Table S1 shows type-specific associations of methylation with CIN3/AIS. Table S2 shows the correlation of methylation across CpG sites, by HPV type

Supplemental Table 1: Screening history of diagnosed cancer cases Supplemental Table 2: Outcome by age and baseline histology among women with conservatively managed CIN1/2, CIN2, CIN2/3 biopsy results

In Brief The majority of cervical cancers after Pap test–positive, human papillomavirus test–negative co-tests are symptomatic or clinically apparent.

Background: State-of-the-art cervical cancer prevention includes human papillomavirus (HPV) vaccination among adolescents and screening/treatment of cervical precancer (CIN3/AIS and, less strictly, CIN2) among adults. HPV testing provides sensitive detection of precancer but, to reduce overtreatment, secondary "triage" is needed to predict women at highest risk. Those with the highest-risk HPV types or abnormal cytology are commonly referred to colposcopy; however, expert cytology services are critically lacking in many regions. Methods: To permit completely automatable cervical screening/triage, we designed and validated a novel triage method, a cytologic risk score algorithm based on computer-scanned liquid-based slide features (FocalPoint, BD, Burlington, NC). We compared it with abnormal cytology in predicting precancer among 1839 women testing HPV positive (HC2, Qiagen, Germantown, MD) in 2010 at Kaiser Permanente Northern California (KPNC). Precancer outcomes were ascertained by record linkage. As additional validation, we compared the algorithm prospectively with cytology results among 243 807 women screened at KPNC (2016-2017). All statistical tests were two-sided. Results: Among HPV-positive women, the algorithm matched the triage performance of abnormal cytology. Combined with HPV16/18/45 typing (Onclarity, BD, Sparks, MD), the automatable strategy referred 91.7% of HPV-positive CIN3/AIS cases to immediate colposcopy while deferring 38.4% of all HPV-positive women to one-year retesting (compared with 89.1% and 37.4%, respectively, for typing and cytology triage). In the 2016-2017 validation, the predicted risk scores strongly correlated with cytology (P<.001). Conclusions: High-quality cervical screening and triage performance is achievable using this completely automated approach. Automated technology could permit extension of high-quality cervical screening/triage coverage to currently underserved regions.

Purpose Obesity has been inconsistently linked to increased cervical cancer incidence and mortality; however, the effect of obesity on cervical screening has not been explored. We investigated the hypothesis that increased body mass might decrease detection of cervical precancer and increase risk of cervical cancer even in women undergoing state-of-the-art screening. Methods We conducted a retrospective cohort study of 944,227 women age 30 to 64 years who underwent cytology and human papillomavirus DNA testing (ie, cotesting) at Kaiser Permanente Northern California (January 2003 to December 2015). Body mass index was categorized as normal/underweight (< 25 kg/m2), overweight (25 to < 30 kg/m2), or obese (≥ 30 kg/m2). We estimated 5-year cumulative risks of cervical precancer and cancer by category of body mass index using logistic Weibull survival models. Results We observed lower risk of cervical precancer (n = 4,489) and higher risk of cervical cancer (n = 490) with increasing body mass index. Specifically, obese women had the lowest 5-year risk of precancer (0.51%; 95% CI, 0.48% to 0.54% v 0.73%; 95% CI, 0.70% to 0.76% in normal/underweight women; P trend < .001). In contrast, obese women had the highest 5-year risk of cancer (0.083%; 95% CI, 0.072% to 0.096% v 0.056%; 95% CI, 0.048% to 0.066% in normal/underweight women; P trend < .001). Results were consistent in subgroups defined by age (30 to 49 v 50 to 64 years), human papillomavirus status (positive v negative), and histologic subtype (glandular v squamous). Approximately 20% of cervical cancers could be attributed to overweight or obesity in the women in our study who underwent routine cervical screening. Conclusion In this large, screened population, overweight and obese women had an increased risk of cervical cancer, likely because of underdiagnosis of cervical precancer. Improvements in equipment and/or technique to assure adequate sampling and visualization of women with elevated body mass might reduce cervical cancer incidence.

Background: Current U. S. cervical cancer screening and management guidelines do not consider previous screening history, because data on multiple-round human papillomavirus (HPV) and cytology "co-testing" have been unavailable. Objective: To measure cervical cancer risk in routine practice after successive negative screening co-tests at 3-year intervals. Design: Observational cohort study. Setting: Integrated health care system (Kaiser Permanente Northern California, Oakland, California). Patients: 990 013 women who had 1 or more co-tests from 2003 to 2014. Measurements: 3-and 5-year cumulative detection of (risk for) cervical intraepithelial neoplasia grade 3, adenocarcinoma in situ, and cervical cancer (>= CIN3) in women with different numbers of negative co-tests, overall and within subgroups defined by previous co-test results or baseline age. Results: Five-year >= CIN3 risks decreased after each successive negative co-test screening round (0.098%, 0.052%, and 0.035%). Five-year >= CIN3 risks for an HPV-negative co-test, regardless of the cytology result, nearly matched the performance (reassurance) of a negative co-test for each successive round of screening (0.114%, 0.061%, and 0.041%). By comparison, >= CIN3 risks for the cytology-negative co-test, regardless of the HPV result, also decreased with each successive round, but 3-year risks were as high as 5-year risks after an HPV-negative co-test (0.199%, 0.065%, and 0.043%). No interval cervical cancer cases were diagnosed after the second negative co-test. Independently, >= CIN3 risks decreased with age. Length of previous screening interval did not influence future >= CIN3 risks. Limitation: Interval-censored observational data. Conclusion: After 1 or more negative cervical co-tests (or HPV tests), longer screening intervals (every 5 years or more) might be feasible and safe.

Although guidelines have recommended extended interval cervical screening using concurrent human papillomavirus (HPV) and cytology ("cotesting") for over a decade, little is known about its adoption into routine care. Using longitudinal medical record data (2003-2015) from Kaiser Permanente Northern California (KPNC), which adopted triennial cotesting in 2003, we examined adherence to extended interval screening. We analyzed predictors of screening intervals among 491,588 women undergoing routine screening, categorizing interval length into early (< 2.5 years), adherent (2.5 < 3.5 years), or late (3.5 < 6.0 years). We also examined repeated early screening in a subgroup of 50,691 women. Predictors examined included: cohort year (defined by baseline cotest, 2003-2009), race/ethnicity, and baseline age. Compared to the 2003 cohort, women in the 2009 cohort were significantly less likely to screen early (aOR = 0.22, 95% CI = 0.21, 0.23) or late (aOR = 0.47, 95% CI = 0.45, 0.49). African American (AA) and Hispanic women were less adherent overall than Non-Hispanic White women, with increased early [(AA: aOR = 1.21, 95% CI = 1.17, 1.25) (Hispanic: aOR = 1.08, 95% CI = 1.06, 1.11)] and late screening [(AA: aOR = 1.23, 95% CI = 1.19, 1.27) (Hispanic: aOR = 1.06, 95% CI = 1.03, 1.08)]. Asian women were slightly more likely to screen early (aOR = 1.03, 95% CI = 1.01, 1.05), and less likely to screen late (aOR = 0.92, 95% CI = 0.90, 0.94). Women aged 60-64 years were most likely to screen early for two consecutive intervals (aOR = 2.09, 95% CI = 1.91, 2.29). Our study found that widespread and rapid adoption of extended interval cervical cancer screening is possible, at least in this managed care setting. Further research examining multilevel drivers promoting or restricting extended interval screening across diverse healthcare settings is needed.

Abstract Purpose: Human papillomavirus (HPV) DNA methylation testing is a promising triage option for women testing HPV positive during cervical cancer screening. However, the extent to which methylation indicates precancer for all 12 carcinogenic HPV types has not been evaluated. Experimental Design: In this nested case–control study, we tested up to 30 cases of precancer [cervical intraepithelial neoplasia grade 3 (CIN3)/adenocarcinoma in situ (AIS)] and 30 normal controls for each carcinogenic type (single infections with 16/18/31/33/35/39/45/51/52/56/58/59). Next-generation bisulfite sequencing was performed on CpG sites within the L1 and L2 genes. We calculated differences in methylation, ORs, and AUC. Using a fixed sensitivity of 80%, we evaluated the specificity and the risk of CIN3/AIS for best performing CpG sites, and compared the performance of an explorative multi-type methylation assay with current triage strategies. Results: Methylation was positively associated with CIN3/AIS across all 12 types. AUCs for the top sites ranged from 0.71 (HPV51 and HPV56) to 0.86 (HPV18). A combined 12-type methylation assay had the highest Youden index (0.46), compared with cytology (0.31) and a 5-type methylation assay, including only previously described types (0.26). The 12-type methylation assay had higher sensitivity (80% vs. 76.6%) and lower test positivity compared with cytology (38.5% vs. 48.7%). The risk of CIN3/AIS was highest for methylation positives and lowest for cytology or HPV16/18 positives. Conclusions: HPV DNA methylation is a general phenomenon marking the transition from HPV infection to precancer for all 12 carcinogenic types. Development of a combined multitype methylation assay may serve as a triage test for HPV-positive women. Clin Cancer Res; 24(9); 2194–202. ©2018 AACR.

As cervical cancer screening shifts from cytology to human papillomavirus (HPV) testing, a major question is the clinical value of identifying individual HPV types. We aimed to validate Onclarity (Becton Dickinson Diagnostics, Sparks, MD), a nine-channel HPV test recently approved by the FDA, by assessing (i) the association of Onclarity types/channels with precancer/cancer; (ii) HPV type/channel agreement between the results of Onclarity and cobas (Roche Molecular Systems, Pleasanton, CA), another FDA-approved test; and (iii) Onclarity typing for all types/channels compared to typing results from a research assay (linear array [LA]; Roche). We compared Onclarity to histopathology, cobas, and LA. We tested a stratified random sample (n = 9,701) of discarded routine clinical specimens that had tested positive by Hybrid Capture 2 (HC2; Qiagen, Germantown, MD). A subset had already been tested by cobas and LA (n = 1,965). Cervical histopathology was ascertained from electronic health records. Hierarchical Onclarity channels showed a significant linear association with histological severity. Onclarity and cobas had excellent agreement on partial typing of HPV16, HPV18, and the other 12 types as a pool (sample-weighted kappa value of 0.83); cobas was slightly more sensitive for HPV18 and slightly less sensitive for the pooled high-risk types. Typing by Onclarity showed excellent agreement with types and groups of types identified by LA (kappa values from 0.80 for HPV39/68/35 to 0.97 for HPV16). Onclarity typing results corresponded well to histopathology and to an already validated HPV DNA test and could provide additional clinical typing if such discrimination is determined to be clinically desirable.

BACKGROUND Human papillomavirus (HPV)-based cervical cancer screening requires triage markers to decide who should be referred to colposcopy. p16/Ki-67 dual stain cytology has been proposed as a biomarker for cervical precancers. We evaluated the dual stain in a large population of HPV-positive women. METHODS One thousand five hundred and nine HPV-positive women screened with HPV/cytology cotesting at Kaiser Permanente California were enrolled into a prospective observational study in 2012. Dual stain cytology was performed on residual Surepath material, and slides were evaluated for dual stain-positive cells. Disease endpoints were ascertained from the clinical database at KPNC. We evaluated the clinical performance of the assay among all HPV-positive women and among HPV-positive, cytology-negative women. We used internal benchmarks for clinical management to evaluate the clinical relevance of the dual stain assay. We evaluated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the dual stain compared with Pap cytology. All statistical tests were two-sided. RESULTS The dual stain had lower positivity (45.9%) compared with cytology at an ASC-US threshold (53.4%). For detection of CIN2+, the dual stain had similar sensitivity (83.4% vs 76.6%, P = .1), and statistically higher specificity (58.9% vs 49.6%, P < .001), PPV (21.0% vs 16.6%, P < .001), and NPV (96.4% vs 94.2%, P = .01) compared with cytology. Similar patterns were observed for CIN3+. Women with a positive test had high enough risk for referral to colposcopy, while the risk for women with negative tests was below a one-year return threshold based on current US management guidelines. CONCLUSION Dual stain cytology showed good risk stratification for all HPV-positive women and for HPV-positive women with normal cytology. Additional follow-up is needed to determine how long dual stain negative women remain at low risk of precancer.

Abstract Background: Human papillomavirus (HPV) testing has been approved as a primary strategy for cervical cancer screening, either alone or in combination with cytology (co-testing), based on its high sensitivity and long-term reassurance against cervical precancer following a negative test result. However, nearly twice as many women will screen positive for HPV compared with cytology-based screening. Thus, effective management of HPV-positive women requires triage markers to distinguish those at high-risk who should be referred to colposcopy from those with benign infections who can safely return to routine screening. p16/Ki-67 dual stain cytology has previously shown good risk stratification for triage of HPV-positive women; however, studies with follow-up extending beyond 3 years are lacking. We evaluated the long-term risk prediction of p16/Ki-67 for detection of cervical precancer (cervical intraepithelial neoplasia grade 3 or worse, CIN3+) in a large population of HPV-positive women. Methods: 1,588 HPV-positive women screened with HPV/cytology co-testing were enrolled in 2012 at Kaiser Permanente Northern California. p16/Ki-67 cytology was performed on residual Surepath material and slides were evaluated for p16/Ki-67 positivity. Cervical histology endpoints were ascertained from the clinical database with follow-up through 2017. We conducted a Kaplan Meier analysis to estimate risk of CIN3+ by p16/Ki-67 and cytology (atypical squamous cells of undetermined significance or worse, ASC-US+, versus normal cytology). Risks were compared to internal benchmarks for colposcopy referral and for a one year return interval. Results: In women testing p16/Ki-67 positive at baseline, the 2-year risk of CIN3+ was 14.3%, compared with 2.2% in p16/Ki-67-negative women. For ASC-US+, the risk was 12.6% and 2.9% for normal cytology. The 5-year risk of CIN3+ in p16/Ki-67-positive women was 21.6% and 5.0% in p16/Ki-67-negatives. The 5-year risk of ASC-US+ was 17.1% compared to 8.2% for normal cytology. Among p16/Ki-67-negatives, the risk remained below the colposcopy referral threshold for 5 years while in women with normal cytology, the colposcopy referral threshold was crossed after year 3. Conclusion: In the first study evaluating long-term risk stratification of p16/Ki-67 dual staining, p16/Ki-67- negativity provided strong reassurance against CIN3+ for at least five years. In contrast, the risk in women with normal cytology crossed the colposcopy referral threshold after three years. These data support use of p16/Ki-67 for triage of HPV-positive women with the possibility of extending surveillance intervals in p16/Ki-67-negative women. Citation Format: Megan A. Clarke, Barbara Fetterman, Mark Schiffman, Philip E. Castle, Eric Stiemerling, Diane Tokugawa, Nancy Poitras, Walter Kinney, Thomas Lorey, Nicolas Wentzensen. Long term risk prediction of p16/Ki-67 dual stain in triage of HPV-positive women [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2202.

Electronic health-records (EHR) are increasingly used by epidemiologists studying disease following surveillance testing to provide evidence for screening intervals and referral guidelines. Although cost-effective, undiagnosed prevalent disease and interval censoring (in which asymptomatic disease is only observed at the time of testing) raise substantial analytic issues when estimating risk that cannot be addressed using Kaplan-Meier methods. Based on our experience analysing EHR from cervical cancer screening, we previously proposed the logistic-Weibull model to address these issues. Here we demonstrate how the choice of statistical method can impact risk estimates. We use observed data on 41,067 women in the cervical cancer screening program at Kaiser Permanente Northern California, 2003-2013, as well as simulations to evaluate the ability of different methods (Kaplan-Meier, Turnbull, Weibull and logistic-Weibull) to accurately estimate risk within a screening program. Cumulative risk estimates from the statistical methods varied considerably, with the largest differences occurring for prevalent disease risk when baseline disease ascertainment was random but incomplete. Kaplan-Meier underestimated risk at earlier times and overestimated risk at later times in the presence of interval censoring or undiagnosed prevalent disease. Turnbull performed well, though was inefficient and not smooth. The logistic-Weibull model performed well, except when event times didn't follow a Weibull distribution. We have demonstrated that methods for right-censored data, such as Kaplan-Meier, result in biased estimates of disease risks when applied to interval-censored data, such as screening programs using EHR data. The logistic-Weibull model is attractive, but the model fit must be checked against Turnbull non-parametric risk estimates.

Objective To inform impending postcolposcopy guidelines, this analysis examined the subsequent risk of CIN 3+ among women with a grade lower than CIN 2 (< CIN 2) colposcopy results, taking into account the referring results that brought them to colposcopy and cotest results postcolposcopy. Methods We analyzed 107,005 women from 25 to 65 years old, recommended for colposcopy at Kaiser Permanente Northern California. We estimated absolute risks of CIN 3+ among women: (1) recommended for colposcopy (precolposcopy), (2) following colposcopy and with histology results < CIN 2 (postcolposcopy), and (3) with cotest results 12 months after a < CIN 2 colposcopy (return cotest). Results After colposcopy showing < CIN 2 (n = 69,790; 87% of the women at colposcopy), the 1-year risk of CIN 3+ was 1.2%, compared with 6.3% at the time of colposcopy recommendation. Negative cotest results 1 year after colposcopy identified a large group (37.1%) of women whose risk of CIN 3+ (i.e., <0.2% at 3 years after postcolposcopy cotest) was comparable with women with normal cytology in the screening population. These risks are consistent with current guidelines recommending repeat cotesting 12 months after colposcopy < CIN 2 and a 3-year return for women with a negative postcolposcopy cotest. Conclusions Most women are at low risk of subsequent CIN 3+ after a colposcopy showing < CIN 2, especially those who are human papillomavirus–negative postcolposcopy, consistent with current management guidelines for repeat testing intervals. Before the finalizing the upcoming guidelines, we will consider additional rounds of postcolposcopy cotesting.

Abstract Cervical intraepithelial neoplasia grade 2 (CIN2) frequently regresses, is typically slow-growing, and rarely progresses to cancer. Some women forgo immediate treatment, opting for conservative management (heightened surveillance with cytology and colposcopy), to minimize overtreatment and increased risk of obstetric complications; however, there are limited data examining clinical outcomes in these women. We performed a retrospective cohort analysis of younger women diagnosed with initially untreated CIN1/2, CIN2 and CIN2/3 lesions at Kaiser Permanente Northern California between 2003 and 2015. Clinical outcomes were categorized into five mutually exclusive hierarchical groups: cancer, treated, returned to routine screening, persistent high-grade lesion, or persistent low-grade lesion. Median follow-up for the 2,417 women was 48 months. Six women were diagnosed with cancer (0.2%), all with history of high-grade cytology, and none after a negative cotest. Thirty percent of women were treated, and only 20% returned to routine screening; 50% remained in continued intensive follow-up, of which 86% had either low-grade cytology/histology or high-risk human papillomavirus (HPV) positivity, but not necessarily persistence of a single HPV type. No cancers were detected after a single negative cotest in follow-up. Almost half of initially untreated women did not undergo treatment, but remained by protocol in colposcopy clinic for 2 or more years in the absence of persisting CIN2+. Their incomplete return to total negativity was possibly due to sequential new and unrelated low-grade abnormalities. The prolonged colposcopic surveillance currently required to return to routine screening in the absence of persisting CIN2+ might not be necessary after a negative cotest. Significance: Many younger women under conservative management following an initial CIN2 result remain in a clinical protocol of prolonged intensified surveillance without a subsequent diagnosis of CIN2 or more severe diagnoses. More research is needed to determine whether such prolonged management might be unnecessary following a negative cotest for those women with an initial CIN2 but otherwise only low-grade findings. Cancer Prev Res; 11(3); 165–70. ©2018 AACR.

Abstract Background: The natural history of human papillomavirus (HPV) and the steps leading to cervical cancer are well-known; the steps include infection with one of the 13 carcinogenic HPV genotypes, viral persistence, progression to precancer, and invasion. Cervical screening programs target treatable cervical precancer to prevent cancer mortality and morbidity. HPV infections are very common and only those causing precancer pose a risk of cancer. In addition to HPV genotype, multiple established co-factors can be combined to predict with unparalleled accuracy and precision the broad range of risks for the critical transition from common HPV infection to uncommon cervical precancer. Thus, there are three types of factors predicting risk of precancer: viral (e.g., HPV genotype and viral load), host (e.g., age, race/ethnicity) and behavioral (e.g., oral contraceptive use, smoking, BMI, co-infection with other sexually transmitted agents). We are building a risk prediction model for clinical use that reflects the determinants of HPV natural history. The absolute-risk based model will consider the three possible HPV outcomes: HPV progression, else HPV “clearance” (immune suppression) signifying low risk of subsequent precancer from that infection, else persistence of HPV infection without either progression or clearance (i.e., still unresolved outcome). To estimate these competing risks for all the factors, cofactors and their combinations requires very large cohorts of HPV-infected women. Methods: Our analysis makes use of data from a uniquely large cohort study of HPV-infected women, specifically, the 35,000 HPV-positive women, 30 years or older, from the NCI-Kaiser Permanente Northern California Persistence and Progression cohort study. The median time of follow-up is 3 years (maximum >7 years). Risk predictors already recorded include: woman's age, HPV infection status, HPV genotype, viral load, concurrent cervical cytology result, and the range of behavioral cofactors. We will present at the meeting the steps leading to the final model: 1) univariate, then multivariate, absolute risks of progression, clearance, or persistence for each HPV genotype; 2) the same risks accounting for time to event and loss-to-followup; and 3) the novel statistic mean risk stratification (MRS), which measures how well the model predicts the crucial dichotomous outcome (progression vs. not). MRS identifies which combination of variables, by virtue of frequency of positive results and strength of risk stratification, is most promising in deciding risk-based clinical management (i.e., who needs colposcopic biopsy due to high risk of precancer). We present the univariate absolute risks for HPV genotypes here, but will show the full multivariate proportional hazards and MRS analyses at the conference. Results: Risk of progression (29.4% for HPV16 to 7.2% for HPV68) varied inversely with risk of clearance (60.1% for HPV16 to 81.6% for HPV68), by HPV type. Relatively few (~10%) of infections of any carcinogenic type persisted without progression. The most important univariate cofactors in preliminary analyses are viral load (for HPV16 mainly), woman's age, and concurrent cytology. No behavioral risk factors are especially important. Time to clearance and time to progression did not vary by HPV type, with median time to events of 1.5-2 years. Conclusions: Based on our preliminary results, the fate of most HPV infections is determined within a few years of first detection, based mainly on characteristics of the virus. MRS summarizes the average risk discrimination of the prediction model compared to pre-test probability, permitting estimation of its expected benefit. We hypothesize and will test whether multivariate calculations of absolute risks and the use of mean risk stratification can lead to improved risk-based clinical management of HPV-infected women. Citation Format: Maria Demarco, Noorie Hyun, Hormuzd Katki, Brian Befano, Li Cheung, Tina R. Raine-Bennett, Barbara Fetterman, Thomas Lorey, Nancy Poitras, Julia C. Gage, Phillip E. Castle, Nicolas Wentzensen, Mark Schiffman. Risk model for clinical management of HPV-infected women. [abstract]. In: Proceedings of the AACR Special Conference: Improving Cancer Risk Prediction for Prevention and Early Detection; Nov 16-19, 2016; Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2017;26(5 Suppl):Abstract nr A28.

Background: Efficient tools are needed to stage liver disease before treatment of patients infected with hepatitis C virus (HCV). Compared to biopsy, several studies demonstrated favorable performance of noninvasive multianalyte serum fibrosis marker panels [fibrosis-4 (FIB-4) index] and aspartate aminotransferase (AST)-to-platelet ratio index (APRI), but suggested cutoffs vary widely. Ourobjectivewasto evaluate FIB-4 index and APRI and their component tests for staging fibrosis in our HCV-infected population and to determine practical cutoffs to help triage an influx of patients requiring treatment.Methods: Transient elastography (TE) results from 1731 HCV-infected patients were mapped to an F0-F4 equivalent scale. Each patient's APRI and FIB-4 index were calculated. Areas under the receiver operator curve (AUROCs) and false-positive and false-negative rates were calculated to retrospectively compare the performance of the indices and their component tests.Results: The highest AUROCs for distinguishing severe (F3-F4) from mild-to-moderate (F0-F2) fibrosis had overlapping 95% CIs: APRI (0.77; 0.74- 0.79), FIB-4 index (0.76; 0.73- 0.78), and AST (0.74; 0.72- 0.77). Cutoffs had false-negative rates of 2.7%-2.8% and false-positive rates of 6.4%-7.4% for all 3 markers.Conclusions: AST was as effective as FIB-4 index and APRI at predicting fibrosis. Published cutoffs for APRI and FIB-4 index would have been inappropriate in our population, with false-negative rates as high as 11%. For our purposes, no serum fibrosis marker was sufficiently sensitive to rule-out significant fibrosis, but cutoffs developed for AST, FIB-4 index, and APRI all had specificities of 79.2%-80.3% for ruling-in severe fibrosis and could be used to triage 1/3 of our population for treatment without waiting for TE or liver biopsy.

Abstract Objectives: Primary HPV testing has been approved in the United States. Screening intervals can be safely extended for HPV-negative women, but the challenge lies in discriminating transient HPV infections from precancers among HPV-positives. In a large study at Kaiser Permanente Northern California (KPNC), candidate strategies for triage of HPV-positive women are being evaluated, including cytology, p16/Ki-67 dual stain (DS, CINtec PLUS) and HPV16/18 genotyping (cobas). Methods: Over 13,000 HPV-positive women participating in cervical cancer screening at KPNC were enrolled. The dual stain and HPV genotyping assays were implemented and conducted at KPNC. Baseline results for 7,124 women are available and detection of CIN3 is currently evaluated only in cytology-positive women. Results: Among 7,124 HPV-positive women, 4,107 (57.7%) were cytology-positive (ASC-US+), 3,056 (42.9%) were DS-positive, and 1,406 (19.7%) were positive for HPV16 or HPV18. Among all 3,017 HPV-positive, cytology-negative women, 911 (30.2%) were DS- positive, and 508 (16.8%) were positive for HPV16 or HPV18. Of 315 CIN3 detected so far, 280 (88.9%) were DS-positive and 176 (55.9%) were positive for HPV16 or HPV18. Major triage strategies including cytology alone, DS alone, as well as combinations of cytology and genotyping, and DS and genotyping will be presented at the meeting. Discussion: In one of the largest clinical implementation studies of triage strategies, we showed that primary HPV screening followed by DS can reduce colposcopy referral compared to HPV-cytology co-testing while achieving high sensitivity. Additional follow-up is underway to evaluate the programmatic performance of several major candidate strategies. Citation Format: Nicolas A. Wentzensen, Barbara Fetterman, Renee Bremer, Philip Castle, Diane Tokugawa, Nancy Poitras, Elizabeth Hosfield, Thomas Lorey, Mark Schiffman, Walter Kinney. Evaluation of p16/Ki-67 dual stain, cytology, and HPV16/18 genotyping for triage of HPV-positive women in a large screening population [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5296. doi:10.1158/1538-7445.AM2017-5296

Inexpensive and easy-to-perform human papillomavirus (HPV) tests are needed for primary cervical cancer screening in lower-resource regions. In a convenience sample of 516 residual exfoliative cervical specimens from the Kaiser Permanente Northern California and U.S. National Cancer Institute Persistence and Progression Study, we assessed the agreement and clinical performance of a simple, inexpensive real-time PCR assay for the detection of 13 carcinogenic HPV types (the H13 assay; Hybribio, Hong Kong) that is marketed in limited-resource settings compared to previous testing by the Hybrid Capture 2 assay (HC2; Qiagen, Germantown, MD) and the Onclarity assay (BD Diagnostics, Sparks, MD). The test set was chosen to include many HPV-positive specimens. The reference standard was a combination of HC2 and Onclarity results for HPV detection and histologic diagnosis of controls (less than cervical intraepithelial neoplasia grade 2 [<CIN2]) or cases (cervical intraepithelial neoplasia grade 2 or higher [CIN2+]) for disease status. In this enriched convenience sample, H13 tested positive for 94.4% of the 108 HC2- and Onclarity-positive CIN2+ specimens and negative for 88.2% of the 51 HC2- and Onclarity-negative <CIN2 specimens. H13 positivity was significantly lower than that of HC2 among women with CIN2+ (89.9% versus 98.7%, respectively) (P < 0.001) and <CIN2 (53.5% versus 72.4%, respectively) (P < 0.001). In conclusion, H13 corresponds well to the combination of HC2 and Onclarity and has good clinical accuracy compared to histologic diagnosis, with less cross-reactivity with untargeted HPV types than HC2. H13 is a lower-cost HPV DNA test that might be useful for primary screening in limited-resource settings.

For cost-effectiveness and efficiency, many large-scale general-purpose cohort studies are being assembled within large health-care providers who use electronic health records. Two key features of such data are that incident disease is interval-censored between irregular visits and there can be pre-existing (prevalent) disease. Because prevalent disease is not always immediately diagnosed, some disease diagnosed at later visits are actually undiagnosed prevalent disease. We consider prevalent disease as a point mass at time zero for clinical applications where there is no interest in time of prevalent disease onset. We demonstrate that the naive Kaplan-Meier cumulative risk estimator underestimates risks at early time points and overestimates later risks. We propose a general family of mixture models for undiagnosed prevalent disease and interval-censored incident disease that we call prevalence-incidence models. Parameters for parametric prevalence-incidence models, such as the logistic regression and Weibull survival (logistic-Weibull) model, are estimated by direct likelihood maximization or by EM algorithm. Non-parametric methods are proposed to calculate cumulative risks for cases without covariates. We compare naive Kaplan-Meier, logistic-Weibull, and non-parametric estimates of cumulative risk in the cervical cancer screening program at Kaiser Permanente Northern California. Kaplan-Meier provided poor estimates while the logistic-Weibull model was a close fit to the non-parametric. Our findings support our use of logistic-Weibull models to develop the risk estimates that underlie current US risk-based cervical cancer screening guidelines. Published 2017. This article has been contributed to by US Government employees and their work is in the public domain in the USA.

Abstract Introduction: We developed risk models for cancer screening cohorts assembled from electronic health records. We applied them to a cohort of 1.4 million women undergoing cervical cancer screening with human papillomavirus (HPV) and Pap “cotesting” at Kaiser Permanente Northern California (KPNC) to develop risk estimates to inform screening guidelines. Methods: Cohorts assembled using electronic health records present 3 challenges that make it inappropriate to use Kaplan-Meier or Cox models. First, the time of disease onset for an individual is unobserved, and falls between screens (interval-censoring). Second, there is also prevalent disease (left-censoring). Third, prevalent disease is not always immediately diagnosed (e.g. in those with negative screening results), and thus some incident disease is actually missed prevalent disease. To address the challenges, we propose a “logistic-Weibull” model that is a logistic regression for prevalent disease and a Weibull survival regression for interval-censored incident disease. We also propose a non-parametric method (no covariates) to check the assumptions of the logistic-Weibull model. We calculate risks of cervical intraepithelial neoplasia grade 3 and cancer (CIN3+) for each possible abnormal cotesting result. As an illustrative example of the biases that can occur in real data, we present the Kaplan-Meier, logistic-Weibull, and non-parametric cumulative risk curves for 34,261 women at KPNC who test Pap negative and HPV positive at enrollment. Results: The non-parametric method estimates 1.99% prevalent risk of CIN3+ and 5.68% 7-year cumulative risk of CIN3+ among women who are Pap-negative/HPV-positive at enrollment. In contrast, the Kaplan-Meier method estimates merely 0.18% prevalent risk of CIN3+, primarily because of prevalent disease not diagnosed at baseline. Furthermore, the Kaplan-Meier method overestimates 7-year cumulative risk as 7.37% because it equates the time of onset with the time of diagnosis, and thus overestimates the hazards of disease onset at later times. In contrast, the logistic-Weibull estimates of 1.87% prevalent risk of CIN3+ and 5.84% 7-year cumulative risk of CIN3+ are close to the risk estimates from the non-parametric method. Discussion: The Kaplan-Meier method provided poor risk estimates while logistic-Weibull model-based risks were close to the risk estimates from the non-parametric method. Our findings support use of the logistic-Weibull models over Kaplan-Meier methods for developing the risk estimates that underlie current U.S. cervical cancer screening guidelines. Citation Format: Li C. Cheung, Qing Pan, Noorie Hyun, Mark Schiffman, Barbara Fetterman, Philip E. Castle, Thomas Lorey, Hormuzd A. Katki. Risk models for cancer screening cohorts assembled from electronic health records: Application to calculating risks that underlie current cervical cancer screening guidelines [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5300. doi:10.1158/1538-7445.AM2017-5300