Biomarkers differentiating regression from progression among untreated cervical intraepithelial neoplasia grade 2 lesions.

DNA methylation occurs when methyl is added to a cytosine with CpG sequences. DNA methylation plays an important role in regulating gene expression. The DNA methylation of HPV has been observed in cancer patients. Previous studies [33] , [34] , [35] have reported that the elevated methylation of HPV16/18 L1 genes correlates with the severity of the histological grade. The nested case-control study suggested that HPV DNA methylation showed higher potential performance as a triage test for precancers than cytology, the common triage strategy [36] . Noriko et al. observed DNA methylation in the L1 HPV gene. Compared to the regression group, the methylation ratio of the L1 gene (L1MR) in the progression group was higher [37] . L1MR might indicate the integration status of the HPV genome in the host cell and act as a biomarker for CIN. Isao et al. reported that the methylation of the HPV52 L1 gene correlated with the clinical progression of CIN1/2 [38] . In studies of the correlation between hrHPV subtype L1 gene methylation status and CIN2 regression/progression, the major limitations of the studies include small sample sizes, inconsistent methylation sites, varying follow-up intervals and heterogeneous CIN2 lesions. Therefore, the biomarkers of HPV methylation must be confirmed in further studies.

DNA methylation is a well-researched epigenetic mechanism. It involves adding methyl (–CH3) modifications to preceding cytosines bound to guanines (called CpG dinucleotides), rather than altering the DNA sequence. As an early event, the local hypermethylation of gene promoter regions in TSGs will lead to silence, thereby regulating gene expression during the disease process. DNA methyltransferase 1 (DNMT1) is involved in maintaining established methylation patterns. It has been reported that hrHPV infection regulates DNA methylation in HPV-related cancers. The methylation of TSGs was decreased by silencing E6 and E7 expression. It was found that the HPV E6 and E7 oncogenes directly affect the activity of DNMTs, which may partly explain frequent hypermethylation events during cervical carcinogenesis. The methylation biomarkers of the genes − including miR-124-2, FAM19A4, EPB41L3, ASTN1, DLX1, ITGA4, RXFP3, SOX17 and ZNF671 − have been demonstrated in the clinical performance of the triage among hrHPV-positive subjects. The methylation-mediated silencing of TSGs has been described in CIN2+ lesions. The methylation level increased with the severity and duration of CIN lesions. Moreover, methylation biomarkers were independent of HPV status and histotype. Hence, the possible use of methylation testing as a CIN2 regression biomarker is promising. Louvanto et al. published the first paper on the value of methylation biomarkers in predicting the progression of CIN2 [49] . Compared with cytology and HPV genotyping, the S5 classifier (including the methylation of EPB41L3 and the late regions of HPV16, 18, 31 and 33) was the best biomarker of outcomes among the regression group vs. progression group (OR = 3.39). A multicentre prospective study included untreated CIN2/3 women who were followed up for 24 months [50] . Women showed a higher proportion of regression with a negative FAM19A4/miR124-2 result (74.7 %) than with a positive result (51.4 %) at the baseline. The rate of regression was high when the FAM19A4/miR124-2 methylation test was negative with abnormal cytology (88.4 %) or with HPV16-negativity (85.1 %). FAM19A4/miR124-2 methylation-negative women with HPV16-negativity or abnormal cytology can be used to support conservative management in CIN2/3 patients with a completely visible transformation zone. A multicentre prospective observational study showed that if the GynTect® test of CIN2 patients was negative at the beginning of the study, 12 of 18 patients (66.7 %) regressed over time [51] . However, the postulated negative predictive value (≥90 %) of the GynTect® test could not be proven. The main limits of these studies were the short follow-up time and the few CIN2 subjects. Although the biomarkers of host gene methylation may be associated with the progression risk, the determination of the monitoring and follow-up of CIN2 patients remains questionable. Further experiments are urged to study the association between the clinical relevance of host gene methylation and the regression of CIN2 lesions.

HLAs are involved in regulating the immune system. Genome-wide association studies (GWASs) of cervical cancer have indicated the susceptibility of HLA variants to cancer [52] , [53] . Class II HLA molecules binding to heterologous proteins formed antigen complexes on antigen-presenting cells and then presented the complex to clusters of differentiation 4+ (CD4+) T lymphocytes. As heterologous proteins, HPV-derived peptides can bind to HLA class II regions where genetic variations may influence the efficiency of binding and immune responsiveness. A prospective study of 454 Japanese women evaluated the effects of the HLA class II subtype on the natural history of cervical lesions [54] . The women with histologic CIN1-2 lesions were continuously monitored by cytology and colposcopy testing throughout 5 years. When women diagnosed with CIN1-2 carried the HLA-DRB1*13:02 allele, no one progressed to CIN3 within 5 years. The data suggested that the HLA-DRB1*13:02 allele had a protective effect against CIN1-2 progression to CIN3. A similar conclusion was reported by another study, which showed that DRB1*1302 -positive women had a lower cumulative progression rate of CIN3 than DRB1*1302 -negative women (2.1 % vs. 14.0 %) for 10 years [55] . The allele of class I HLA has also been reported in disease outcomes. Trimble et al. performed an observational cohort study to evaluate the prognostic variables in CIN2/3 women [56] . Compared with HLA*A201 -negative women, the regression was lower (14.3 % vs. 42.3 %) for non-16 HPV CIN2/3 women who were HLA*A201 -positive. They also found that the regression was significantly lower among women with HPV16 only. The possible mechanisms of immunology relate to the inability of certain HLA alleles to efficiently present antigen peptides. However, more prospective studies are needed, especially considering that HLA variants vary in ethnic populations.

Not applicable.

Not applicable.

The new medical technology of the Third Xiangya Hospital of Central South University: 2024 , No. 20.

Although HPV infection is undoubtedly the primary pathogenic factor of CC, cervical and vaginal microbiota are involved in the immune response and metabolic process, which may influence the presence of subsequent CIN [68] . A prospective longitudinal study indicated that the progression of CIN2+ may be associated with elevated microbial diversity and Gardnerella [69] . The relatively decreasing concentration of certain Lactobacillus species and increasing vaginal microbiota diversity should impact HPV clearance and the regression of CIN [70] . A cohort study among Korean women was conducted to evaluate whether vaginal microbiota correlated with the progression of cancer [71] . Likewise, the study suggested that Lactobacillus and Gardnerella had discriminatory values between healthy and CIN subjects, whereas Gardnerella and Streptococcus could discern invasive cancer patients from CIN patients. Dong et al. reported that the increased abundance of Gardnerella and Prevotella was associated with persistent HPV16 infection. The increased abundance of Prevotella was associated with persistent HPV18 infection, whereas the decreased abundance of Lactobacillus was found in women with persistent HR‐HPV infection [72] . The findings showed that the mechanism of increasing Prevotella , which leads to hrHPV infection‐related CIN lesions, may pass through the pathway of the host NF‐κB and C‐myc. Except for bacteria, viruses controlled by the host’s immune system also have adjunct effects on the disease progression of the host [73] . Li et al. investigated the associations between the cervical disease’s status and the eukaryotic virome [74] . The data suggested that vaginal Anelloviridae may be involved in aggravating the severity of the disease. Evidence implied associations between HPV infection, vaginal microbiota composition and the course of CIN disease. However, the correlations between CIN2 regression and microbiota composition remain uncertain. Mitra et al. found that the depletion of Lactobacillus spp. and the presence of Megasphaera , Prevotella timonensis and Gardnerella vaginalis are associated with CIN2 persistence or slower CIN2 regression [75] . Based on these findings, the composition of the vaginal microbiota may be an applicable surveillance biomarker for predicting CIN2 regression.

Moreover, p16 INK4A (also termed p16), a tumour suppressor protein, decelerates the progression of cells to the S phase through the inhibition of cyclin-dependent kinases 4 and 6. Ki-67 is expressed in the active phase of the cell cycle and represents the proliferative activity of tumours. Immunohistochemical staining for p16 and Ki-67 was used to discriminate low-risk lesions from high-risk transformation lesions. It has been reported that CIN1 lesions with p16-positive staining have a significantly higher risk of progressing to CIN3 than p16-negative cases. In addition, p16-negative CIN1 lesions have rarely progressed [40] , [41] , [42] . Miyamoto et al. conducted a retrospective study to compare the frequency of p16 and Ki-67 positivity among Japanese women with the regression, persistence and progression of CIN2 [43] . The analysis illustrated that 122 women with CIN2 had a high tendency of progression (p16+ vs. p16-: 59.5 % vs. 27.9 %; Ki-67+ vs. Ki-67−: 57.9 % vs. 32.6 %) when the immunohistostaining for p16 and Ki-67 was positive with more than 50 % of cells. Miralpeix et al. recruited 96 CIN2 patients without treatment and followed up with them for 12 months at 4-month intervals to evaluate their outcomes according to their p16 statuses [44] . Compared to p16-positive patients, p16-negative patients had a higher spontaneous regression rate (100 % vs. 57 %) during follow-up. Ferreira et al. conducted a cross-sectional study to analyse the correlation of p16 and Ki-67 status in CIN2 lesions with disease regression [45] . The negative staining of p16 and Ki-67 can prevent CIN2 from progressing to CIN3+ in more than 85 % of treated and untreated patients. However, the status of p16 and Ki-67 did not work to predict CIN2 regression or progression among untreated patients. This corroborated Zhang et al.’s study that E6/E7 mRNA may provide predictive information for CIN2 regression, while p16 and Ki-67 proteins provide little value [31] . Guedes et al. prospectively followed up with 42 Brazilian women diagnosed with CIN2 [46] . The results showed no significant failed regression (incomplete regression, persistence or progression) between p16-negative and p16-positive CIN2 lesions. Brun et al. examined the prediction value of clinical factors, p16/Ki-67 status and HPV genotype in the spontaneous regression of CIN2 [47] . Compared to non-16 hrHPV and low-risk HPV patients, the regression rate was lower in patients with HPV16 (46.9 % vs. 75 %). The outcome of CIN2 was not significantly influenced by the status of p16 and Ki-67. Koeneman et al. observed women with hrHPV-positive CIN2 lesions to identify potential predictors for regression without treatment [48] . The results of prognostic factors demonstrated that p16 and Ki-67 staining were insignificantly correlated with spontaneous regression in hrHPV-positive CIN2 patients. Testing positive for p16 and Ki-67 staining was strongly correlated with progression in CIN2 patients. However, many studies have found no significance between p16/Ki-67 staining and CIN2 outcome. We supposed that incompatible results may be caused by the varied interpretation criteria for P16 and Ki-67 immunohistochemistry. Taking Ki-67 as an example, some researchers categorised the expression of Ki-67 into two intervals according to the number of stained cells − low expression and high expression − while others scored the expression of Ki-67 according to the size of the stained area: 0, 1, 2 and 3. In addition, the number of CIN2 patients enrolled in the various studies was relatively small, even less than 100 or 50, and the results were not highly reproducible. Well-designed studies are still needed to confirm the value of P16 and Ki-67 to predict CIN2 regression.

Most cervical cancers (>95 %) are caused by HPV, of which more than 60 % are caused by HPV16/18. Moreover, HPV16/18 are also common types in women without cancer. A cohort study reported that women infected with HPV 16/18 with a negative cytology result have a 13.6 %–17.2 % 10-year risk of CIN3+, but only a 3.0 % risk when infected with other hrHPV subtypes [18] . Therefore, when the exfoliated cells are tested as HPV16/18 positive regardless of the cytology results, the patients should be directly referred to a colposcopy. HPV16 is associated with an increased risk of progression [19] . The clearance rate of HPV16 is low, even in immune-competent women. HPV-16 persistence decreased the rate of regression of CIN2 during follow-up. This retrospective study included 128 patients diagnosed with CIN2 without treatment. The results showed that non-16 hrHPV-positive infection (odds ratio [OR] = 5.4) predicted the spontaneous regression of CIN2 during a 25-month median follow-up [20] . The CIN2 lesion disappeared or regressed in 44 (75.9 %) patients with non-16 hrHPV, while only 24 (40 %) patients with HPV16 disappeared or regressed. A prospective observational study was conducted on Spanish women diagnosed with CIN2 who were invited for conservative management for 2 years [21] . The results reported that the regression rate was 100 % among CIN2 patients without hrHPV infection, whereas it was only 48.2 % among CIN2 patients with HPV16 and previous high-grade squamous intraepithelial lesions (HSIL). HPV16 and previous HSIL cytology significantly increased the risk of CIN2+. A Japanese cohort study [22] reported that CIN1/2 women infected with HPV subtypes 16, 18, 31, 33, 35, 52 and 58 (20.5 %) had a significantly higher risk of developing CIN3 within 5  years compared with other hrHPV (6.0 %) and low-risk HPV or HPV-negative women (1.7 %). Sykes et al . confirmed that most CIN2 women under 25 years would regress to CIN1− within 2 years with observational management, and HPV16-negativity could help to predict regression [23] . Katarina et al . found that the regression of CIN2 lesions was 82.9 % (68/82) in the absence of HPV16 and 51.1 % (23/45) in the presence of HPV16 among patients aged between 25 and 30 without treatment [24] . This study also concluded that CIN2 lesions with a fully visible SCJ and HPV16-negativity could be recommended with active surveillance for 15 months. Damgaard et al. described an association between progression and HPV type-specific among untreated CIN2 women [25] . The results indicated that the rate of regression was only 26.2 % among HPV16-positive women with high-grade cytology who had the highest risk of persistence or progression. Timely interventions should be considered for these women. All these studies showed that HPV16 testing may further help stratification management in younger CIN2 women. Meanwhile, some researchers have suggested that HPV testing combined with cytology may better identify patients at risk of persistence and progression. Thus, more potential markers associated with lesions could be included in the study. In addition, large prospective studies are warranted to validate the predictive value of the HPV genotype in CIN2 outcomes.

CIN2 is not an endpoint because it is heterogeneous and moderately reproduces diagnoses. More prudence is needed for CIN2 patients, particularly women of reproductive age, to minimize the risk of infertility and obstetric complications. Although clinical parameters probably influence the spontaneous regression of CIN2 and can provide certain instructions to clinicians in making treatment decisions [76] , biomarkers for predicting CIN2 regression are needed to prevent overtreatment, especially when CIN2 patients are of reproductive age. The natural history of CIN2 is commonly determined by the interplay of viral factors, host factors and other factors, which influence the outcome of CIN2: regression, persistence, or progression. Therefore, we summarized that these biomarkers correlated with HPV viral factors, host factors and other factors. The most common hrHPV subtype is HPV16, which is regarded as having the greatest impact on the progression of cervical lesions and cervical cancer. Epigenetic changes have been observed in host DNA and hrHPV DNA induced by persistent infection with hrHPV. An effective immune response would clear the hrHPV infection before cell deregulation when women are infected with hrHPV. HLA genotypes and differences in immunoreactive cells are associated with the disease’s prognosis in different individuals. The expression of p16 and Ki-67 could discriminate low-risk lesions from high-risk transformation lesions. The ideal biomarkers would be highly associated with the CIN2 prognosis. Current studies on biomarkers are promising, however, the quality of these studies is uneven. A single biomarker could not offer the predictive value of the CIN2 outcome. It has been reported that the predictive utility of biomarkers was increased according to the HPV genotype combined with cytology, the methylation of the host gene combined with the L1 gene of a specific hrHPV type, the methylation of the host gene with the cytology or with the HPV genotype. We speculate that integrated biomarkers that complement each other will be the trend in CIN2 management. The clinical outcome of CIN2 lesions is unpredictable depending on the histopathological examination. If the predictive value of the biomarkers were valid, CIN2 patients with fertility requirements could be treated conservatively using the results of biomarkers to minimize the risk of infertility and obstetric complications, thereby reducing overtreatment and unnecessary complications. However, there is a real lack of studies with larger samples, longer observation times and consistent outcome indicators. Therefore, it is worth continuing research. Before using relatively reliable predictive biomarkers in clinical practice, it is urgent to employ prospective cohort validation studies with expanded enrolments, longer observational periods and the tracking of more cases to help confirm these conclusions

The World Health Organization called for the elimination of cervical cancer (CC) globally in 2018 [1] . One of the strategies for elimination is widespread screening, which reduces the incidence and mortality of CC by identifying pre-cancer lesions and treating them to prevent invasive cancers [2] . High-risk human papillomavirus (hrHPV) is the major cause of cervical lesions and carcinogenesis, especially persistent infections. The multiple stages of hrHPV-mediated carcinogenesis are now classified as cervical intraepithelial neoplasia (CIN) grade 1 (CIN 1), CIN 2, CIN 3 and invasive cancer according to the histologic diagnosis of the cervical transformation zone. CIN1 lesions could probably regress to no lesion in 60 % of cases, whereas CIN2/3 lesions have a higher proportion of progression into CC. However, the clinical course of CIN2 is unpredictable [3] , [4] . A nationwide cohort study [5] suggested that the rate of regression exceeded 60 % among untreated CIN2 patients. A meta -analysis extracting 36 studies summarized the outcomes of nonpregnant CIN2 patients, who were managed conservatively at different follow-up time intervals [3] . At 24-month time points, 819 of 1,470 untreated CIN2 women (pooled rate: 50 %) regressed to CIN1 or less (CIN1−), 334 of 1,257 women (pooled rate: 32 %) persisted with CIN2, and 282 of 1,445 women (pooled rate: 18 %) progressed to CIN3 or worse (CIN3+). In the analysis of patients under 30, the pooled rates are 60 %, 29 % and 11 %, respectively. The other meta -analysis also estimated the natural history of CIN2/3 in pregnant women [4] . In the subgroup analysis, CIN2 lesions showed pooled rates of regression of 59 %, 40 % persistence and only 1 % progression. Although heterogeneity exists in different types of research and bias in histological classification results, the rate of CIN2 spontaneous regression is still high after conservative management, especially among young women [6] . Therefore, the management of CIN2 patients is debatable. CIN1 lesions with conservative management could mostly regress to normal epithelia. When a biopsy sample is histologically diagnosed with CIN2 or worse (CIN2+), CIN2 is often considered the treatment threshold [7] , [8] . The American Society for Colposcopy and Cervical Pathology (ASCCP) risk-based management consensus recommends immediate treatment without observation in all nonpregnant patients diagnosed with CIN3 and partial CIN2 when the lesion is not incompletely visualised in the squamous-columnar junction (SCJ) or the upper limit or when the endocervical sampling is tested as CIN2+ or ungraded. Only observation will be considered if the CIN2 patient’s concerns about side effects are more about the future pregnancy than cancer. The preferred therapy for CIN2/3 lesions is surgical excision, which is an invasive surgery with side effects. Although untreated CIN2 may be associated with a higher long-term risk of CC than immediate treatment [9] , the treatment of all CIN2+ leads to over-treatment and unnecessary complications. One side effect after treatment is probably the increased risk of subfertility; the other is the risk of premature birth in pregnancies. Preserving fertility is important while the childbearing age is delayed and the multiple-child policy is popular. In addition, the study demonstrated that women with CIN2 were willing to undergo conservative treatment [10] . Thus, if the natural history of the CIN2 lesion is predictable, it will help women with CIN2 select suitable management techniques. Since cytology and histology cannot recognise the development of cervical lesions [11] , it is necessary to develop biomarkers to identify the potential regression of CIN2. The biomarkers might have appropriate potentialities in differentiating CIN2 women who require immediate excision from those who only need close monitoring over time [12] . The natural progression of CIN is influenced by the interplay of human papillomavirus (HPV) viral factors, host factors and other factors ( Fig. 1 ). In this review, the biomarkers ( Table 1 ) associated with HPV infection, host genes and microenvironments for CIN2 regression are summarized. We aim to present an overview of the biomarkers that discern regression from progression among untreated CIN2 lesions, and their application for predicting CIN2 regression is considered. Fig. 1 The natural development of CIN2. The natural development of CIN2 is commonly regulated by the interaction of HPV viral factors (HPV genotype and HPV methylation), host factors (p16/Ki-67 status, host gene methylation effects, human leukocyte antigen subtypes and immune microenvironment) and other factors (vaginal microbiota). CIN1: Cervical intraepithelial neoplasia grade 1. CIN2: Cervical intraepithelial neoplasia grade 2. CIN3: Cervical intraepithelial neoplasia grade 3. HPV: Human papillomavirus. HLA: Human leukocyte antigens. Table 1 Candidate biomarkers for the prediction of CIN2 spontaneous regression. Biomarker Study Number Age Diagnosis Follow-up time Outcome Evaluation indicators Ref HPV viral factors − HPV genotype HPV16 a retrospective study 128 ＜40 HSIL/CIN2 25-month median Disa: no lesion Re: CIN1- Per: CIN2 Pro: CIN3+ Disa/Re: HPV16:24/60 (40.0 %) non-16 hrHPV+: 44/58 (75.9 %) Per/Pro: HPV16: 36/60 (60.0 %) non-16 hrHPV+: 14/58 (24.1 %) 20 HPV16 a prospective study 291 16–64 CIN2 24 months Re: CIN1- Per: CIN2 Pro: CIN3+ Re: HPV16 and HSIL: 26/54 (48.1 %) hrHPV-: 21/21 (100 %) Per/Pro: HPV16 and HSIL: 28/54 (51.9 %) hrHPV-: 0/21 (0) 21 HPV genotype a prospective cohort study CIN1:479; CIN2: 91 18–54 CIN1/2 39.1-month medium Pro: CIN3 Pro: HPV16/18/31/33/35/45/52/58: 20.5 % other hrHPV: 6.0 % low-risk HPV or HPV-: 1.7 % 22 HPV16 a multicenter prospective study 506 <25 CIN2 24 months Re: CIN1- Re: HPV16- and low-grade: 79 % 23 HPV16 a prospective study 127 25–30 CIN2 24 months Re: CIN1- Per: CIN2 Pro: CIN3+ Re: HPV16: 23/45 (51.1 %) Non-16 HPV: 68/82 (82.9 %) 24 HPV16 a historical cohort study − 23–40 CIN2 24 months Per/Pro: CIN2+ Re: HPV16 + and HSIL: 22/84 (26.2 %) 25 

 HPV viral factors − HPV E6/E7 mRNA E6/E7 mRNA a prospective study 108 19–66 CIN2 at least 6 months Re: CIN1- Per: CIN2 Pro: CIN3+ Re: E6/E7 mRNA+: 32/69 (46.4 %); E6/E7 mRNA-: 20/39 (51.3 %) Pro: E6/E7 mRNA+: 18/69 (26.1 %) E6/E7 mRNA-: 2/39 (5.1 %) 31 E6/E7 mRNA a cohort study 42 17–47 CIN2 12 months Re: no lesion Per: CIN1/CIN2 Pro: CIN3 Re: E6/E7 mRNA-: 68.3 % E6/E7 mRNA+: 82.0 % HPV16+: 61.4 % other HPV types or HPV-: 89.5 % 32 

 HPV viral factors − HPV methylation L1 HPV methylation − 15 − CIN1-3 − Re: no lesion Pro: CIN3 L1MR in the progression group was higher than in the regression group. ( p  < 0.05) 37 HPV52 L1 methylation − 54 − CIN1 and CIN2 at least 12 months Re: no lesion Per: CIN1/2 Pro: CIN3 Re: 15.0 % Per/Pro: 35.0 % 38 

 Host factors − Protein expression p16 Ki-67 a retrospective study 122 − CIN2 more than 2 years Re: no lesion Per: CIN2 Pro: CIN3 Pro: p16+: 47/79 (59.5 %) p16-: 12/43 (27.9 %) ki67+: 44/76 (57.9 %) ki67-: 15/46 (32.6 %) 43 p16 a prospective study 96 − HSIL/CIN2 12 months Re: CIN1- Per: CIN2 Pro: CIN3 Re: p16+: 46/81 (56.8 %) p16-: 15/15 (100 %) Pro: p16+: 8/81 (9.9 %) p16-: 0/15 (0) 44 p16 Ki-67 a cross-sectional study 23 17.8–79 CIN2 6–12 months Re: CIN2- Re: p16+: 6/6 (100 %); p16-:17/17 (100 %); ki-67+: 22/22 (100 %); ki-67-: 1/1 (100 %) 45 p16 Ki-67 a prospective study 108 19–66 CIN2 at least 6 months Re: CIN1- Per: CIN2 Pro: CIN3+ Re: p16+: 34/82 (41.5 %) p16-: 18/26 (69.2 %) Ki-67＜5: 6/8 (75.0 %) Ki-67＞25: 17/46 (37.0 %) Pro: p16+: 20/82 (24.4 %) p16-: 0/26 (0) Ki-67＜5: 0/8 (0) Ki-67＞25: 15/46 (32.6 %) 31 p16 − 42 18–61 CIN2 with hrHPV+ at least 12 months Failed regression: lesions Failed regression of CIN2: p16: HR=1.15 ( p  = 0.63) 46 p16 Ki-67 a retrospective study 60 < 40 years CIN2 2 years Disa: no lesion Re: CIN1 Per: CIN2 Pro: CIN3+ Disa/Re: p16+: 34/57 (59.6 %) p16-: 2/3 (66.7 %) Ki-67+: 26/43 (60.5 %) Ki-67-: 6/11 (54.5 %) HPV16+: 15/32 (46.9 %) Low-risk HPV and non-16 hrHPV: 21/28 (75 %) Per/Pro:p16+: 23/57 (40.4 %) p16-: 1/3 (33.3 %) Ki-67+: 17/43 (39.5 %) Ki-67-: 5/11 (45.5 %) HPV16+: 17/32 (53.1 %) Low-risk HPV and non-16 hrHPV: 7/28 (25.0 %) 47 p16 Ki-67 a retrospective cohort study 56 − CIN2 with hrHPV-positive 24 months Re: CIN1- Re: weak p16: 36/56 (64.3 %) weak Ki-67: 14/56 (25.0 %) 48 

 Host factors − Host gene methylation S5 a prospective cohort study 149 25.9–27 CIN2 24 months Re: no lesion Per: CIN1/CIN2 Pro: CIN3+ S5: Re vs Pro (OR = 3.39) 49 FAM19A4/miR124-2 a multicenter prospective study 114 18–55 CIN2/3 lesion was 50 % of the visible cervix 24-month Re: CIN1- CIN2 Pro: CIN3 + or AIS CIN3 Pro: Ca or AIS Re: FAM19A4/miR124-2-negative (74.7 %) FAM19A4/miR124-2-positive (51.4 %) FAM19A4/miR124-2-negative with abnormal cytology (88.4 %) FAM19A4/miR124-2-negative with HPV16-negative (85.1 %) 50 GynTect a multicenter prospective study CIN2: 24; CIN3: 36 18–29 CIN2/3 CIN2: 24 months CIN3: 12 months CIN2/CIN3 Re: CIN1-/CIN2- Per: CIN2/CIN3 Pro: CIN3+ GynTect® test negative: CIN2 regression: 12 of 18 (66.7 %) CIN3 regression: 15 of 27 (55.6 %) 51 

 Host factors − HLA genotype HLA-DRB1*13:02 a prospective study 454 19–54 LSIL and CIN1-2 a mean follow-up of 39.0 months Pro: CIN3 Pro: DRB1*13:02+: 0/47 (0) p  < 0.03 DRB1*13:02-: 39/407 (9.6 %) 54 HLA-DRB1*13:02 a prospective cohort study 454 − LSIL 10 years Pro: CIN3 10-year pro: DRB1*13:02+: 1/47 (2.1 %) p  < 0.03 DRB1*13:02-: 57/407 (14.0 %) 55 HLA*A201 a prospective cohort study 100 18–67 CIN2/3 over 15 weeks Re: CIN1- Per: CIN2/3 Pro: CIN3+ Re: HLA*A201-: 42.3 % HLA*A201 + with non-16 HPV: 14.3 % 56 

 Immune microenvironment specific CD4 + T-cell a cohort study 14 − HPV16 and HSIL 15 weeks Re: no lesion specific CD4 + T-cell immune response was significantly higher in the regression group than in the non-regression group ( p  < 0.05). 60 local immune response − 55 19–49 CIN2-3 12 weeks Re: CIN1- Re: CD4+/CD25+: high: 7/12 (58.3 %); low: 3/42 (7.1 %) CD138+: high: 3/36 (8.3 %); low: 6/16 (37.5 %) CD8+: high: 6/18 (33.3 %); low: 4/37 (10.8 %) 61 local immuneresponse − 53 19–49 CIN2-3 12 weeks Re: CIN1- Re: CD8+/CD25+: high: 7/18 (38.9 %); low: 3/37 (8.1 %) pRb+, deep layer: high: 10/17 (58.8 %); (low): 0/37 (0) p53+, deep layer: high: 8/20 (40.0 %); low: 3/34 (8.8 %) 62 pRb, CIN-lesion, CD4+ a prospective cohort study 162 25–40 CIN2-3 16 weeks Re: CIN1- Re: pRb > 30 % (30 %) pRb > 30 %, a CIN-lesion < 2.5 mm and CD4+ ≤195 (53 %) 63 CD4 + cells infiltration a cohort study 115 23–49 CIN2 3.9–60 months Re: no lesion Per: CIN1/CIN2 Pro: CIN3 Re: HPV16- patients with high CD4 infiltration: 78 % at 60 months 64 FoxP3 − 96 18–58 CIN2 1–27 months − High expression of FoxP3 associated with the CIN2 progression. 65 TLRs − 63 16.3–25 CIN2 3 years Re: no lesion Per: CIN1/CIN2 The levels of TLR2 and TLR7 are significantly higher in women with CIN2 regression. 66 Blimp–1 − 69 <25 CIN2 with hrHPV+and p16+ 2 years Pro: CIN3+ Progression was associated with high Blimp-1 in the epidermis. 67 

 Other factors − Vaginal microbiota Lactobacillus spp. − 87 16–26 CIN2 2 years Re: no lesion Re: Lactobacillus -dominant: 45/57 (78.9 %) Lactobacillus spp.-depleted: 18/30 (60.0 %) 75 CIN2: Cervical intraepithelial neoplasia grade 2. HPV: Human papillomavirus. HLA: Human leukocyte antigens. FAM19A4: Family with sequence similarity 19 (chemokine (C–C motif)-like) member A4. miR124-2: MicroRNA 124-2. CD4: Cluster of differentiation 4. CD8: Cluster of differentiation 8. CD25: Cluster of differentiation 25. CD138: Cluster of differentiation 138. FoxP3: Forkhead box protein P3. TLRs: Toll-like receptors. Ref: Reference. Disa: Disappearance. Re: Regression. Per: Persistence. Pro: Progression. The natural development of CIN2. The natural development of CIN2 is commonly regulated by the interaction of HPV viral factors (HPV genotype and HPV methylation), host factors (p16/Ki-67 status, host gene methylation effects, human leukocyte antigen subtypes and immune microenvironment) and other factors (vaginal microbiota). CIN1: Cervical intraepithelial neoplasia grade 1. CIN2: Cervical intraepithelial neoplasia grade 2. CIN3: Cervical intraepithelial neoplasia grade 3. HPV: Human papillomavirus. HLA: Human leukocyte antigens. Candidate biomarkers for the prediction of CIN2 spontaneous regression. CIN2: Cervical intraepithelial neoplasia grade 2. HPV: Human papillomavirus. HLA: Human leukocyte antigens. FAM19A4: Family with sequence similarity 19 (chemokine (C–C motif)-like) member A4. miR124-2: MicroRNA 124-2. CD4: Cluster of differentiation 4. CD8: Cluster of differentiation 8. CD25: Cluster of differentiation 25. CD138: Cluster of differentiation 138. FoxP3: Forkhead box protein P3. TLRs: Toll-like receptors. Ref: Reference. Disa: Disappearance. Re: Regression. Per: Persistence. Pro: Progression.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.