PAX1 and SEPT9 methylation analyses in cervical exfoliated cells are highly efficient for detecting cervical (pre)cancer in hrHPV-positive women

Abstract

Studies have investigated PAX1 and SEPT methylation were closely associated with cervical cancer. For this study, we verified the expressions of PAX1 and SEPT9 methylation in 236 hrHPV women cervical exfoliated cells by using quantitative methylation-specific PCR and we further explored their diagnostic value in cervical (pre)cancer detection. Our results identified that the methylation rates and levels of PAX1 and SEPT9 increased with cervical lesion severity. For a diagnosis of cervical (pre)cancer, the area under the curve (AUC) of PAX1 methylation was 0.77 (95% CI 0.71–0.83) and the AUC of SEPT9 methylation was 0.86 (95% CI 0.81∼0.90). Analyses of the PAX1 and SEPT9 methylation statuses alone or combined with commonly used tests can efficiently identify cervical (pre)cancer. In particular, SEPT9 methylation might serve as an effective and powerful biomarker for the diagnosis of cervical (pre)cancer and as an alternative triage test in HPV-based cervical (pre)cancer screening programs.

Impact Statement

• What is already known on this subject? This subject showed that PAX1 and SEPT9 methylation were closely associated with cervical cancer. The methylation rates and levels of PAX1 and SEPT9 increased with cervical lesion severity and reached a peak in cervical cancer exfoliated cells. We further assessed the diagnostic performances of PAX1 and SEPT9 methylation in cervical cancer screening. In detecting cervical (pre)cancer, the sensitivity values of PAX1 and SEPT9 methylation were up to 61.18% and 82.35%, respectively, and the specificity values of PAX1 and SEPT9 methylation were up to 95.36% and 86.75%, respectively. Moreover, the ROC curve analysis showed AUC values of 0.77 for PAX1 methylation and 0.86 for SEPT9 methylation tests, which were significantly superior to other commonly used tests. These findings suggest that PAX1 and SEPT9 methylation detection may have great clinical potential in cervical cancer screening.

• What the results of this study add? The rates and levels of PAX1 and SEPT9 methylation increased with the severity of the cervical lesions. For a diagnosis of cervical (pre)cancer, the area under the curve (AUC) of PAX1 methylation was 0.77 (95% CI 0.71–0.83), and the sensitivity and specificity values were 61.18% and 95.36%, respectively. The AUC value of the SEPT9 methylation was 0.86 (95% CI 0.81 ∼ 0.90), and the sensitivity and specificity values were 82.35% and 86.75%, respectively. Compared with the various tests we conducted, the PAX1 methylation showed the highest specificity (95.36%), and the SEPT9 methylation demonstrated the highest accuracy(86.00%).

• What the implications are of these findings for clinical practice and/or further research? The methylation levels of PAX1 and SEPT9 had a certain predictive effect on the severity of cervical lesions in hrHPV-positive women. In addition, SEPT9 methylation analysis performs better than PAX1 methylation analysis and commonly used tests in cervical exfoliated cells for detecting cervical (pre)cancer in hrHPV-positive women. SEPT9 methylation analysis merits consideration as an effective and objective, alternative triage test in HPV-based cervical (pre)cancer screening programs.

Keywords:

• PAX1
• SEPT9
• DNA methylation
• human papillomavirus
• cervical (pre)cancer
• biomarker

1. Introduction

The latest statistics show cervical cancer ranks as the fourth most frequently diagnosed cancer and the most common gynaecological cancer, with an estimated 604,000 new cases and 342,000 deaths per year (Sung et al. 2021). High-risk human papillomavirus (hrHPV) is the virtually necessary but not sufficient cause of cervical cancer. It takes decades to develop from virus infection to cervical cancer. Worldwide, cervical cancer incidence and mortality rates reportedly have been in decline in many high-income countries due to hrHPV vaccination and early detection screening programs. However, high incidence rates are observed in low- and middle-income resource-limited countries where is the majority of the global cervical cancer burden (Sung et al. 2021, Huang et al. 2022).

Around the world, cytology and/or HPV-based screening are currently the most commonly used methods for cervical cancer screening (Partanen et al. 2021). There is strong evidence that hrHPV testing has superior sensitivity for detecting cervical (pre)cancer (Elfstrom et al. 2014). However, 90% of HPV infections clear the virus spontaneously; thus, HPV testing cannot identify transient infections (Giorgi-Rossi et al. 2012, Vink et al. 2020). The combined use of cytology and HPV testing maximises sensitivity, while increasing specificity to detect certain cervical (pre)cancers (Sahasrabuddhe et al. 2011). However, the obvious disadvantages of cytology testing include high false-negative rates and the fact that cytology testing is easily affected by sampling, i.e. laboratory and pathologist-subjective interpretations. To date, even high-quality Pap cytology may miss >30% of cervical (pre)cancer cases (Lorincz 2016). Thus, it is important to find an efficient triage strategy that is sensitive enough to detect cervical (pre)cancer but that has high enough specificity to rule out HPV-positive women without potential for progression.

Aberrant DNA methylation leads to the abnormal expression of genes in cervical cancer, which results in the activation of oncogenes; the silencing and inactivation of tumour suppressor genes; cell transformation; and the loss of imprinting, leading to genomic instability and cancer (Bhat et al. 2016, Tu et al. 2022). DNA methylation has been shown to be associated with the severity of cervical intraepithelial neoplasia (CIN) lesions and the risk of invasive cancer. Prior studies have identified some genes that are hypermethylated in cervical cancer, such as PAX1, ZNF582, FAM19A4, CADM1, JAM3, MIR124 and SOX1 (Liang et al. 2020, Zhang et al. 2022a). However, not all of these targets are applicable to clinical practice due to their relatively low sensitivity and specificity, which need serial validation studies to confirm whether these targets can be used effectively in clinical situations. A large number of clinical studies have confirmed that PAX1 methylation, as a biomarker, can effectively detect cervical (pre)cancer in HPV-positive women with a high degree of accuracy (Fang et al. 2019). Studies have evaluated the feasibility of combining HPV methylation with host gene methylation to improve the detection of CIN3 lesions (Louvanto et al. 2015, Hsu et al. 2017). In 2016, methylated PAX1 assays have been approved as an adjunct screening method for cervical cancer by Taiwan FDA (Hsu et al. 2017). However, carcinogenesis is a complex biological process involving multiple genes not just individual ones, thus scientists are actively searching for more effective biomarkers. SEPT9 (Septin9) is defined as a novel tumour biomarker, a cyclin and a member of the Septin gene family involved in cell division, cell cycle control, vesicle trafficking, cytoskeleton and tumorigenesis biological behaviour (Kuhlenbaumer et al. 2005, Lyu et al. 2020). Methylated SEPT9 has been approved and used clinically as the first plasma-based biomarker for colorectal cancer screening by FDA (Zhao et al. 2019). Additionally, hypermethylated SEPT9 was also reported in various cancers including cervical cancer and described as a promising biomarker for Nasopharyngeal Cancer, breast cancer, hepatocellular cancer, etc (Zhao et al. 2019, Lyu et al. 2020, Li et al. 2022). Jiao et al (Jiao et al. 2019) revealed that SEPT9 methylation exhibited significantly differential expression between cervical cancer and normal tissues. Therefore, all these studies made us realise that PAX1 and SEPT9 methylation has the potential to become an effective triage for detection of cervical (pre)cancer.

In the present study, we verified the expressions of PAX1 and SEPT9 methylaiton in cervical (pre)cancer in hrHPV women cervical exfoliated cells. we further compared the diagnosis performance of PAX1 and SEPT9 methylation for detection of cervical (per)cancer and investigated the value of the two genes methylation and other conventional tests in cervical cancer screening.

2. Materials and methods

2.1. Study population

We collected cervical scrapings on 837 hrHPV-positive patients aged 20–67 years, who visited the Guangdong Women and Children Hospital between March 2020 and December 2020 (Figure 1). All samples were taken by experienced gynaecologists using a cervical brush and all patients voluntarily accept HPV DNA genotyping and cytology test. HPV DNA genotyping tests were detected in the Department of Clinical Laboratory at our hospital by fluorescence quantification PCR in accordance with the manufacturer’s instructions (Kaipu Company, China). This method could identify 14 hrHPV genotypes (HPV-16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68) and other low-risk HPV genotypes. Cytologic smear specimens were prepared using a customised liquid-based cytology method – the ThinPrep Pap system – according to the manufacturer’s protocol (Hologic, MA, USA). The cytologic diagnoses were evaluated by two or more academic cytopathologists in accordance with the Bethesda 2001 criteria.

Figure 1. Study population. LLETZ: large loop excision of the transformation zone; UE: uterus expiration; CIN: cervical intraepithelial neoplasia; ASCUS: atypical squamous cells of unknown significance; No CIN: no cervical intraepithelial neoplasia or normal cervix/cervicitis; HSIL: high-grade squamous intraepithelial lesion; LSIL: low-grade squamous intraepithelial lesion; CC: cervical cancer; hrHPV: HPV31, 33, 35, 39, 45 51, 52, 56, 58, 59, 66 and 68.

Cervical exfoliated cells that tested for HPV genotyping preserved by RNAstore solution (CWBiotech company, Beijing, China) were stored at −20 °C for subsequent methylation analysis. Eligible patients were referred for colposcopy and cervical biopsies were taken from every visible lesion or two random biopsies (6 and 12 o’clock) if no lesions are visible, according to international criteria. Eventually, a total of 236 cases who were histologically confirmed as different cervical lesions by two experienced pathologists were included into this study. Exclusion criteria were as follows: pregnancy, lactation, previous hysterectomy, previous history of abnormal cervical epithelial lesions and previous history of cancer.

This study was approved by the Guangdong Women and Children Hospital Ethics Committee (reference number, 201701005), We obtained informed written consent prior to specimen collection, according to institutional guidelines.

2.2. DNA extraction and bisulphite modification

Genomic DNA from cervical exfoliated cells was isolated using a magnetic bead-based TIANamp Genomic DNA Extraction Kit (Tiangen, Shanghai, Jingshan Biotechnology). DNA concentrations were measured using a Quawell Q5000 UV spectrophotometer. The extracted DNA was treated with bisulphite using an EZ DNA methylation-Gold kit (Zymo Research, USA) according to the manufacturer’s instructions. The bisulphite-modified DNA was immediately tested by qMS-PCR assay.

2.3. Quantitative methylation-specific PCR (qMS-PCR)

The primer sequences of target genes (PAX1 and SEPT9) and internal reference gene (housekeeping gene beta actin (ACTB) are shown in Table 1. Every sample from bisulphite-modified DNA was performed in triplicate by ABI 7500 PCR System (Life Tech, USA). The thermal cycling conditions were 1 cycle at 94 °C for 10 min, followed by 50 cycles at 93 °C for 20s, 56 °C for 60 s and 65 °C for 30 s. The circulating threshold (CT value) and the amplification curve of ACTB and SEPT9 were obtained by collecting the fluorescent signal to determine the amplification results.

Table 1. Primers and probes used for qMS-PCR.

Primer/probe	Sequence (5′-3′)	Amplicon length (bp)
PAX1-F	TATTTTGGGTTTGGGGTCGC	153
PAX1-R	CCCGAAAACCGAAAACCG	153
PAX1-probe	FAM-TTTTTGTTTTAGAGAGGTTAGTAAT-BHQ1
SEPT9-F	TTTAGTTAGCGCGTAGGGTTC	81
SEPT9-R	AACTAATAAACAACGAATCGCG	81
Septin 9-probe	FAM-ACGCCCCCGACGAAACC-BHQ1
ACTB-F	ATAATAAAAAGGAGGTTGGAT	125
ACTB-R	CTCCCRCAAAACAACCAC	125
ACTB-probe	VIC-CCACCTTACCCTAAACACTACAAC-BHQ1

Samples with a CT value >45 for SEPT9 were considered to represent a negative test result. All samples had a CT value ≥32 for ACTB and were considered to represent an invalid result. Each sample with two valid amplification curves of three PCR replicates and an average △CT(target gene − ACTB) <9 was considered to represent a positive test result. Methylation scores were calculated using the following formula: 2^{[Ct (ACTB) -Ct (target gene)]} × 100 (De Strooper et al. 2014).

2.4. Statistical analysis

Statistical analyses were performed using the IBM SPSS Statistics Version 26 (IBM Corp, Armonk, NY, USA), graphs were done using GraphPad Prism 7.0 and Medcale software applications, and all analyses were two-sided. The normal continuous variables were expressed as means ± σ. The median (P¼−P¾) was used for continuous variables of abnormal distribution. A chi-squared test and Fisher’s test were used to analyse the categorical variables. Kruskal–Wallis H and Mann–Whitney U tests were used for univariate analysis of continuous variables. The Cochran–Armitage trend tests were used to analyse the linear correlations. A p value <.05 was considered to be statistically significant for the above statistical methods.

3. Results

3.1. Clinical data comparison of patients

The average age of the 236 patients was 38.27 ± 10.48 years. HPV genotyping revealed 91 HPV16/18-positive patients and 145 other hrHPV-positive patients. There were 154 patients with cytology ≥ atypical squamous cells of unknown significance (ASCUS) and 82 patients with cytology < ASCUS. The histology results revealed that 22 patients had cervical cancer (CC, including 17 squamous cell carcinomas, 4 adenocarcinomas and 1 gastric-type endocervical adenocarcinoma), 63 patients had high-grade squamous intraepithelial lesions (HSIL), 70 patients had low-grade squamous intraepithelial lesions (LSIL) and 81 patients had normal cervixes and chronic cervicitis (No CIN).

3.2. Methylation rates and levels of PAX1 and SEPT9 in different cervical lesions

The PAX1 and SEPT9 methylation rates among the four groups showed significant differences (p value <.001) (Table 2). The Cochran–Armitage trend tests confirmed that there was a linear trend between cervical lesions and both PAX1 methylation and SEPT9 methylation (p value <.001), which meant that the PAX1 and SEPT9 methylation rates gradually increased with the severity of the cervical lesion.

Table 2. Comparisons of methylated Septin 9 and PAX1 in different severities of cervical lesions.

	PAX1 methylation			SEPT9 methylation
Groups	Methylation rates% (n1/N)	Median methylation scores (P¼−P¾)	p Value	Methylation rates% (n2/N)	Median methylation scores (P¼−P¾)	p Value
No CIN	3.70% (3/81)	0.01 (0.000–0.03)	1.73× 10^−16	9.88% (8/81)	0.01 (0.002-0.05)	3.98× 10^−22
LSIL^a	5.71% (4/70)	0.000 (0.000–0.02)		17.14% (12/70)	0.02 (0.005–0.15)
HSIL^b,c	47.62% (30/63)	0.06 (0.000–3.39)		76.19% (48/63)	0.91 (0.22–9.32)
CC^d,e,f	100.00% (22/22)	63.76 (18.89–86.11)		100.00% (22/22)	149.70 (61.93–240.78)

Note: n1: number of PAX1 methylation positive; n2: number of SEPT9 methylation positive; N: total number in the group; p value <.05, Kruskal–Wallis H test.

^aCompared with No CIN, p value >.05, Kruskal–Wallis H test.

^bCompared with No CIN, p value <.001, Kruskal–Wallis H test.

^cCompared with LSIL, p value <.001, Kruskal–Wallis H test.

^dCompared with HSIL, p value <.001, Kruskal–Wallis H test.

^eCompared with LSIL, p value <.001, Kruskal–Wallis H test.

^fCompared with No CIN, p value <.001, Kruskal–Wallis H test.

Table 3. Sensitivity, specificity, PPV, NPV, YI and AUC of different tests for detecting cervical (pre)cancer.

	Sensitivity% (95%CI)	Specificity % (95%CI)	PPV% (95%CI)	NPV %(95%CI)	YI %	AUC （95%CI）
SEPT9 methylation	82.35^▲▲	86.75^▲▲	77.78^▲▲	89.73^▲▲	69.10	0.86^▲
	(72.90–89.00)	(80.42–91.26)	(68.16–85.13)	(83.75–93.68)		(0.81–0.90)
PAX1 methylation	61.18**	95.36**	88.14**	81.36**	56.54	0.77*
	(50.55–70.84)	(90.74-97.73)	(77.48–94.14)	(74.98–86.41)		(0.71–0.83)
HPV16/18 genotyping	49.41*^▲▲	67.55*^▲	46.15*^▲	70.34*^▲	16.96	0.59*^▲
	(39.04–59.83)	(59.73–74.50)	(36.27–56.34)	(62.46–77.17)		(0.52–0.65)
Cytology	80.00**^▲▲	43.05*^▲	44.16*^▲	79.27*^▲	23.05	0.62*^▲
	(70.28–87.12)	(35.42–51.02)	(36.55–52.05)	(69.29–86.63)		(0.55–0.68)
HPV16/18 genotyping and/or Cytology	91.76**^▲▲	17.22*^▲	38.42*^▲	78.79*^▲	8.98	0.55*^▲
	(83.96–95.95)	(12.03–24.04)	(32.00–45.27)	(62.25–89.33)		(0.48–0.61)
SEPT9 methylation and/or HPV16/18 genotyping	87.06**	59.60*	54.81*	89.11**	46.66	0.73*
	(78.30–92.62)	(51.63–67.09)	(46.40–62.96)	(81.54–93.81)		(0.67–0.79)
SEPT9 methylation and/or Cytology	92.94**	37.75*	45.66*	90.48**	30.69	0.65*
	(85.44–96.72)	(30.41–45.70)	(38.41–53.10)	(80.74–95.56)		(0.59–0.71)
PAX1 methylation and/or HPV16/18 genotyping	77.65^▲▲	65.56^▲	55.93^▲	83.90^▲▲	43.21	0.72^▲
	(67.71–85.20)	(57.68–72.67)	(46.93–64.56)	(76.22–89.44)		(0.65–0.78)
PAX1 methylation and/or Cytology	88.24^▲▲	40.40^▲	45.45^▲	85.92^▲▲	28.64	0.64^▲
	(79.69–93.49)	(32.91–48.37)	(38.04–53.06)	(75.98–92.17)		(0.57–0.71)
SEPT9 and/or PAX1 methylation	82.35**^▲▲	84.77**^▲▲	75.27**^▲▲	89.51**^▲▲	67.12	0.84**^▲
	(72.90–89.00)	(78.18–89.63)	(65.62–82.93)	(83.41–93.54)		(0.78–0.89)

Note: AUC: area under curve; YI: Youden index; PPV: positive predictive value; NPV: negative predictive value.

*Compared with SEPT9 methylation test alone, p value <.05; **compared with SEPT9 methylation test alone, p value >.05; ^▲ compared with PAX1 methylation test alone, p value <.05; ^▲▲ compared with PAX1 methylation test alone, p value >.05.

The methylation levels of PAX1 and SEPT9 both increased linearly with an increase in cervical lesion severity by calculating methylation scores (p value <.05) (Figure 2). Pairwise comparisons of the groups showed that the methylation scores of PAX1 and SEPT9 in the HSIL, LSIL and No CIN groups were all lower than those of the CC group (p value <.05). The methylation scores of the HSIL group were also higher than those of the LSIL and No CIN groups (p value <.05). The methylation scores of the LSIL and the No CIN groups showed no significant differences (p value >.05).

Figure 2. The PAX1 and SEPT9 methylation levels were determined by using a qMS-PCR assay. Methylation scores of every sample were calculated using the above formula. In the scatter plots, the lines represent the 25th median and 75th percentiles.

3.3. Clinical performance of different methods for detecting cervical (pre)cancer

We calculated the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and Youden index (YI) for all the individual and combined tests in our study (Table 3). According to a comprehensive comparison of all the screening methods, the results showed that the sensitivity of the SEPT9 methylation test was 82.35%, which was higher than that of other individual tests but lower than that of the combined tests. The specificity (86.75%) and PPV (77.78%) values of the SEPT9 methylation test were higher than those of other tests, except for the PAX1 methylation test. The NPV of the SEPT9 methylation test was 89.73%, which was higher than those of other tests, except the SEPT9 methylation combined with cytology test. The sensitivity of the PAX1 methylation test was higher than that of the HPV16/18 genotyping (61.18% vs. 49.41%, p value >.05) but lower than other tests, but the PAX1 methylation test had the highest specificity and PPV values. The NPV of the PAX1 methylation test was higher than those of other tests, except the SEPT9 methylation combined with cytology (89.73% vs. 90.48%, p value >.05). Subsequently, we constructed the receiver operating characteristic (ROC) curves and calculated the area under the curve (AUC) (Figure 3). SEPT9 methylation, PAX1 methylation and the combination of the two genes’ methylation tests had higher AUC values, which were 0.86, 0.77 and 0.84 respectively. The AUC values of the SEPT9 methylation test and the combination of the two genes methylation test were greater than those of the PAX1 methylation test, and the differences were statistically significant (0.86 vs. 0.77 and 0.84 vs. 0.77, both p values <.05), but the former two tests had no significant differences (0.86 vs. 0.84, p value >.05).

Figure 3. Comparison of PAX1 and SEPT9 methylation levels in different histological lesions.

4. Discussion

In this study, we used the qMS-PCR assay to analyse the methylation rates and levels of the PAX1 and SEPT9 genes in different cervical lesion samples and assessed the diagnostic performances of PAX1 and SEPT9 methylation in cervical cancer screening. Our results identified that the methylation rates and levels of PAX1 and SEPT9 increased with cervical lesion severity and reached a peak in cervical cancer samples. For cervical cancer screening, the sensitivity values of PAX1 and SEPT9 methylation were up to 61.18% and 82.35%, respectively, and the specificity values of PAX1 and SEPT9 methylation were up to 95.36% and 86.75%, respectively. Moreover, the ROC curve analysis showed AUC values of 0.77 for PAX1 methylation and 0.86 for SEPT9 methylation tests, which were significantly superior to other commonly used tests. These findings suggest that PAX1 and SEPT9 methylation detection may have great clinical potential in cervical cancer screening.

Compelling evidence has demonstrated that aberrant DNA methylation is one of the earliest events during the tumorigenic process and often occurs in the precursor lesions of human cancers, including genital tract cancers (Tian et al. 2017, Kremer et al. 2021). It has been reported that PAX1 and SEPT9 act as tumour suppressor genes and are involved in numerous cellular processes of hypermethylation during the carcinogenesis process (Connolly et al. 2011, Wu et al. 2022). A series of studies have detected PAX1 and SEPT9 methylation in a broad spectrum of tumour tissues including cervical cancer tissue (Jiao et al. 2019, Lyu et al. 2020, Li et al. 2021). Some studies have highlighted the potential of PAX1 methylation in cervical cancer screening (Fang et al. 2019, Liang et al. 2020). However, to the best of our knowledge, this study was the first to investigate the diagnostic capacity of SEPT9 methylation for detecting cervical (pre)cancer in cervical exfoliated cells from hrHPV-positive women. Liu et al. (2020) and Jiao et al. (2019) found that the methylation levels of PAX1 and SEPT9 were elevated with worsening cervical centercancer, closely resembling our results. A review reported on a variety of methylation markers that were clinically evaluated for detecting cervical (pre)cancer, with sensitivity values of 55.00–88.00% and specificity values of 60.00–91.00%, which were consistent with this study (Luttmer et al. 2016). In the present study, both PAX1 and SEPT9 were hypermethylated in all cervical cancer exfoliated cells, with higher methylation frequencies as compared with other genes reported being hypermethylated in cervical cancer (Xu et al. 2019, Li et al. 2021). The methylation levels of PAX1 and SEPT9 in cervical cancer exfoliated cells were the highest, which suggested that even when the methylation test was set at a very high threshold, the probability of missing cervical cancer was small. It is worth noting that, in this study, 22 cervical cancers with PAX1 and SEPT9 hypermethylation also included special types of cervical cancer, which were prone to occur in young women with poor prognosis and no significant association with HPV (Pal et al. 2015). Recent research have indicated that HPV-negative cervical cancers may represent a biologically distinct subset of tumours, relying on a distinct pathogenetic pathway and carrying a poorer prognosis than HPV-positive cervical cancers (Arezzo et al. 2021). Another study revealed the adoption of HPV testing alone would miss the diagnosis of 10–15% HPV-negative HSIL (Bogani et al. 2021). From this perspective, the analyses of PAX1 and SEPT9 methylation levels are both conducive to early detection and treatment of special types of cervical (pre)cancer and HPV-negative cervical (pre)cancer, which is helpful for fertility preservation.

A long-term follow-up study demonstrated that hypermethylated cervical scrapes had a remarkable tendency to progress to cancer after five years, even if they were non-cancer tissues at the time of sample collection (Katki et al. 2013). A study among 149 women with CIN2 indicated that the S5 classifier had the highest sensitivity to predict CIN2 lesions that progressed to CIN3 from those that spontaneously regressed to ≤ CIN1 over a two-year follow-up period as compared with cytology with a sensitivity of 86.9% for S5 versus 75.4% for cytology (Louvanto et al. 2020). In addition, De Strooper et al. (2018) found that negative methylation levels indicated a lower 14-year cervical cancer risk. As previously described, all these findings highlight that methylation analysis specifically detects advanced cervical lesions that have a high short-term risk of progression to cancer, and DNA methylation analysis may act as a promising prognostic test for cervical lesions in hrHPV-positive women (Katki et al. 2013, Koeneman et al. 2015). In this study, hypermethylation was also observed in non-cancer samples, due to two possible reasons: One reason was the quality of the samples and the other reason was that the host DNA had undergone hypermethylation, suggesting a high risk of cervical cancer in the future; however, this hypothesis needs to be further verified by long-term follow-up.

We further compared the diagnostic performances of PAX1 methylation analysis and SEPT9 methylation analysis to those of other commonly used tests alone or in combination for detecting cervical (pre)cancer. In this study, the PAX1 and SEPT9 methylation analyses both provided an advantage over HPV16/18 genotyping and showed markedly higher sensitivity and specificity values for cervical (pre)cancer, which were in accordance with previous studies (Vink et al. 2020, Kremer et al. 2021). The results can be explained by the inherent genotype restriction, HPV16/18 genotyping does not suffice as a direct stand-alone triage strategy for hrHPV-positive women regardless of the sample material (Verhoef et al. 2014). Studies have suggested that methylation markers combined with HPV16/18 genotyping were complementary for detecting cancer and advanced lesions (Kalantari et al. 2014, Lorincz 2016, Luttmer et al. 2016). However, after combining HPV16/18 genotyping with PAX1 and SEPT9 methylation analyses in our study, the sensitivity values for detecting cervical (pre)cancer were slightly increased, but with markedly decreased specificity values; these results were similar to FAM19A4 methylation research (Bu et al. 2018). Our explanation for this phenomenon may be that the results were influenced by the characteristics of high sensitivity and low specificity of HPV16/18 genotyping. Currently, cytology is the most advocated common triage strategy for hrHPV-positive woman in China. Our data revealed slightly increased sensitivity of SEPT9 methylation analysis for detecting cervical (pre)cancer as compared with cytology, while displaying superior specificity values. Although the sensitivity of PAX1 methylation analysis was slightly lower than that of cytology, the specificity also showed a significant advantage over that of cytology, which was in agreement with reported studies (Luttmer et al. 2016, Kelly et al. 2019). The researchers’ explanation for this is that due to cytology relies on subjective interpretation and is highly dependent on experienced pathologists for correct evaluation, which results in a certain rate of misdiagnose (Ronco et al. 2014, Zhang et al. 2021). Although repeated cytology has been recommended to increase specificity, in the meantime, repeated testing is correlated with loss to follow-up (Luttmer et al. 2016). The results of methylation analysis combined with cytology or HPV16/18 genotyping as compared with methylation analysis alone showed slightly increased sensitivity but a very low specificity and PPV for identifying cervical (pre)cancer, which indicated that these joint screening strategies might have induced some over-referrals to specialists and over-treatments. Liang et al. (2020) and Liu et al. (2020) respectively, found that the AUC values of PAX1 methylation analysis for detecting cervical cancer screening were 0.87 and 0.91, which were higher than our results. We speculated that it might be related to the sample size. We were pleasantly surprised to find that the diagnostic value of SEPT9 methylation analysis performed better than PAX1 methylation analysis, which indicated that SEPT9 methylation analysis may be more suitable for cervical (pre)cancer screening, which is expected to be a new hrHPV shunting method. As the ultimate way to diagnose (pre)cancer is by cervical biopsies, the methylation status of SEPT9 might prompt the clinicians to proceed to colposcopy and diagnostic biopsy in the meantime.

The strengths of this study include the fact that it is the first study to investigate the diagnostic capacity of SEPT9 methylation in cervical exfoliated cells. The quantitative analysis of the PAX1 and SEPT9 gene methylation levels can be considered as an objective and automatic method for diagnosis with maximum avoidance of personal errors and rapid mass screening. Despite the promising results, some limitations are present. First, the sample size of each group was relatively small, which may have affected the statistical power. Second, the present study lacked follow-up examinations. We plan to collect longitudinal clinical samples to conduct a large-scale, multicentre, prospective study to further verify the diagnostic performance of PAX1 and SEPT9 methylation analyses. Studies have found that methylation testing is applicable to self-collected cervicovaginal material, even urine material, which would simplify sample collection and increase the screening rate (van den Helder et al. 2022, Zhang et al. 2022b). On the basis of this study, it is important to further explore PAX1 and SEPT9 methylation analyses for detecting cervical (pre)cancer in different cervical self-collected specimens.

5. Conclusion

In summary, this study showed that SEPT9 methylation and PAX1 methylation were closely associated with cervical cancer, and that the methylation levels of SEPT9 and PAX1 had a certain predictive effect on the severity of cervical lesions in hrHPV-positive women. In addition, SEPT9 methylation analysis performs better than PAX1 methylation analysis in cervical exfoliated cells for detecting cervical (pre)cancer in hrHPV-positive women. SEPT9 methylation analysis merits consideration as an effective and objective, alternative triage test in HPV-based cervical (pre)cancer screening programs.

Ethical approval

All patients provided informed consent and agreed to participate in the study. The present study was approved by the Guangdong Women and Children Hospital Ethics Committee (reference number, 201901093).

Informed consent

All participants gave written informed consent.

Acknowledgments

We thank all the participants who took part in these trials and contributed to this research. We thank Jingshan Biotechnology in China for the help of qMS-PCR detection technology. We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

According to Norwegian data legislation, the data of this study cannot be made generally available. Requests should be sent to the corresponding author.

Additional information

Funding

The study was funded by the Technology Project of Guangzhou (Reference: No.202002030174).