number of PCR cycles required for the fluorescence to exceed a value significantly higher than the background fluorescence, using the TaqMan small RNA assay (Applied Biosystems by Life Technologies) with miRNA specific primers according to manufacturer’s instructions. Briefly, 1.4 μL of cDNA was added to 10 μL of probe qPCR mix and 7.6 μL of nuclease free water. The following TaqMan small RNA assay (Applied Biosystems) primers were used: hsa-miR-9-5p, hsa-miR-21-5p, hsa-miR-155-5p, and RNU6B. All analyzed miRNAs are of human (Homo sapiens) origin and therefore, the prefix “hsa” is omitted throughout the text. RT-qPCR reactions were performed using a CFX96 Real-Time PCR System Detector (Bio-Rad, Hercules, CA, USA). Samples were run in duplicate for each experiment. Data were analyzed using the comparative Ct (2-ΔΔC_T) method using the small nuclear RNA, RNU6B, as an endogenous control. To monitor reagent contamination, negative controls were included for each primer pair. PCR cycling conditions were as follows: 95 °C for 3 min 40 cycles of 95 °C for 15 s and 60 °C for 60 s.

To detect HPV E6/E7 mRNA in FFPE cervical tissues, multiplex RT-qPCR was performed using the TaqMan assay with the OPTIMYGENE HPV E6/E7 mRNA RT-qDx assay kit (Optipharm, Osong, Republic of Korea). PCR primers and the corresponding TaqMan probes were designed for three different sets of HPV regions, with each set of probes targeting their conserved sequence (FAM: HPV genotypes 16, 31, 33, 35, 52, and 58; CY5: HPV genotypes 18, 39, 45, 51, 59, and 68; and HEX: HPV genotypes 53, 56, 66, and 69).

RT-qPCR reactions consisted of 10 μL of 2 × Thunderbird probe qPCR mix (Toyobo, Osaka, Japan), 5 μL of primers and TaqMan probe mixture, 2 μL of template cDNA, and distilled water for a final reaction volume of 20 μL. The multiplex RT-qPCR assay detected the HPV E6 and E7 genes simultaneously in a single tube by incorporating two targets (E6 and E7) using specific TaqMan probes, which were labeled with different fluorophores (FAM, HEX, and Cy5). Positive and negative controls were included throughout the procedure. PCR cycling conditions were as follows: 95 °C for 3 min 45 cycles of 95 °C for 20 s and 60 °C for 40 s. To avoid false negatives because of mRNA degradation, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an endogenous control.

Statistical analysis was performed using GraphPad Prism software version 5.02 (GraphPad, La Jolla, CA, USA) and MedCalc 9.0 software (MedCalc Software Inc., Mariakerke, Belgium). Student’s t-test and Mann Whitney U test were used to determine statistical significance between cervical cancer and normal cervical tissue samples as well as investigate miRNA expression in patients according to HPV infection status. Receiver operating characteristic (ROC) curves were generated to assess diagnostic accuracy of each miRNA, and the area under the ROC curve (AUC) was calculated to measure discriminatory capacity. The best sensitivity/specificity pair was selected based on the maximum likelihood ratio. Univariate and multivariate logistic regression by odds ratio (OR) and 95% confidential interval (95% CI) were performed to assess predictors for cervical cancer diagnosis using the XLSTAT software (Addinsoft, New York, USA). All statistical tests were two-sided, and a P value ≤0.05 was considered statistically significant.

Prior to investigating miRNA expression levels, we first examined HPV E6/E7 mRNA expression in 52 FFPE cervical cancer tissue samples and 50 FFPE normal control samples. Fifteen (28.8%) of the 52 FFPE cervical cancer tissue samples were negative for HR-HPV E6/E7 expression (termed HR-HPV E6/E7-negative), while 37 (71.2%) samples were positive (termed HR-HPV E6/E7-positive). We found that all 50 FFPE normal cervical control samples were negative for HPV E6/E7 mRNA expression (Table 1).

Expression levels of miR-9, miR-21, and miR-155 were investigated in our 52 FFPE cervical cancer tissue samples and 50 FFPE normal cervical tissue controls. All three miRNAs were significantly up regulated in FFPE cervical cancer tissues compared to FFPE normal cervical tissues (P < 0.0001) (Fig. 1a-c). The AUC was 0.7565 [95% confidence interval (CI) = 0.6624–0.8507] in miR-9, 0.8325 (95% CI = 0.7530–0.9120) in miR-21, and 0.8492 (95% CI = 0.7736–0.9249) in miR-155, all of which indicate these miRNAs may be used as potential biomarkers for cervical cancer (Fig. 1d-f).

To assess the potential diagnostic value of these three miRNAs, the performance characteristics sensitivity, specificity, positive predictive value, and negative predictive value were determined and evaluated. The cut-off values of miR-9, miR-21, and miR-155 as determined using the likelihood ratio, were 4.035, 1.975, and 3.880 respectively, for optimal sensitivity and specificity. The sensitivity of miR-9, miR-21, and miR-155 was 67.3% (95% CI = 52.9–79.7), 82.7% (95% CI = 69.7–91.8), and 65.4% (95% CI = 50.9–78.0) respectively, while the specificity was 80.0% (95% CI = 66.3–90.0) for miR-9, 72.0% (95% CI = 57.5–83.8) for miR-21, and 96.0% (95% CI = 86.3–99.5) for miR-155. Positive predictive values (PPVs) of miR-9, miR-21, and miR-155 were 77.8%, 75.4%, and 94.3% respectively, while the respective negative predictive values (NPVs) were 70.2%, 80.0%, and 71.6% (Table 2).

To investigate the expression of miR-9, miR-21, and miR-155 with cervical cancer cases that were HPV E6/E7 mRNA-positive or -negative for HPV E6/E7 mRNA, expression levels of the three miRNAs were analyzed in three groups: HPV E6/E7-positive cancer samples, HPV E6/E7-negative cancer samples, and normal samples. We found that all three miRNAs were significantly up regulated in HR-HPV E6/E7-positive cancer tissue samples compared to normal tissue samples (P < 0.0001), while miR-21 and miR-155 were up regulated in HPV E6/E7-negative cancer tissue samples compared to normal controls (P = 0.0079 and P = 0.0384, respectively). We found no significant difference in miR-9 expression levels between HR-HPV E6/E7-negative cancer samples and normal cervical samples (Fig. 2).

The effectiveness of each three miRNA combined with HPV E6/E7 mRNA assay was investigated by analysis of AUC. In all cases of cervical cancer, the value of AUC was 0.8558 (95% CI 0.7773–0.9343), 0.8135 (95% CI 0.7257–0.9013), 0.8215 (95% CI 0.7349–0.9082), 0.8935 (95% CI 0.8243–0.9626) in HPV E6/E7, HPV E6/E7 + miR-9, HPV E6/E7 + miR-21, and HPV E6/E7 + miR-155, respectively. Especially, in HPV negative cervical cancer, miR-21 and miR-155 in conjunction with the HPV E6/E7 mRNA was shown increased AUC value of 0.7267 (95% CI 0.5776–0.8757) and 0.7000 (95% CI 0.5152–0.8646) respectively, compared to HPV E6/E7 assay (Table 3).

We analyzed the following risk predictors of cervical cancer: age, HPV E6/E7 mRNA-expressing status, and miR-9, miR-21, and miR-155 expression. We found that the highest risk factor was HPV E6/E7 mRNA expression (OR = 244.4, 95% CI = 13.6–4376.4) followed by expression of miR-155 (OR = 41.7, 95% CI = 9.1–191.2), miR-21 (OR = 12.3, 95% CI = 4.8–31.7), and miR-9 (OR = 8.2, 95% CI = 3.3–20.3). Ages below 60 years were not a significant risk factor but being over the age of 60 years did confer a higher risk to cervical cancer (Table 4).

To investigate the predictive power for diagnosing cervical cancer in HPV E6/E7 mRNA negative cases, the odds ratio of the miRNAs was determined based on HPV E6/E7 mRNA expression using multivariate analysis. We found that miR-21 and miR-155 in HPV E6/E7-negative cervical cancer patients conferred a 7.0 (95% CI = 1.3–37.6) and 10.3 (95% CI = 1.5–70.7) fold risk to cervical cancer, respectively (Table 5).