p16/Ki67 dual stain triage versus cytology in primary human papillomavirus‐based cervical cancer screening with limited genotyping

The introduction of primary human papillomavirus (HPV) cervical cancer screening requires the implementation of an appropriate triage strategy that will be effective in detecting high‐grade cervical disease without losing diagnostic specificity. From the 30.066 screening tests results, a total of 1086 with available high‐risk human papillomavirus (HRHPV) with limited genotyping, cytology, and p16/Ki67 dual‐stain were selected. Two triage strategies for primary HPV screening were analyzed retrospectively based on the study group. Performance characteristics for p16/Ki67 and cytology triage in the detection of cervical intraepithelial neoplasia grade 2 or worse (CIN2+) and grade 3 or worse (CIN3+) were calculated, detected in colposcopic biopsy. In HPV16/18‐positive cases, primary HPV with p16/Ki67 triage was significantly more specific than cytology (53.1%/16.8% for CIN2+; p < 0.0001; 45.9%/17.0% for CIN3+; p < 0.0001), with yielded sensitivity (95.7%/84.8% for CIN2+; p = 0.0955; 100.0%/87.5% for CIN3+; p = 0.0832). In other HRHPV‐positive cases (N16/N18), p16/Ki67 triage was also significantly higher specific (51.3%/15.3% for CIN2+; p < 0.0001; 44.5%/16.5% for CIN3+; p < 0.0001), with sensitivity (92.3%/74.4% for CIN2+; p = 0.0522; 90.9%/81.8% for CIN3+; p = 0.5637). Diagnostic predictive values were significantly higher for p16/Ki67 triage with the highest PPV in HPV16/18‐positive cases for CIN2+ (45.4%; 95% confidence interval [CI]: 35.2–55.8; p < 0.0001) and very high NPV in all HPV‐positive cases regardless of detected genotype (96.3%–100.0%). The risk (1‐NPV) for CIN3+ in HRHPV16/18‐positive/p16/Ki67‐negative women was 0.0%. Superior diagnostic performance compared to cytology for detecting cervical cancer precursors indicates that p16/Ki67 dual‐immunostain may be a highly effective tool of triage in primary HPV screening with limited HPV 16/18 genotyping in secondary cervical cancer prevention.


| INTRODUCTION
[3][4] Primary HPV screening strategy is the recommended for all countries, regardless of resource settings. 5,6This is due to higher sensitivity of HRHPV testing for the detection of cervical precancers compared to primary cytology-based screening, as well as better reproducibility and substantially lower subjectivity. 7,8The World Health Organization (WHO) has initiated a global call to eliminate cervical cancer as a population burden, and its guidelines released in 2021 recommend using a primary HPV detection for cervical cancer screening. 6,9wever, the implementation of an effective triaging of patients with a positive high-risk human papillomavirus (HRHPV) test result remains an unresolved question and a challenge.Persistent infection of HRHPV types is the etiological factor of most cervical precancers leading to the development of cervical cancer.As it is not possible to differentiate between transient and persistent infection in women with HRHPV positive tests results, there is no doubt that primary HPV screening requires a triage test.Otherwise, too many patients would be referred for colposcopy. 10Different triaging options have been proposed. 11,12Triage testing should be assessed jointly with the primary screening testing, which is to ensure the safety and effectiveness of further clinical management. 11al immunocytochemical staining of cervical cytology specimens using p16 and Ki67 proteins is a morphologic-independent biomarker of a precancerous cervical lesions risk.4][15][16] p16/Ki67 testing is characterized by high sensitivity and high specificity for the detection of CIN2+.9][20][21] p16/Ki67 evaluation by a qualified pathologist has been approved by the Food and Drug Administration (FDA) as a triage method for HRHPVpositive N16/N18 cases in primary HPV screening and for HRHPVpositive N16/N18 NILM women who undergone cotesting. 22The Polish Interim cervical cancer screening guidelines have recommended a wider p16/Ki67 usage during the SARS-CoV-2 pandemic. 4cently, it has also been studied for use in self-sampling, 23 which may open up new possibilities in the management of abnormal screening test results in non-responders in cervical cancer systems of prevention.
The scientific evidence for the use of p16/Ki67 as a triage in HRHPV-positive women undergoing primary HPV screening is limited, and after limited 16/18 genotyping there is very insufficient.
Due to the need to increase the specificity resulting with the better identification of risk groups and improving of the effectiveness of cervical cancer precursors detection in primary HPV screening, we investigated whether primary HPV screening with incorporating p16/Ki67 triage of HPV-positive cases may be an alternative screening strategy for the commonly used cytologic triage.For this purpose, we conducted a retrospective analysis of cytological, virological and immunocytochemical results with histologic correlation, and assessment of the diagnostic performance of the two triage approaches for high-grade squamous intraepithelial lesion (HSIL) detection at the CIN2+ and CIN3+ thresholds (HSIL/CIN2+, HSIL/CIN3+) in primary HPV screening.

| Study design and participants
This is another study in the series, the methodology was described in detail previously. 19,24This retrospective analysis concerns the results of

| HRHPV detection and genotyping
Limited HPV genotyping was performed using the qualitative automated in vitro PCR Abbott RealTime High Risk HPV assay

| Liquid-based screening (LBS)
The cervical samples were collected from all patients once on the SurePath medium (Becton Dickinson) in a typical manner, using Cervex-Brush device (Rovers Medical Devices) according to the manufacturer's procedure, and were the basis for performing all screening tests (LBC, HRHPV, p16/Ki67 testing).Residual cytological material from all cervical samples was stored for 3 months by laboratory in conditions specified by the manufacturer.

| Liquid-based cytology
All liquid-based SurePath cytology samples were processed in the automatic PrepStain Slide system (Becton Dickinson) according to manufacturer's instructions.The cytology slides were reported by a gynecological cytopathologist according to the Bethesda 2014 system.The quality assesment and control procedures for gynecological cytopathology were based on benchmarks published by US laboratories accredited by the College of American Pathologists, and reporting rates in the study were within normal ranges reported. 24screening was amended to include all NILM HPV-positive cases.
Abnormal cytology was defined as atypical squamous cells of undetermined significance (ASC-US) or worse (ASC-US+).

| p16/Ki67 dual immunostaining
Dual immunocytochemical staining of cervical samples were processed in the automatic BenchMark XT laboratory system (Ventana Medical Systems Inc.), using p16 and Ki67 proteins in CINtec PLUS detection kit (Roche, MTM AG laboratories) according to the manufacturer's protocol.A control specimen was present in each run.p16/Ki67 testing was performed from the same sample as cytology was, using residual cellular material stored in laboratory in the original SurePath vials.An immunoprofile evaluation was done by a qualified gynecological cytopathologist, who was blinded to cytology results.p16/Ki67 slides were classified as positive, negative, or unsatisfactory.A positive result was the presence of at least one cell in the slide meeting the following criteria: simultaneous red nuclear stain for Ki67 and brown cytoplasmatic stain for p16 in the same epithelial cell.In the case of the immunoexpression assessment of the cell group, the positive diagnosis was determined by a strong diffuse p16 stain and the presence of at least one cell with nuclear Ki67 staining and cytoplasmic p16 staining seen on the periphery of the cell group or sheet.If no immunostaining or single staining of p16 or Ki67 was noted within the epithelial cells, the slide was classified as negative.p16/Ki67 testing was not performed in cellular residual pellets, in which previously obtained cytology was diagnosed as inadequate.

| Colposcopy and histology
All HRHPV-positive patients with positive p16/Ki67 test result or with abnormal cytology were referred for colposcopy.The management of abnormal screening test results was based on the Polish

| Statistical analysis
The PQStat Software in a 1. and HSIL/CIN3+ were the cut-off points.Additionally, the positivity rate with a normal approximation method used was calculated.
Specificity and sensitivity for p16/Ki67 and cytology triage were compared using the McNemar's chi-square test.For comparison of PPV and NPV a method developed by Leisenring et al. 29 was applied using DTComPair package in R. Differences in diagnostic value between the analyzed triage approaches were evaluated with exact p-values, where p < 0.05 was considered as a significant.

| Study participants and characteristics
The initial study group included patients with three screening tests performed: HRHPV, LBC, and p16/Ki67.Table 2

| Combining primary HPV with limited genotyping with p16/Ki67 versus cytology in age stratification
In the initial group, there were 878 of HRHPV-positive cases, of which HPV 16/18 was positive in 305 cases, including 134 positive p16/Ki67 cases, and 573 positive HRHPV N16N18 cases, including 162 p16/Ki67 positive cases (Table 3).Table 4 shows the results of cytological diagnoses for three age groups (<25, 25-65, and >65 years old) including the size of each group.
The correlation between the cytological diagnosis (NILM, ASC-US, LSIL, ASC-H, HSIL) with the HRHPV status (HPV-, HPV+) and the p16/Ki67 result (DS+, DS-) is presented in PPV % (95% CI) 32.4 (23.9, 42.N16/N18).FDA approval document reported from approximately 10% to 20% higher sensitivity rates (depending on the taken CIN2+ or CIN3+ threshold and HRHPV positivity type) for p16/Ki67 than for cytology triage, similar to our results.The sensitivity of p16/Ki67 immunotesting triage higher than almost 20% compared to cytology as in our investigation (92.3%-95.7% vs. 74.4%-84.8%for CIN2+; 90.9%-100.0%vs. 81.8%-87.5% for CIN3+) were also noted by Wright et al. 20 and by Giorgi Rossi et al. 12 Our study results revealed a relatively low specifitity of triage with cytology (14.3%-16.8%for CIN2+; 16.5%-17% for CIN3+) compared with levels reported, 44.6% for CIN2+ and 42.9% for CIN3+ by Wentzensen et al., 30 76.6% for CIN2+ by Giorgi Rossi et al., 12 or 75.0% for CIN3+ reported by Wright et al. 20 It must be noted that the specificity may have been understimated since the proportion of true negative cytology results in HPV-positive women with NILM cytology was relatively small due to limited indications for colposcopy in that group (especially for N16/N18-positive cases) and concurrently large number of false positive cytology rates in our population.Whilst it should be rememebered a good quality of a gynecological cytopathology in the study. 24r analysis is one of the few large studies evaluating p16/Ki67 diagnostic performance as a triage in HRHPV-positive patients in primary HPV screening setting with limited genotyping for HPV16/ 24.7% vs. 9.0% for HSIL/CIN3+, respectively), regardless of the cut-off point.Similar differences in PPV levels between HRHPVpositivity in p16/Ki67 triage have been reported by others for CIN2+ and CIN3+. 20,22Our results highlight the higher effectiveness of p16/ Ki67 immunotesting incorporated as a single triage tool in detecting cervical precancers than strategies where cytology is used in HPVpositive women who undergone primary HPV screening, regardless of the detected HRHPV type.Though, the highest detection efficiency was observed for p16/Ki67 triage in HPV 16/18-positive women.Moreover, a negative p16/Ki67 test result was associated with a high safety of women with a positive HRHPV and with the lowest of cervical precancers risk in the most oncogenic types of HRHPV, 16 and 18 (1-NPV: 0.00%-3.2%).
Based on high or very high negative predictive value levels, not lower than 96%, obtained in both cut-off points taken, our data confirm high safety of HRHPV-positive patients with a negative p16/ Ki67 test performed as a triage test.Similar estimates were obtained by Wright et al., 20 Wentzensen et al. 30  status and negative p16/Ki67 test result was noted at the level of 0.0% in our study, while in the other study the lowest risk for CIN3+ (0.8%) was observed in HRHPV N16/N18-positive patients. 22reover, very low levels of the negative likelihood ratios obtained in the study (0.08-0.15 for HSIL/CIN2+; 0.00-0.20 for HSIL/CIN3+), that were even two to four-fold lower compared to the others: (0. where Hawaii Medicare reimbursement schedules were available for performed gynecologic and pathologic procedures. 33In turn, no management guidelines for p16/Ki67 use in a public-based screening has been adressed, which was equivalent with lack of pricing and reimbursement schedules for this testing and associated gynecologi- In conclusion, our study showed that the application of p16/Ki67 dual-staining as a triage strategy in women with positive HRHPV test results with limited genotyping has superior diagnostic performance for detecting cervical cancer precursors compared to cytology triage in primary HPV-based cervical cancer screening.Significantly higher specificity of dual-stain triage indicates that this strategy might be associated with a substantial reduce the number of colposcopies, both in HPV 16/18-positive women, as well as in women positive for patients participating in private funds-based opportunistic cervical cancer screening (August 2015-July 2020).All analyzed data come from the electronic registry one of the largest private-based outpatient gynecologic clinics in Lower Silesia in Poland, Corfamed Woman's Health Center (Center).A total of 30.066 screening tests results were analyzed, including 20.605 liquid-based cytologies, 8.331 HPV tests and 1.130 of p16/Ki67 immunostains.It was the initial study group, from which, in the first phase, patients with the performed HRHPV and LBC tests were selected (n = 8331), in the second phase with the additionally performed p16/Ki67 tests (n = 1086), and in the third phase a pre-final group with available histopathology (n = 375).In the prefinal group with colposcopic biopsy results, patients with positive HRHPV results were then selected (n = 352), and it was the final study group.Histopathological diagnoses at the HSIL/CIN2+ and HSIL/CIN3+ thresholds were clinical endpoints of the study.The final group has been retrospectively analyzed along with the diagnostic performance assessment for the primary HPV screening model, with two different triaging approaches, as follows: (1) p16/Ki67 dual-stain testing; (2) cytology.The study population was divided into three age groups (<25, 25-65, and >65 years of age) presented in Table 1.The group of HRHPV-positive patients was divided into two subgroups depending on HRHPV type detected: 16/18 or N16/N18.The primary exclusion criteria were hysterectomy, pregnancy, history of treatment for cervical intraepithelial lesion or cancer, current cancer, missing data, or colposcopy performed outside the Center.The ethics committee approval (ID: 118.6120.36.2023).

(
Abbott Molecular) according to the manufacturer's instructions.The assay was designed to detect 14 HRHPV DNA types with parallel directed HPV 16 and/or 18 genotyping (HPV 16/18), and pooled phenotyping of the remaining 12 non-16 and -18 HRHPV genotypes (HRHPV N16/N18).HPV16/18-positive result was classified as detecting HPV DNA 16 or 18, or both.HRHPV N16/N18 positive was classified as detecting one or more of HRHPV N16/N18 genotypes.To amplify HPV DNA a primer combination consisting of three forward and two reverse primers targeting a conserved L1 region is used, which is also a sequence identity for the HPV classification (family Papillomaviridae, subfamily Firstpapillomavirinae, genus Alphapapillomavirus, species Alphapapillomavirus 9). 25

TRZESZCZ
the extension to American Society for Colposcopy and Cervical Pathology 2012 and 2015 guidelines for cases not covered by the Polish guidelines. 26-28The extended colposcopic protocol used included the endocervical sampling using endocervex brushing and curettage in all cases, targeted biopsy when any abnormal colposcopic findings were found, and a random biopsy from four quadrants in the absence of any abnormal cervical lesions and visualization in the relevant quadrant the new squamocolumnar junction and major screening abnormalities present.The number of sampled biopsies ranged from 1 to 5. The International Federation of Cervical Pathology and Colposcopy 2011 nomenclature was used in colposcopic protocols.The Centre's colposcopists participate in a nationwide Colposcopy 2020 Project for i.a.colposcopy procedure standardization.The LAST 2012/WHO 2014 terminology was used for reporting histologic diagnoses and reviewed by a gynecological pathologist.
6.0 full version (2015 PQStat Statistical Calculation Software) was used for the statistical analysis.A diagnostic value of analyzed screening approaches including primary screening test with secondary test, measured with sensitivity, specificity, positive (PPV) and negative predictive values (NPV), positive (PLR) negative (NLR) likelihood ratios, was calculated according with standard definitions.Histologic results HSIL/CIN2+ shows the number of patients with positive test results in three groups: HRHPVpositive, HRHPV-positive with p16/Ki67-positive, and HRHPVpositive with abnormal LBC.The total number of patients with three tests (HRHPV, LBC, and p16/Ki67) was 1086 of which 878 cases were HRHPV-positive, HRHPV-positive/p16/Ki67-positive were 296 cases, and HRHPV-positive with abnormal LBC were 488 cases.45% (n = 159/352) of women referred for colposcopy were HPV 16/18positive, whereas patients with one or more positive type of HRHPV N16/N18 were 55% (n = 193/352).
cal and pathological procedures by the National Health Fund in Poland.In recently published paper of Harper et al. on costeffectiveness of p16/Ki67 following cotesting with high-risk HPV genotyping, invasive cervical cancer death and costs related to this diagnosis were decreasing, despite increasing costs of screening tests during lifetime. 34We would like to point it out, that our large population-based study demonstrated a superior diagnostic performance of p16/Ki67 triage for detecting cervical cancer precursors in a primary HPV-based cervical cancer screening, which is less expensive option than cotesting analyzed in the referred paper.Significantly increased detection rate for HSIL/CIN2+ and HSIL/CIN3+ combined with decreased number of colposcopies needed to detect of these lesions translate into very good effectiveness of the secondary prevention systems with the biomarker incorporation.This study has several strengths: (1) one of the few large studies evaluating p16/Ki67 diagnostic performance as a triage in HPVpositive patients in primary HPV screening with limited genotyping; (2) one of the first such comprehensive analyses including data from private funds-based opportunistic screening in the Central European population with correlation of virological-cytologicalimmunocytochemical results along with histology; (3) a wellorganized system of management with abnormal screening tests results, which determines good disease ascertainment at all stages of screening and further diagnostics; (4) performing all screening tests from one cervical sampling with a triage testing performed shortly after the visit (which meant short-term storage of residual cervical samples); (5) p16/Ki67 was evaluated by a qualified gynecological cytopathologist; (6) a strict adherence to extended colposcopy protocol used; (7) short interval between abnormal screening tests results and referral to colposcopy allowed immediate histologic correlation (interval not exceeding 3 months).Limitations: (1) this is a retrospective study; (2) loss of patients at various study stages; (3) data in this study comes from a real-life practice, that is not a clinical trial, which further increases heterogeneity and affects the proportions of individual studied subgroups.
Four-level selection of the final study group.
would have a crucial impact on an assessment of cost-effectiveness of introducing p16/Ki67 biomarker triage making this reliable and comparable.As it was in a study ofKilleen et al.,