Brian Befano

Impact of repeatedly screening negative on cervical cancer risk

Abstract Background We demonstrated that cervical cancer risk following any screening result is lower if there is a known prior negative screening history vs an unknown screening history. We extended these findings to look at how screening performs following repeatedly negative screening results. Methods Approximately 1.7 million women aged 30-64 years underwent triennial human papillomavirus (HPV) and cytology co-testing from 2003 to 2021. We modeled 5-year risks of cervical intraepithelial neoplasia grade 3 or more severe diagnoses (CIN3+) and invasive cervical cancer for the initial co-test, and then successive rounds following negative co-testing. A logistic-Weibull prevalence-incidence model was used to model risks. Results HPV test positivity was greater than cytology positivity for only the first co-test, and both positivity rates decreased with each screening round. Diagnostic yields of CIN3+ and cancer declined with each round of screening so the first screen yielded 8-fold more CIN3+ and invasive cancer than the fifth screen following 4 consecutive negative co-tests. Five-year risks of CIN3+ for positive and negative HPV and cytology results, individually or combined, decreased considerably after the first screen, with smaller decreases in each subsequent round. For cancer, we noticed a considerable decrease with the first screen only. Five-year CIN3+ risks were greater for positive HPV or cytology results, with a longer antecedent screening interval and younger age at screening (Ptrend < .001). Conclusions Triennial screening that includes HPV testing becomes inefficient after a single and more so after multiple negative screens. These data support the use of longer screening intervals, especially following negative screen(s).

The Improving Risk Informed HPV Screening (IRIS) Study: Design and Baseline Characteristics

Abstract Background: Cervical cancer screening with high-risk human papillomavirus (HrHPV) testing is being introduced. Most HrHPV infections are transient, requiring triage tests to identify individuals at highest risk for progression to cervical cancer. Head-to-head comparisons of available strategies for screening and triage are needed. Endometrial and ovarian cancers could be amenable to similar testing. Methods: Between 2016 and 2020, discarded cervical cancer screening specimens from women ages 25 to 65 undergoing screening at Kaiser Permanente Northern California were collected. Specimens were aliquoted, stabilized, and stored frozen. Human papillomavirus (HPV), cytology, and histopathology results as well as demographic and cofactor information were obtained from electronic medical records (EMR). Follow-up collection of specimens was conducted for 2 years, and EMR-based data collection was planned for 5 years. Results: Collection of enrollment and follow-up specimens is complete, and EMR-based follow-up data collection is ongoing. At baseline, specimens were collected from 54,957 HPV-positive, 10,215 HPV-negative/Pap-positive, and 12,748 HPV-negative/Pap-negative women. Clinical history prior to baseline was available for 72.6% of individuals, of which 53.9% were undergoing routine screening, 8.6% recently had an abnormal screen, 30.3% had previous colposcopy, and 7.2% had previous treatment. As of February 2021, 55.7% had one or more colposcopies, yielding 5,563 cervical intraepithelial neoplasia grade 2 (CIN2), 2,756 cervical intraepithelial neoplasia grade 3 (CIN3), and 146 cancer histopathology diagnoses. Conclusions: This robust population-based cohort study represents all stages of cervical cancer screening, management, and posttreatment follow-up. Impact: The IRIS study is a unique and highly relevant resource allowing for natural history studies and rigorous evaluation of candidate HrHPV screening and triage markers, while permitting studies of biomarkers associated with other gynecologic cancers.

The development of “automated visual evaluation” for cervical cancer screening: The promise and challenges in adapting deep‐learning for clinical testing

AbstractThere is limited access to effective cervical cancer screening programs in many resource‐limited settings, resulting in continued high cervical cancer burden. Human papillomavirus (HPV) testing is increasingly recognized to be the preferable primary screening approach if affordable due to superior long‐term reassurance when negative and adaptability to self‐sampling. Visual inspection with acetic acid (VIA) is an inexpensive but subjective and inaccurate method widely used in resource‐limited settings, either for primary screening or for triage of HPV‐positive individuals. A deep learning (DL)‐based automated visual evaluation (AVE) of cervical images has been developed to help improve the accuracy and reproducibility of VIA as assistive technology. However, like any new clinical technology, rigorous evaluation and proof of clinical effectiveness are required before AVE is implemented widely. In the current article, we outline essential clinical and technical considerations involved in building a validated DL‐based AVE tool for broad use as a clinical test.

Different human papillomavirus types share early natural history transitions in immunocompetent women

AbstractNecessary stages of cervical carcinogenesis include acquisition of a carcinogenic human papillomavirus (HPV) type, persistence associated with the development of precancerous lesions, and invasion. Using prospective data from immunocompetent women in the Guanacaste HPV Natural History Study (NHS), the ASCUS‐LSIL Triage Study (ALTS) and the Costa Rica HPV Vaccine Trial (CVT), we compared the early natural history of HPV types to inform transition probabilities for health decision models. We excluded women with evidence of high‐grade cervical abnormalities at any point during follow‐up and restricted the analysis to incident infections in all women and prevalent infections in young women (aged <30 years). We used survival approaches accounting for interval‐censoring to estimate the time to clearance distribution for 20 529 HPV infections (64% were incident and 51% were carcinogenic). Time to clearance was similar across HPV types and risk classes (HPV16, HPV18/45, HPV31/33/35/52/58, HPV 39/51/56/59 and noncarcinogenic HPV types); and by age group (18‐29, 30‐44 and 45‐54 years), among carcinogenic and noncarcinogenic infections. Similar time to clearance across HPV types suggests that relative prevalence can predict relative incidence. We confirmed that there was a uniform linear association between incident and prevalent infections for all HPV types within each study cohort. In the absence of progression to precancer, we observed similar time to clearance for incident infections across HPV types and risk classes. A singular clearance function for incident HPV infections has important implications for the refinement of microsimulation models used to evaluate the cost‐effectiveness of novel prevention technologies.

Artificial intelligence–based image analysis in clinical testing: lessons from cervical cancer screening

Abstract Novel screening and diagnostic tests based on artificial intelligence (AI) image recognition algorithms are proliferating. Some initial reports claim outstanding accuracy followed by disappointing lack of confirmation, including our own early work on cervical screening. This is a presentation of lessons learned, organized as a conceptual step-by-step approach to bridge the gap between the creation of an AI algorithm and clinical efficacy. The first fundamental principle is specifying rigorously what the algorithm is designed to identify and what the test is intended to measure (eg, screening, diagnostic, or prognostic). Second, designing the AI algorithm to minimize the most clinically important errors. For example, many equivocal cervical images cannot yet be labeled because the borderline between cases and controls is blurred. To avoid a misclassified case-control dichotomy, we have isolated the equivocal cases and formally included an intermediate, indeterminate class (severity order of classes: case>indeterminate>control). The third principle is evaluating AI algorithms like any other test, using clinical epidemiologic criteria. Repeatability of the algorithm at the borderline, for indeterminate images, has proven extremely informative. Distinguishing between internal and external validation is also essential. Linking the AI algorithm results to clinical risk estimation is the fourth principle. Absolute risk (not relative) is the critical metric for translating a test result into clinical use. Finally, generating risk-based guidelines for clinical use that match local resources and priorities is the last principle in our approach. We are particularly interested in applications to lower-resource settings to address health disparities. We note that similar principles apply to other domains of AI-based image analysis for medical diagnostic testing.

Initial evaluation of a new cervical screening strategy combining human papillomavirus genotyping and automated visual evaluation: the Human Papillomavirus–Automated Visual Evaluation Consortium

Abstract The HPV-Automated Visual Evaluation Consortium is validating a cervical screening strategy enabling accurate cervical screening in resource-limited settings. A rapid, low-cost human papillomavirus (HPV) assay permits sensitive HPV testing of self-collected vaginal specimens; HPV-negative women are reassured. Triage of positive participants combines HPV genotyping (4 groups in order of cancer risk) and visual inspection assisted by automated cervical visual evaluation that classifies cervical appearance as severe, indeterminate, or normal. Together, the combination predicts which women have precancer, permitting targeted management to those most needing treatment. We analyzed CIN3+ yield for each HPV-Automated Visual Evaluation risk level (HPV genotype crossed by automated cervical visual evaluation classification) from 9 clinical sites (Brazil, Cambodia, Dominican Republic, El Salvador, Eswatini, Honduras, Malawi, Nigeria, and Tanzania). Data from 1832 HPV-positive participants confirmed that HPV genotype and automated cervical visual evaluation classification strongly and independently predict risk of histologic CIN3+. The combination of these low-cost tests provided excellent risk stratification, warranting pre-implementation demonstration projects.

Validation of a Lab-free Low-cost Screening Test for Prevention of Cervical Cancer

The purpose of this study is to validate Automated Visual Evaluation (AVE), specifically the CINFinder version developed by DL Analytics, a point-of-care screening and triage diagnostic tool for cervical cancer based on the assessment of digital images through artificial intelligence. Several teams around the world have developed versions of AVE as a triage technology but none as a screening tool.