Over Expressed TKTL1, CIP-2A, and B-MYB Proteins in Uterine Cervix Epithelium Scrapings as Potential Risk Predictive Biomarkers in HR-HPV-Infected LSIL/ASCUS Patients

High oncogenic risk human papillomaviruses (HR-HPVs) promote cervical carcinoma development, the fourth most common feminine cancer. A slow oncodevelopmental phase—defined histopathologically as Cervical Intraepithelial Neoplasia (CIN) grades 1–3, or cytologically as Low- or High-grade Squamous Intraepithelial Lesions (LSIL or HSIL)—precedes the malignancy. Cervical carcinoma screenings through HR-HPV genotyping and Pap smears are regularly performed in Western countries. Faulty cytology screening or genotyping or patients' non-compliance with follow-ups can let slip an oncoprogression diagnosis. Novel biomarker tests flanking HR-HPV genotyping and cytology could objectively predict the risk of disease progression thus helping triage LSIL/ASCUS patients. Here, anonymized leftovers of fresh cervical epithelium scrapings from twice (LSIL/ASCUS and HR-HPV DNA)-positive and twice (Pap smear- and HR-HPV DNA)-negative (control) patients in a proteome-preserving solution served to assess the biomarker worth of three cervical carcinoma-related proteins, i.e., B-MYB (or MYBL2), Cancerous Inhibitor of PP2A (CIP-2a), and transketolase-like1 (TKTL1). Leftovers anonymity was strictly kept and storage at −80°C, protein extraction, immunoblotting, and band densitometry were blindly performed. Only after tests completion, the anonymous yet code-corresponding HR-HPV-genotyping and cytology data allowed to assign each sample to the twice-positive or twice-negative group. Descriptive statistics showed that the three proteins levels significantly increased in the twice-positive vs. twice-negative scrapings. Diagnostic ROC curve analysis identified each protein's Optimal Decision Threshold (OTD) showing that TKTL1 and CIP-2a are stronger risk predictive biomarkers (Sensitivity, 0.91–0.93; Specificity, 0.77–0.83) than B-MYB. Logistic Regression coupled with Likelihood-Ratio Tests confirmed that a highly significant relation links increasing TKTL1/CIP-2a/B-MYB protein levels in twice-positive cervical scrapings to the risk of HR-HPV-driven oncoprogression. Finally, a 3 year clinical follow-up showed that 13 patients (50% of total) of the twice-positive group with biomarker values over OTDs compliantly underwent scheduled colposcopy and biopsy. Of these, 11 (i.e., 84.7%) received a positive histological diagnosis, i.e., CIN1 (n = 5; 38.5%) or CIN2/CIN2+ (n = 6; 46,2%). Therefore, TKTL1/CIP-2a/B-MYB protein levels could objectively predict oncoprogression risk in twice (HR-HPV- and Pap smear)-positive women. Further studies will assess the translatability of these findings into clinical settings.

Each Western country addresses this somewhat murky diagnostic situation differently. The Italian Association of Cervical Screening Programs (GISCI) binding guidelines emphasize the need for using first the HR-HPV genotyping followed, if positive, by the cytology test and again, if positive, by colposcopy; negative women should be retested after 1-3 years (26,27). However, it would be both scientifically and clinically helpful to rely also upon specific biomarkers objectively predicting the actual risk of progression for any cytological precancerous lesion in HR-HPV DNA-positive and Pap smear-positive patients (22,28). Therefore, the present work aimed at assessing the potential worth of three proteins-i.e., B-MYB (or MYBL2), Cancerous Inhibitor of PP2A (CIP-2a), and transketolase-like1 (TKTL1)-as risk predictive biomarkers of disease progression. We chose these proteins because they are over expressed in advanced dysplastic lesions and cervical carcinomas [(13, 29-35); see the Discussion for more details]. We used as starting materials, under strictly anonymized conditions, the leftovers of cervical epithelium scrapings from HR-HPV DNA-positive and LSIL/ASCUS diagnosed (i.e., twice-positive) patients and, as controls (Ctr), scraped leftovers from HPV DNA-negative and Pap smearnegative (i.e., twice-negative) patients. Our herein reported biochemical results integrated by the results of a subsequent 3-year follow-up of the compliant fraction (50%) of the same twice-positive patients indicate that increases in TKTL1, CIP-2a, and B-MYB proteins above their respective Optimal Decision Threshold (ODT) values may robustly predict the risk of an active HR-HPV-driven oncogenesis.

Samples
Following the suggestions of European Union guidelines, population-based organized cervical screening programs have been ongoing since 1998 in the Veneto and Trentino-Alto Adige Regions (North-East Italy) (36)(37)(38). Every 3 or 5, years women aged 19-64 are invited to undergo Pap smear cytology and HPV-DNA genotyping at no personal expense. The screening procedures are organized and overseen by the Local Healthcare Units. The screening protocol follows the GISCI guidelines (26) and a Health Technology Assessment Report on HPV DNA-based primary screening (27). Besides the official letter of invitation, the eligible women receive an a leaflet with information about HPVs, the HR-HPV test, and the screening program. If women do not act in response to the first invitation, they receive a second informing reminder by mail. Women coming of their own accord to the Hospital test site sign the informed consent prior to the sampling of their cervical cells. The collection of the anonymized leftover samples of human cervical epithelial cells took place in Veneto and Trentino Hospitals. Given the particular study type ethical review and approval was not required as per local legislation. Just after sampling, the cervical leftovers were dipped into a freezing solution (90% v/v fetal bovine serum (FBS) + 10% v/v DMSO), coded to anonymize them, and next stored at −80 • C until brought to our lab where they received the second coding. Preliminary investigations (not detailed here) led to choose the FBS+DMSO freezing solution as it preserved the integrity of the cervical cells' proteome. Several other commercial solutions evaluated in parallel, which contained organic and fixative solvents (e.g., Thin Prep TM Hologic USA, Cytofast TM Hospitex International Italy, etc.), could not preserve the cells' proteome integrity. HR-HPV DNA genotyping and Pap smear cytology tests were carried out on portions of the original scrapings having the same code as their leftover counterparts at the laboratory services of the Hospitals involved. However, information about the age, HR-HPV DNA genotype(s), and Pap smear diagnosis was kept strictly secret until all the leftovers had undergone the protein immunoblotting tests and their densitometric evaluations. Leftovers with an inadequate protein amount were directly cast out. Leftovers whose HR-HPV DNA genotype could not be retrieved were also disposed of. Leftovers with coexisting infectious conditions other than HR-HPVs were too left out. However, the data from six pilot samples with a cytological diagnosis of inflammation but with a negative result of a complete TVAG (Trichomonas, Candida, Bacterial vaginosis, Chlamydia, and Mycoplasma) test were kept apart as specificity controls. Eventually, 16 both PAP smear-negative and HR-HPV-DNA-negative ["twice-negative" or control (CTR)] samples, and 26 LSIL/ASCUS and HR-HPV-DNA positive ("twice-positive") samples constituted the main populations studied. Their mean age was for both groups was 34.5 years with a range from 16 to 56 years. Cytological diagnoses proportions were: LSIL, 2/3 and ASCUS, 1/3 of the total cases. LSIL and ASCUS data were pooled together because of the limited numbers of the latter. Moreover, ASCUS HPV-DNA-positive patients were categorized together with LSIL ones because GISCI's guideline indicates that the ASCUS diagnostic category, being borderline, must be avoided or limited to a minimum, and the relative samples classified as clearly as possible as LSIL or negative (26,27). No samples from patients diagnosed as positive for the HR-HPV molecular test and HSIL or ASC-H were obtained as they were directly referred to colposcopy.

Samples Lysis
This procedure and the subsequent Western immunoblotting was carried out on the anonymized leftover samples blindly, i.e., with no knowledge of their cytological and HR-HPV DNA characteristics. Frozen leftover samples were fast thawed at 45 • C and next spun at 200 × g for 10 min at 4 • C. The resulting pellets were resuspended in 1.0 ml of phosphate buffer saline (PBS) added with complete EDTA-free protease inhibitor cocktail (Roche, Milan). The cells' morphological integrity was checked under an inverted phase-contrast light microscope (IM35, Zeiss). Next, after three washings with PBS + antiproteolytic cocktail, cell pellets were homogenized in T-PER TM tissue protein extraction reagent (Thermo Scientific, Rockford, USA) also added with the complete EDTA-free protease inhibitor cocktail. Finally, to achieve total cell lysis samples were subjected to one freeze/thaw cycle at −20 • C and ice-homogenized with an UltraTurrax TM . The protein contents of the extracts were assayed according to Chiarini et al. (39,40) using bovine serum albumin (BSA) as standard.

Western Immunoblotting (WB)
WB was performed as detailed in Chiarini et al. (39,40). Equal amounts of protein lysates (15 µg for the analysis of TKTL1 and CIP-2a and 20 µg for B-MYB) were heatdenatured for 10-min at 70 • C in an appropriate volume of 1X NuPAGE LDS Sample Buffer supplemented with 1X NuPAGE Reducing Agent (Life Technologies, Italia). In parallel, 10 µg of protein lysate obtained from cervical carcinoma C4-I cells (39,40) served as the positive control (PC). The amount of sample to be loaded was determined for each target protein to produce a linear signal response for both the target protein and the loading control (LC). The samples were next loaded on NuPAGE Novex 4-12% Bis-Tris polyacrylamide gel (Life Technologies Italia). After electrophoresis in NuPAGE MOPS SDS Running Buffers using the Xcell SureLock TM Mini-Cell (Life Technologies Italia) (50 min runtime at 200 V constant), proteins were blotted onto nitrocellulose membranes (0.2 µm) by means of iBlot TM Dry Blotting System (Life Technologies Italia). The membranes were probed with the following rabbit polyclonal primary antibodies: (a) anti-TKTL1 (ab 155662, Abcam, UK), diluted 1:1000 in TBST/1% BSA; (b) anti-B-MYB (ab 191064, Abcam), at the concentration of 0.5 µg ml −1 in TBST/1% BSA; (c) anti-CIP-2a (ab 99518, ABCAM) diluted 1:1000 in TBST/1% BSA; (d) anti-β actin mouse monoclonal antibody (Santa Cruz Biotechnology) which served as loading control (LC) and was used at 1.0 µg ml −1 in TBST/1% BSA. After overnight incubation at 4 • C, the membranes were probed with the secondary, alkaline phosphatase conjugated, donkey antirabbit IgG (Abcam), diluted 1:3000 in TBST or goat antimouse IgG (Thermo) diluted 1:1000 in TBST. BCIP/NBT (cat. # 72091, Sigma) was used as chromogenic blue substrate to detect alkaline phosphatase activity. The membranes were incubated in the substrate solution for 10-min until color development; thereafter, the reaction was stopped by washing the membranes in several changes of bidistilled water. The membranes were scanned in an Odissey TM (LI-COR, Inc. USA) using the 700 channel. A digital image of each western blot was saved in TIFF format. The densitometric integrated intensity values of the specific bands for the B-MYB, CIP-2a, and TKTL1 antigens and of the loading controls (LC) were evaluated by means of the Image Studio TM Software, version 5.2, developed by LI-COR Inc. for WB analysis. The densitometric analysis of the specific protein bands pertaining to each sample was performed on three independent immunoblots and the results for each tested protein were presented as the mean of signal intensities ± SEM. To ease calculations all densitometric values were multiplied by E4. In detail, the "Add Rectangle" tool in Image Studio Software served to draw a rectangular-shaped box around the bands of interest to quantify the target proteins. The "Median background method" allowed to compute the median pixels intensity at the border contouring the band's shape and to subtract the background from the shape intensity. Finally, the "Signal value" for each shape was calculated by determining its "Total signal" and by subtracting from the latter the product of the "Background" multiplied for the "Area." Human Cervical Carcinoma C4-I Cells C4-I cells were plated in 175 cm 2 plastic flasks (Sarstedt S.r.l., Verona, Italy), and incubated at 37 FBS (Lonza AG, Basel, Switzerland), and gentamycin (0.1 mg ml −1 ; Lonza). Before reaching confluence, cultures were split at a ratio of 1:6 after briefly incubating them at 18 ± 2 • C with 0.025% (w/v) trypsin (Sigma-Aldrich). Protein extracts from the cells of six culture flasks were processed as previously detailed (39,40) and served as positive controls for B-MYB, CIP-2a, and TKTL1 proteins.

Cytology and HR-HPV-DNA Genotyping
Pap smear tests were performed on all the cervical epithelium specimens from which the leftovers derived in the laboratories of the involved Hospitals. In parallel, the same scrapings were partly dipped into Thin Prep TM (Hologic USA) to analyze the genotypes of the infecting HPVs using the INFINITY HPV QUAD TM assay developed by AutoGenomics Inc, (Vista, CA, USA), which detects HR-HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68 (41) or COBAS HPV test (Roche Molecular Diagnostics, Pleasanton, CA, USA) which detects HR-HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68. Only after completion of immunoblotting and band densitometric intensity assessments on all the leftover samples and prior to be sent to the Verona lab, the results of the cytological and HR-HPV genotyping and the ages of patients were anonymized and each labeled with the same code as the corresponding leftovers. The pairing of the data with the same code allowed their later assignment to the twice-positive or twicenegative group to carry out the statistical work-up. Therefore, no breach of sample anonymity whatsoever took place during all such procedures.

Statistical Analysis
Descriptive and diagnostic statistical analyses were carried out using the Analyse-it TM software package (www.analyse-it.com). Shapiro-Wilk's test revealed that a normal distribution could not be assumed for most of the data groups. Welch's t-test or Wilcoxon rank Z test (42) were used for the descriptive statistical comparisons of the densitometric integrated intensity results concerning each protein's specific immunoblot bands from twice-negative controls vs. twice (LSIL/ASCUS and HPV DNA)-positive groups. For diagnostic statistics a prevalence of HPV infection of 0.31 was assumed by averaging observations from whole Italy (43,44). The performance of diagnostic tests, i.e., their ability to correctly identify positive and negative cases over a range of medical decision points, was assessed using nonparametric empirical receiver-operating characteristic (ROC) curve analysis (45). The values ± SEMs of the areas under each ROC curve (AUC) distinguished every sample distribution vs. the no discrimination or chance area, the latter having a value of 0.5 (46). Differences between AUC pairs were assessed via DeLong's non-parametric Z test (47). Youden's J index (or Youden's statistic), which optimizes biomarker performance when equal weight is allotted to Sensitivity and Specificity, was calculated by the non-parametric method of Delong et al. (48)(49)(50). The maximum J index value indicated the Optimal Decision Thresholds (ODTs) of the densitometric integrated intensity values for each of the three proteins studied at values of: Sensitivity, 0.933-0.909; and Specificity, 0.833-0.769 (51). Finally, binary Logistic Regression was used to independently confirm that a continuous predictor or explanatory variable did indeed impact on a dependent binary or dichotomous variable (52)(53)(54)(55)(56). The tests used were two-sided wherever possible and the level of statistical significance was a P-value < 0.05. For more details on statistical methods used see the online Supplemental Material.

RESULTS
The present work aimed at assessing the worth of three proteins, i.e., B-MYB, CIP-2a, and TKTL1, as risk predictive biomarkers of HR-HPV-driven precancerous disease progression. Anonymized fresh cervical epithelium scrapings leftovers which were "twicepositive" having a cytological diagnosis of LSIL/ASCUS and being also HR-HPV DNA-positive for 14 different oncogenic subtypes (see Table 1 for details in this regard) comprised the test group. As observed also by others (57), multiple HR-HPV subtypes infections, with HPV-16, -51, -52, and -56 as the more prevalent ones, occurred in 58% of the leftovers ( Table 1). "Twice-negative" (both Pap test-negative and HR-HPV DNA-negative) specimens served as controls. To fully preserve the protein complement of the leftovers an ad hoc freezing medium (90% v/v FBS + 10% v/v DMSO) was devised and thoroughly tested prior to starting the study. The results proved its clear superiority over any other standard commercial solutions containing organic agents or fixatives, e.g., Thin Prep TM (Hologic, USA), Cytofast TM (Hospitex International, Italy), etc., which were also tested in parallel (not shown). All the leftovers were processed blindly,  and their belonging to the twice-positive or twice-negative group became known only after immunoblotting procedures and specific band protein immunoblot densitometry evaluations had been carried out for all the specimens involved. Figure 1 shows typical immunoblot bands of the three putative biomarker proteins studied. The protein bands from twice (Pap test-and HR-HPV)-negative (Ctr) samples are flanked by corresponding ones from twice (LSIL/ASCUS and HR-HPV)positive samples. Typical matching proteins bands from premetastatic human C4-I cervical carcinoma cells (39,40) are also displayed as positive controls (PCs).
In absolute terms, the densitometric integrated intensity values of B-MYB immunoblot bands were the lowest while those of CIP-2a the highest ones. In relative terms, the immunoblot bands mean densitometric values were significantly higher-B-MYB, 4.7-fold (P = 0.0023); CIP-2a, 7.7-fold (P < 0.0001); and TKTL1's 14.6-fold (P < 0.0001)-in the twice (LSIL/ASCUS and HR-HPV)-positive samples than in their twice-negative (control) counterparts (Figure 2; Table 2). Therefore, HR-HPVinfection drove a significant overexpression of the three proteins with intensities decreasing in the order TKTL1 > CIP-2a > B-MYB which could be detected through WB and densitometry evaluations in the LSIL/ASCUS cervical epithelium scrapings.
Receiver operating characteristics (ROC) curves plot the diagnostic accuracy of a putative biomarker across the whole range of potential decision thresholds (45,46). Figure 3 displays the ROC curves pertaining to each of the three proteins studied. In each instance, the mean value of the area under the respective ROC curve (AUC) (47) significantly (P < 0.0001) differed from the value (= 0.5) of the no discrimination or chance area (Figure 3; Table 3). Moreover, the three proteins' AUCs significantly (P < 0.001) differed from each other, and both TKTL1's and CIP-2a's AUC values were significantly (P < 0.001) higher than B-MYB's. This was the first indication that TKTL1 and CIP-2a likely are more robust biomarkers of HR-HPV-driven Precision statistical analysis showed how the values of ROC curve Sensitivity and Specificity and Youden's J index (or Youden's statistic) concomitantly changed for each of the proteins tested along with increasing densitometric values of specific protein bands (Figures 4A-C). Notably, Youden's J index peak value embodies a criterion to choose the Optimum Decision Threshold (ODT)-in this case the specific protein band densitometric integrated intensity values (x E4)-of a dichotomous (0-to-1) diagnostic numeric test (i.e., normal or active oncoprogression) (48,49).
Remarkably Hence, TKTL1 occupied the first rank as candidate biomarker since it exhibited a Youden's J index peak value of 0.833 (at 0.933  Table 2). Sensitivity, 0.833 Specificity, 87.9% Accuracy, with a Likelihood Ratio (+) of 5.60, Predictive Value (+) of 0.97, and Cost of 0.103) ( Table 3) Table 3). Thus, the ODT values of TKTL1 and CIP-2a gained from uterine cervical epithelium scrapings could have a direct clinical significance as they would allow to pinpoint the twice-positive patients with a significant risk of active oncoprogression.
An assessment of how the probability of HR-HPV-driven active oncogenesis changed with increasing densitometric values of B-MYB, CIP-2a, and TKTL1 in the cervical epithelium scrapings and its statistical level of significance was carried out via Logistic Regression (54) combined with Likelihood Ratio Test (LTRs or Effect of model) (55). Figure 5 displays the respective fitted Logistic Regression curves indicating the Log Odds ratio (the Logit) of the probability of the oncogenic outcome in relation to increasing densitometric values of each protein in the cervical leftovers examined. Table 4 reports the respective β 0 and β 1 x Logistic Regression parameters estimates, with their corresponding 95% CIs, and SEs.
The LTR tests demonstrate that in all the three protein instances, the Log Likelihood (or the Deviance from the theoretical best fit) of the Full Model with all the β 0 + β 1 x predictor data was significantly smaller than the Log Likelihood of the Null Model with only the β 0 intercept and no β 0 + β 1 x predictor data ( Table 5). Therefore, the predictor significantly improved the Model fit, meaning that the increasing densitometric values of the three proteins tested in scraped cervical epithelium samples of patients with an ASCUS/LSIL cytology diagnosis could significantly impact on the outcome of active HR-HPV-driven oncogenesis. Hence, though having different degrees of clinical translatability, B-MYB, CIP-2a, and TKTL1 are biomarkers of HR-HPV-driven risk of disease progression which can be detected in cervical epithelium scrapings with a LSIL/ASCUS cytological diagnosis.
Finally, a pilot study involving six HR-HPV DNA-negative cervical epithelial scrapings with a generic cytological diagnosis of inflammation but with a negative result of the complete TVAG (Trichomonas, Candida, Bacterial vaginosis, Chlamydia, and Mycoplasma) test showed that no significant difference could be detected in the expression of TKTL1, CIP-2a, and B-MYB proteins and in ROC curve and Logistic Regression/LRT diagnostic parameters vs. 16 double-negative control samples (data not shown). Though preliminary, these findings further strengthen the specificity of TKTL1 and CIP-2a as robust biomarkers of HR-HPV-driven active oncogenesis detectable in cervical LSIL/ASCUS scrapings.

Clinical Three-Year Follow-Up Results
A total of 13 patients (50%) of twice-positive patients, whose TKTL, CIP-2A, and B-MYB markers values were above corresponding OTDs (Figure 4 Middle and Bottom Panels; Table 3), compliantly underwent a scheduled enrollment colposcopy coupled with cervical biopsy during the following three years in order to discriminate benign from precancerous lesions. In this group, 11 women (84.6%) received a positive histological diagnosis: the most frequent histological upshot was CIN1 (n = 5; 38.5% of total cases), followed by CIN2+ (n = 4; 30.8% of total cases) and by CIN2 (n = 2; 15.4% of total cases). These follow-up histology results strengthen the view that the biomarkers we analyzed in the cervical scrapings are endowed with a risk predictive value of disease progression.  (58). Reportedly, multiple HPV types are found in low-grade cervical lesions, but in highgrade lesions HPV-16 and -18 types prevail (28,57,59). As data in Table 1 show, HP-16, -51, -52, -56, and -18 were in decreasing order the most prevalent subtypes in twice-positive cervical epithelium samples, and the identification of multiple infecting HPV DNA subtypes in the same specimen occurred in 58.3% of the samples. Most important, HR-HPVs can drive the development of cancers in both sexes at the level of vulva, anus, penis, prostate, oropharynx, skin, and neck, particularly in HIV-infected patients (28,(60)(61)(62). Moreover, HPV infections' social impact is striking due to (i) emotional reactions of the patients and family members aroused by the awareness of the HR-HPVs infection and by the reiteration of follow-up screenings; and to (ii) growing health service diagnostic and care costs (63). In HR-or LR-HPVinfected women, cytological screening identifies the presence of LSIL or HSIL or ASCUS lesions (14,16,(64)(65)(66). The triage of LSIL/ASCUS-diagnosed patients is problematic because not only of the huge numbers involved but also of the lack of generally accepted management protocols. In addition, follow-up patients may let slip their tests; and faulty cytology screening results or incomplete or false-negative HR-HPV genotyping (for LR-HPV subtypes no commercial diagnostic kits are available) may fail to spot patients with a HPV-driven oncogenic progression (17,(19)(20)(21)(22)67). To solve this diagnostic problem, a proper and fast triage of ASCUS/LSIL cervical lesions particularly in young women would require a third leg, i.e., the assay of a novel biomarker(s) objectively predicting the risk of an active HR-HPV-driven disease and might pinpoint the patients needing a quick colposcopy referral besides being clinically useful worldwide including the underdeveloped countries (4,11,62).

DISCUSSION
In this regard, several biomarker quests have been undertaken. Co-detection of p16 and Ki67, two well-known surrogate biomarkers of cell-cycle de-regulation, was initially suggested to unmask HPV-induced oncogenic transformation (68,69). We independently devised to directly test the worth as biomarkers of three proteins as HPV-oncogenesis risk beacons in cervical epithelium scrapings from "twice-positive" patients, that is having a LSIL/ASCUS cytological diagnosis and a demonstrated and identified HR-HPV-DNA genotype infection as compared to "twice-negative, " that is both Pap smear testnegative and HR-HPV DNA-negative patients. The expression of the three putative protein biomarkers we evaluated, i.e., B-MYB, CIP-2A, and TKTL1, was known to increase in histopathology sections of CIN2+ and of cervical carcinoma cones (29)(30)(31)(32).  Table 3 (q. v.). In all three panels the value of the area under the curve (AUC) is shown between brackets as well as the no discrimination (chance) line joining the bottom left angle with the top right angle, the area of which is 0.5. In all the three instances the AUC values significantly differ (P < 0.0001) from the no discrimination (chance) area value (see also Table 3). TPF, true positive fraction or Sensitivity. FPF, false positive fraction (1-Specificity). In more detail, B-MYB is a DNA-binding transcription factor, the expression and transcriptional activity of which increase in cells progressing from G1 phase to S phase (or DNA synthesis) of the mitotic cycle. Immunohistochemically, being weakly expressed B-MYB is hard to detect in most normal cervical epithelial cells. Conversely, B-MYB is over expressed and localizes in the cell nuclei of CIN 2+, cervical glandular intraepithelial neoplasia, and invasive cervical carcinoma specimens (13, 29, 33). B-MYB associates with the HPV E7 protein and interacts with phosphorylated retinoblastoma protein (pRb), a tumor suppressor protein, and with p107, a pRb-related protein (13).
In turn, CIP-2a promotes tumorigenesis and, in general, tumor progression by inhibiting the activity of protein phosphatase 2A (PP2A)-a prerequisite for the oncogenic transformation of human cells-and by stabilizing c-MYC in different tumors. CIP-2A is overexpressed in cervical carcinomas as it positively correlates with HPV-16 E6 and E7 proteins expression in CIN3 and cervical cancer tissues and cells (33). CIP-2a expedites the G1/S transition by modulating Cdk1 and Cdk2 activities in a B-MYB-dependent manner. Indeed, B-MYB might be the downstream target of CIP-2a, since CIP-2a silencing inhibited the expression of B-MYB in human cervical SiHa cancer cells (34,35). CIP-2A identification in HPVassociated cancers implies the clinical prominence of this protein as a cancer biomarker and a potential therapeutic target (30).
Finally, TKTL1 is a key enzyme in the pentose phosphate pathway, which plays as a crucial role in the progression of cervical neoplasia (31). TKTL1 expression increases in CIN 2/3 or frank neoplasia stages and in neoplastic HeLa cells (32).
The present findings prove that our method of preserving cervical epithelium scraping leftovers to obtain from them integral protein lysates brings about results agreeing with histopathological data reported in the Literature (29)(30)(31)(32)(33)(34)(35). The procedures we used are simple and the results highly reproducible. Hence, TKTL1, CIP-2a, and B-MYB testing could be fully automatized using an ad hoc equipment and software (this is a current undertaking of our lab). Arguably, such automated procedures could be used in mobile labs crisscrossing underdeveloped countries to screen for HR-HPV-infected women at impending risk of cervical carcinoma. Odd ratios between a categorical dependent outcome variable ("0" or normal vs. "1" or active oncogenesis) and the continuously increasing values of the predictor or independent variable, that is the specific band densitometry values of each protein studied. The binary Logistic Regression parameters (or coefficient) estimates of the three curves are reported in Table 4. The results of the Likelihood Ratio Tests (LRT or Effect of Model) (not shown here: see Table 5 for details) indicate a statistically significant impact of the increasing independent predictor (i.e., densitometry values) values on the dependent outcome "1" (i.e., ongoing HPV-driven oncogenesis).
Previously, the majority cervical carcinoma-related biomarkers were discovered in histological sections of cervical tissue cone biopsies taken at colposcopy. Although such biomarkers may increase the accuracy of the histological diagnosis of cancer, histopathology cannot be the primary methodological approach in cervical carcinoma screenings.  This would be impractical as it could hardly deal with large numbers of LSIL/ASCUS bioptic samples particularly in underdeveloped countries. Conversely, using the very same cervical epithelium scrapings to perform standard cytology, HR-HPV DNA genotyping, and biomarker(s), such as TKTL1 and/or CIP-2a expression levels analysis via standard protein lysates Western immunoblotting and specific band densitometry is a feasible undertaking.
In keeping with the just mentioned Literature, our present findings reveal for the first time that it is feasible to detect an increased expression of TKTL1, CIP-2A, and B-MYB in the cervical epithelium scrapings of HR-HPV DNA-positive women having a LSIL/ASCUS cytological diagnosis. Notwithstanding B-MYB's weaker expression, its statistically significant diagnostic indexes make it a still potentially translatable risk predictive biomarker of HR-HPV-driven disease progression. B-MYB's "weakness" may be due to its role as a transcription factor (33), which does not require its synthesis in huge amounts. Moreover, B-MYB's near absence in twice-negative (control) scraping samples is due to its declining synthesis in the terminally differentiating cells of the higher layers of the cervical epithelium (33).
Strengths of this works are: the novel diagnostic approach using cervical epithelium scraping leftovers; feasible and practical methodology used for protein expression analysis; use of biomarkers of the cervical epithelium as objective indicators of HR-HPV-driven disease; the precise assessment of the ODT values for each biomarker which reduces to a minimum the size of false-positive and false-negative findings; and the matching of the biomarker data with follow up data in 85% of the compliant patients.
Limitations of this work are: grouping of ASCUS and LSIL patients in one category because, according to Italian GISCI directive, LSIL or negative diagnoses have mostly supplanted the ASCUS one; the relatively small numbers of patients examined which warrants further studies and follow-ups with larger cohorts.
In conclusion, our biochemical, statistical and threeyear follow up findings indicate TKTL1, CIP-2A, and B-MYB as potential robust risk predictive biomarkers of an active onco-developmental process driven by HR-HPV subtypes.

DATA AVAILABILITY
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

ETHICS STATEMENT
We considered our study exempt from the above ethics requirements because (1) the anonymized cervical epithelium scrapings leftovers we used for our study would have been otherwise totally destroyed and scientifically wasted; (2) we could strictly maintain the complete anonymity of the same scrapings leftovers along all the procedural steps of our research: no subject can be identified from our data; (3) upon our preliminary verbal query Vicenza's Hospital Ethics Committee members answered that there was no need to submit a research protocol for approval since we would have been using anonymized leftovers and keeping them anonymous; and (4) FDA's regulations which exempt the use of leftovers from ethics requirements are taken as reference indications by Hospital Ethics Committees of Veneto and Trentino (Italy).

AUTHOR CONTRIBUTIONS
AC, DL, and ID equally contributed to design the study and its methodology, to carry out the laboratory work, to collect and interpret the data, and to draft the manuscript. UA and MR supervised the technological aspects of the project and contributed to the interpretation of the data. MR and CE assisted in the clinical supervision and sample collection. UA and CE performed the statistical descriptive and diagnostic analyses of the data and wrote substantial parts of the manuscript. UA revised the manuscript draft. All authors read and approved the manuscript in its present final form.