FLEXIBLE RISK PREDICTION MODELS FOR LEFT OR INTERVAL-CENSORED DATA FROM ELECTRONIC HEALTH RECORDS

Electronic health records are a large and cost-effective data source for developing risk-prediction models. However, for screen-detected diseases, standard risk models (such as Kaplan-Meier or Cox models) do not account for key issues encountered with electronic health record data: left-censoring of pre-existing (prevalent) disease, interval-censoring of incident disease, and ambiguity of whether disease is prevalent or incident when definitive disease ascertainment is not conducted at baseline. Furthermore, researchers might conduct novel screening tests only on a complex two-phase subsample. We propose a family of weighted mixture models that account for left/interval-censoring and complex sampling via inverse-probability weighting in order to estimate current and future absolute risk: we propose a weakly-parametric model for general use and a semiparametric model for checking goodness of fit of the weakly-parametric model. We demonstrate asymptotic properties analytically and by simulation. We used electronic health records to assemble a cohort of 33,295 human papillomavirus (HPV) positive women undergoing cervical cancer screening at Kaiser Permanente Northern California (KPNC) that underlie current screening guidelines. The next guidelines would focus on HPV typing tests, but reporting 14 HPV types is too complex for clinical use. National Cancer Institute along with KPNC conducted a HPV typing test on a complex subsample of 9258 women in the cohort. We used our model to estimate the risk due to each type and grouped the 14 types (the 3-year risk ranges 21.9-1.5) into 4 risk-bands to simplify reporting to clinicians and guidelines. These risk-bands could be adopted by future HPV typing tests and future screening guidelines.

Discovery and validation of candidate host DNA methylation markers for detection of cervical precancer and cancer

Human papillomavirus (HPV) testing has been recently introduced as an alternative to cytology for cervical cancer screening. However, since most HPV infections clear without causing clinically relevant lesions, additional triage tests are required to identify women who are at high risk of developing cancer. We performed DNA methylation profiling on formalin-fixed, paraffin-embedded tissue specimens from women with benign HPV16 infection and histologically confirmed cervical intraepithelial neoplasia grade 3, and cancer using a bead-based microarray covering 1,500 CpG sites in over 800 genes. Methylation levels in individual CpG sites were compared using a t-test, and results were summarized by computing p-values. A total of 12 candidate genes (ADCYAP1, ASCL1, ATP10, CADM1, DCC, DBC1, HS3ST2, MOS, MYOD1, SOX1, SOX17 and TMEFF2) identified by DNA methylation profiling, plus an additional three genes identified from the literature (EPB41L3, MAL and miR-124) were chosen for validation in an independent set of 167 liquid-based cytology specimens using pyrosequencing and targeted, next-generation bisulfite sequencing. Of the 15 candidate gene markers, 10 had an area under the curve (AUC) of ≥ 0.75 for discrimination of high grade squamous intraepithelial lesions or worse (HSIL+) from <HSIL cytology using at least one assay. Overall, SOX1, DCC, and EPB41L3 showed the best discrimination with AUC values of ≥0.80, irrespective of methylation detection assay. In addition to verifying candidate markers from the literature (e.g., SOX1 and EPB41L3), we identified novel markers that may be considered for detection of cervical precancer and cancer and warrant further validation in prospective studies.

Abstract A28: Risk model for clinical management of HPV-infected women

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 &gt;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.

Cervical cancer incidence after screening with HPV, cytology, and visual methods: 18-Year follow-up of the Guanacaste cohort

Testing negative for human papillomavirus (HPV) predicts long-term reassurance against invasive cervical cancer (ICC). To provide realistic estimates of effectiveness for new screening programs, we studied ICC risk after a 7-year repeated multimethod screening effort. In 1993-1994, 10,049 women aged 18-97 years were enrolled into a population-based cohort study of cervical HPV in Guanacaste, Costa Rica. Women were screened at different intervals according to enrollment results. Each visit (mean 3.2, 90% attendance) included split-sample conventional, automated, and liquid-based cytology, visual inspection, cervicography, and PCR-based HPV testing. Abnormal screening led to colposcopy and excisional treatment as appropriate during the study. Referral to colposcopy for HPV in the absence of other findings was introduced only at the last visit. Population-based Costa Rica Cancer Registry linkage identified cohort women diagnosed with ICC in the 18 years following cohort enrollment. The ICC cumulative risk was 0.4% (n = 38); 18 were diagnosed with ICC after study participation. Of these, 9 were missed at the screening step (negative screening or below the referral threshold, refused screening or colposcopy), 5 attended colposcopy but were not diagnosed as CIN2+, and 4 were treated for CIN2/3 but progressed to ICC nonetheless. Decreasing age-standardized ICC rates for the 1993-2011 period were observed in Guanacaste; cohort women showed additional 31% ICC incidence reduction with apparent downstaging of cancers that occurred. ICC risk following negative HPV testing in the optimal age range 30-50 years was extremely low. Real-life screening effectiveness following introduction is lower than the potential near-complete efficacy predicted by HPV natural history.

Evaluation of Type Replacement Following HPV16/18 Vaccination: Pooled Analysis of Two Randomized Trials

Background

Current HPV vaccines do not protect against all oncogenic HPV types. Following vaccination, type replacement may occur, especially if different HPV types competitively interact during natural infection. Because of their common route of transmission, it is difficult to assess type interactions in observational studies. Our aim was to evaluate type replacement in the setting of HPV vaccine randomized controlled trials (RCTs).

Methods

Data were pooled from the Costa Rica Vaccine Trial (CVT; NCT00128661) and PATRICIA trial (NCT001226810)-two large-scale, double-blind RCTs of the HPV-16/18 AS04-adjuvanted vaccine-to compare cumulative incidence of nonprotected HPV infections across trial arms after four years. Negative rate difference estimates (rate in control minus vaccine arm) were interpreted as evidence of replacement if the associated 95% confidence interval excluded zero. All statistical tests were two-sided.

Results

After applying relevant exclusion criteria, 21 596 women were included in our analysis (HPV arm = 10 750; control arm = 10 846). Incidence rates (per 1000 infection-years) were lower in the HPV arm than in the control arm for grouped nonprotected oncogenic types (rate difference = 1.6, 95% confidence interval [CI] = 0.9 to 2.3) and oncogenic/nononcogenic types (rate difference = 0.2, 95% CI = -0.3 to 0.7). Focusing on individual HPV types separately, no deleterious effect was observed. In contrast, a statistically significant protective effect (positive rate difference and 95% CI excluded zero) was observed against oncogenic HPV types 35, 52, 58, and 68/73, as well as nononcogenic types 6 and 70.

Conclusion

HPV type replacement does not occur among vaccinated individuals within four years and is unlikely to occur in vaccinated populations.

Type-dependent association between risk of cervical intraepithelial neoplasia and viral load of oncogenic human papillomavirus types other than types 16 and 18

Studies of the clinical relevance of human papillomavirus (HPV) DNA load have focused mainly on HPV16 and HPV18. Data on other oncogenic types are rare. Study subjects were women enrolled in the atypical squamous cells of undetermined significance (ASC-US) and low-grade squamous intraepithelial lesion (LSIL) triage study who had ≥1 of 11 non-HPV16/18 oncogenic types detected during a 2-year follow-up at 6-month intervals. Viral load measurements were performed on the first type-specific HPV-positive specimens. The association of cervical intraepithelial neoplasia grades 2-3 (CIN2/3) with type-specific HPV DNA load was assessed with discrete-time Cox regression. Overall, the increase in the cumulative risk of CIN2/3 per 1 unit increase in log₁₀ -transformed viral load was statistically significant for four types within species 9 including HPV31 (adjusted hazard ratio [HR _adjusted ] = 1.32: 95% confidence interval [CI], 1.14-1.52), HPV35 (HR _adjusted  = 1.47; 95% CI, 1.23-1.76), HPV52 (HR _adjusted  = 1.14; 95% CI, 1.01-1.30) and HPV58 (HR _adjusted  = 1.49; 95% CI, 1.23-1.82). The association was marginally significant for HPV33 (species 9) and HPV45 (species 7) and was not appreciable for other types. The per 1 log₁₀ -unit increase in viral load of a group of species 9 non-HPV16 oncogenic types was statistically significantly associated with risk of CIN2/3 for women with a cytologic diagnosis of within normal limits, ASC-US, or LSIL at the first HPV-positive visit but not for those with high-grade SIL. Findings suggest that the viral load-associated risk of CIN2/3 is type-dependent, and mainly restricted to the species of HPV types related to HPV16, which shares this association.

Similar Risk Patterns After Cervical Screening in Two Large U.S. Populations: Implications for Clinical Guidelines

OBJECTIVE

To compare the risks of histologic high-grade cervical intraepithelial neoplasia (CIN) or worse after different cervical cancer screening test results between two of the largest U.S. clinical practice research data sets.

METHODS

The New Mexico Human Papillomavirus (HPV) Pap Registry is a statewide registry representing a diverse population experiencing varied clinical practice delivery. Kaiser Permanente Northern California is a large integrated health care delivery system practicing routine HPV cotesting since 2003. In this retrospective cohort study, a logistic-Weibull survival model was used to estimate and compare the cumulative 3- and 5-year risks of histologic CIN 3 or worse among women aged 21-64 years screened in 2007-2011 in the New Mexico HPV Pap Registry and 2003-2013 in Kaiser Permanente Northern California. Results were stratified by age and baseline screening result: negative cytology, atypical squamous cells of undetermined significance (ASC-US) (with or without HPV triage), low-grade squamous intraepithelial lesion, and high-grade squamous intraepithelial lesion.

RESULTS

There were 453,618 women in the New Mexico HPV Pap Registry and 1,307,528 women at Kaiser Permanente Northern California. The 5-year CIN 3 or worse risks were similar within screening results across populations: cytology negative (0.52% and 0.30%, respectively, P<.001), HPV-negative and ASC-US (0.72% and 0.49%, respectively, P=.5), ASC-US (3.4% and 3.4%, respectively, P=.8), HPV-positive and ASC-US (7.7% and 7.1%, respectively, P=.3), low-grade squamous intraepithelial lesion (6.5% and 5.4%, respectively, P=.009), and high-grade squamous intraepithelial lesion (53.1% and 50.4%, respectively, P=.2). Cervical intraepithelial neoplasia grade 2 or worse risks and 3-year risks had similar trends across populations. Age-stratified analyses showed more variability, especially among women aged younger than 30 years, but patterns of risk stratification were comparable.

CONCLUSION

Current U.S. cervical screening and management recommendations are based on comparative risks of histologic high-grade CIN after screening test results. The similar results from these two large cohorts from different real-life clinical practice settings support risk-based management thresholds across U.S. clinical populations and practice settings.

Carcinogenic human papillomavirus infection

Infections with human papillomavirus (HPV) are common and transmitted by direct contact. Although the great majority of infections resolve within 2 years, 13 phylogenetically related, sexually transmitted HPV genotypes, notably HPV16, cause - if not controlled immunologically or by screening - virtually all cervical cancers worldwide, a large fraction of other anogenital cancers and an increasing proportion of oropharyngeal cancers. The carcinogenicity of these HPV types results primarily from the activity of the oncoproteins E6 and E7, which impair growth regulatory pathways. Persistent high-risk HPVs can transition from a productive (virion-producing) to an abortive or transforming infection, after which cancer can result after typically slow accumulation of host genetic mutations. However, which precancerous lesions progress and which do not is unclear; the majority of screening-detected precancers are treated, leading to overtreatment. The discovery of HPV as a carcinogen led to the development of effective preventive vaccines and sensitive HPV DNA and RNA tests. Together, vaccination programmes (the ultimate long-term preventive strategy) and screening using HPV tests could dramatically alter the landscape of HPV-related cancers. HPV testing will probably replace cytology-based cervical screening owing to greater reassurance when the test is negative. However, the effective implementation of HPV vaccination and screening globally remains a challenge.

Trends in cervical cancer incidence in younger US women from 2000 to 2013

OBJECTIVE

This study aimed to assess the temporal trends in invasive cervical cancer (ICC) incidence rates among 21-25year-olds. US guidelines no longer recommend screening prior to age 21, and concerns have been raised that delayed screening initiation may increase ICC incidence among young women.

METHODS

This study utilized ICC incidence data from 18 US population-based cancer registries in SEER from 2000 to 2013 and Pap test prevalence data from the Behavioral Risk Factor Surveillance System from 1996 to 2012. Trends were evaluated with annual percent changes (APCs) using Joinpoint regression.

RESULTS

The prevalence of never having a Pap test before age 21 increased from 22.0% in 1996-2004 to 38.3% in 2006-2012 (APC=+5.48, 95%CI=+4.20, +7.50). Despite this decline in screening, ICC incidence among 21-23year olds significantly declined between 2000 and 13 (APC=-5.36, 95%CI=-7.83,-2.82), particularly from 2006 to 2013 (APC=-9.70, 95%CI=-15.79, -3.17). ICC incidence remained constant among 24-25year olds (APC=+0.45, 95%CI=-2.00, 2.97). Compared to women born in 1978-1985, women born in 1986-1991 had a higher prevalence of never receiving a Pap test prior to 21 (35.4% vs. 22.1%, p<0.001), but a lower ICC incidence at 21-23 (0.98 vs. 1.55 per 100,000, p<0.001).

CONCLUSION

While US females born in 1986-1991 were less likely to receive a Pap test before age 21, diagnoses of ICC in the early 20s were rare and lower than for those born in earlier years. This provides reassurance that the updated guidelines to delay screening until 21 has not resulted in a population-level increase in ICC rates among young women.

Population-Based Precision Cancer Screening: A Symposium on Evidence, Epidemiology, and Next Steps

Precision medicine, an emerging approach for disease treatment that takes into account individual variability in genes, environment, and lifestyle, is under consideration for preventive interventions, including cancer screening. On September 29, 2015, the National Cancer Institute sponsored a symposium entitled "Precision Cancer Screening in the General Population: Evidence, Epidemiology, and Next Steps". The goal was two-fold: to share current information on the evidence, practices, and challenges surrounding precision screening for breast, cervical, colorectal, lung, and prostate cancers, and to allow for in-depth discussion among experts in relevant fields regarding how epidemiology and other population sciences can be used to generate evidence to inform precision screening strategies. Attendees concluded that the strength of evidence for efficacy and effectiveness of precision strategies varies by cancer site, that no one research strategy or methodology would be able or appropriate to address the many knowledge gaps in precision screening, and that issues surrounding implementation must be researched as well. Additional discussion needs to occur to identify the high priority research areas in precision cancer screening for pertinent organs and to gather the necessary evidence to determine whether further implementation of precision cancer screening strategies in the general population would be feasible and beneficial. Cancer Epidemiol Biomarkers Prev; 25(11); 1449-55. ©2016 AACR.

Proof-of-principle study of a novel cervical screening and triage strategy: Computer-analyzed cytology to decide which HPV-positive women are likely to have ≥CIN2

A challenge in implementation of sensitive HPV-based screening is limiting unnecessary referrals to colposcopic biopsy. We combined two commonly recommended triage methods: partial HPV typing and "reflex" cytology, evaluating the possibility of automated cytology. This investigation was based on 1,178 exfoliated cervical specimens collected during the enrollment phase of The Study to Understand Cervical Cancer Early Endpoints and Determinants (SUCCEED, Oklahoma City, OK). We chose a colposcopy clinic population to maximize number of outcomes, for this proof-of-principle cross-sectional study. Residual aliquots of PreservCyt were HPV-typed using Linear Array (LA, Roche Molecular Systems, Pleasanton, CA). High-risk HPV typing data and cytologic results (conventional and automated) were used jointly to predict risk of histologically defined ≥CIN2. We developed a novel computer algorithm that uses the same optical scanning features that are generated by the FocalPoint Slide Profiler (BD, Burlington, NC). We used the Least Absolute Shrinkage and Selection Operator (LASSO) method to build the prediction model based on a training dataset (n = 600). In the validation set (n = 578), for triage of all HPV-positive women, a cytologic threshold of ≥ASC-US had a sensitivity of 0.94, and specificity of 0.30, in this colposcopy clinic setting. When we chose a threshold for the severity score (generated by the computer algorithm) that had an equal specificity of 0.30, the sensitivity was 0.91. Automated cytology also matched ≥ASC-US when partial HPV typing was added to the triage strategy, and when we re-defined cases as ≥CIN3. If this strategy works in a prospective screening setting, a totally automated screening and triage technology might be possible.

The population impact of human papillomavirus/cytology cervical cotesting at 3-year intervals: Reduced cervical cancer risk and decreased yield of precancer per screen

BACKGROUND

The objective of cervical screening is to detect and treat precancer to prevent cervical cancer mortality and morbidity while minimizing overtreatment of benign human papillomavirus (HPV) infections and related minor abnormalities. HPV/cytology cotesting at extended 5-year intervals currently is a recommended screening strategy in the United States, but the interval extension is controversial. In the current study, the authors examined the impact of a decade of an alternative, 3-year cotesting, on rates of precancer and cancer at Kaiser Permanente Northern California. The effect on screening efficiency, defined as numbers of cotests/colposcopy visits needed to detect a precancer, also was considered.

METHODS

Two cohorts were defined. The "open cohort" included all women screened at least once during the study period; > 1 million cotests were performed. In a fixed "long-term screening cohort," the authors considered the cumulative impact of repeated screening at 3-year intervals by restricting the cohort to women first cotested in 2003 through 2004 (ie, no women entering screening later were added to this group).

RESULTS

Detection of cervical intraepithelial neoplasia 3/adenocarcinoma in situ (CIN3/AIS) increased in the open cohort (2004-2006: 82.0/100,000 women screened; 2007-2009: 140.6/100,000 women screened; and 2010-2012: 126.0/100,000 women screened); cancer diagnoses were unchanged. In the long-term screening cohort, the detection of CIN3/AIS increased and then decreased to the original level (2004-2006: 80.5/100,000 women screened; 2007-2009: 118.6/100,000 women screened; and 2010-2012: 84.9./100,000 women screened). The number of cancer diagnoses was found to decrease. When viewed in terms of screening efficiency, the number of colposcopies performed to detect a single case of CIN3/AIS increased in the cohort with repeat screening.

CONCLUSIONS

Repeated cotesting at a 3-year interval eventually lowers population rates of precancer and cancer. However, a greater number of colposcopies are required to detect a single precancer. Cancer 2016;122:3682-6. © 2016 American Cancer Society.

HPV16 Sublineage Associations With Histology-Specific Cancer Risk Using HPV Whole-Genome Sequences in 3200 Women

BACKGROUND

HPV16 is a common sexually transmitted infection although few infections lead to cervical precancer/cancer; we cannot distinguish nor mechanistically explain why only certain infections progress. HPV16 can be classified into four main evolutionary-derived variant lineages (A, B, C, D) that have been previously suggested to have varying disease risks.

METHODS

We used a high-throughput HPV16 whole-genome sequencing assay to investigate variant lineage risk among 3215 HPV16-infected women. Using sublineages A1/A2 as the reference, we assessed all variant lineage associations with infection outcome over three or more years of follow-up: 1107 control subjects (

RESULTS

A4 sublineage was associated with an increased risk of cancer, specifically adenocarcinoma (OR = 9.81, 95% CI = 2.02 to 47.69, P = 4.7x10(-03)). Lineage B had a lower risk of CIN3 (OR = 0.51, 95% CI = 0. 28 to 0.91, P = 02) while lineage C showed increased risk (OR = 2.06, 95% CI = 1.09 to 3.89, P = 03). D2/D3 sublineages were strongly associated with an increased risk of CIN3 and cancer, particularly D2 (OR for cancer = 28.48, 95% CI = 9.27 to 87.55, P = 5.0x10(-09)). D2 had the strongest increased risk of glandular lesions, AIS (OR = 29.22, 95% CI = 8.94 to 95.51, P = 2.3x10(-08)), and adenocarcinomas (OR = 137.34, 95% CI = 37.21 to 506.88, P = 1.5x10(-13)). Moreover, the risk of precancer and cancer for specific variant lineages varied by a women's race/ethnicity; those women whose race/ethnicity matched that of the infecting HPV16 variant had an increased risk of CIN3 + (P < 001).

CONCLUSIONS

Specific HPV16 variant sublineages strongly influence risk of histologic types of precancer and cancer, and viral genetic variation may help explain its unique carcinogenic properties.

Collapse

Key Words Collapse

MESH Headings

• Adenocarcinoma/ethnology
• Adenocarcinoma/virology
• Adenocarcinoma in Situ/ethnology
• Adenocarcinoma in Situ/virology
• Adult
• Aftercare
• California/epidemiology
• Carcinoma/ethnology
• Carcinoma/virology
• Carcinoma, Squamous Cell/ethnology
• Carcinoma, Squamous Cell/virology
• Female
• Genome
• High-Throughput Nucleotide Sequencing
• Human papillomavirus 16/classification
• Human papillomavirus 16/genetics
• Human papillomavirus 16/pathogenicity
• Humans
• Middle Aged
• Papillomavirus Infections/ethnology
• Papillomavirus Infections/virology
• Phylogeny
• Precancerous Conditions/ethnology
• Precancerous Conditions/virology
• Risk Factors
• Uterine Cervical Neoplasms/ethnology
• Uterine Cervical Neoplasms/virology
• Young Adult
• Uterine Cervical Dysplasia/ethnology
• Uterine Cervical Dysplasia/virology

Collapse

Grants

• U01 CA078527 NCI NIH HHS

Collapse

Distribution of cell types differs in Papanicolaou tests of squamous cell carcinomas and adenocarcinomas

INTRODUCTION

Successful cervical cancer screening has led to decreasing numbers of malignant Papanicolaou tests in most laboratories. A previous study demonstrated a greater trend to unsatisfactory Papanicolaou tests in women with squamous carcinoma when compared with adenocarcinoma cases. However, adenocarcinomas were less frequently recognized as malignant.

MATERIALS AND METHODS

In an effort to elucidate differences in Papanicolaou tests from these tumor types, the relative distribution of cells was blindly and semi-quantitatively assessed in ThinPrep Papanicolaou slides from 332 women with biopsy-proven squamous carcinoma (237 cases), adenocarcinoma (45), and noninvasive lesions (50).

RESULTS

Significant differences (P < 0.0001) among the three categories were observed in total cellularity, amount of blood and diathesis, normal endocervical cells, and normal squamous cells. When slides from squamous carcinomas and adenocarcinomas were compared, the amount of blood (P < 0.4) and presence of diathesis (P > 0.004) were more prominent in squamous carcinomas. The number of endocervical cells (P < 0.0001) was greater in adenocarcinomas, but adenocarcinomas were less likely to be recognized as malignant.

CONCLUSIONS

This systematic evaluation reinforces earlier suggestions that the presence of blood and tumor diathesis allow easier recognition of squamous carcinoma. A more detailed analysis of adenocarcinoma's cellular characteristics in Papanicolaou tests is needed to understand the reasons for undercalls in this tumor type.

Impact of human papillomavirus (HPV) 16 and 18 vaccination on prevalent infections and rates of cervical lesions after excisional treatment

BACKGROUND

Human papillomavirus (HPV) vaccines prevent HPV infection and cervical precancers. The impact of vaccinating women with a current infection or after treatment for an HPV-associated lesion is not fully understood.

OBJECTIVES

To determine whether HPV-16/18 vaccination influences the outcome of infections present at vaccination and the rate of infection and disease after treatment of lesions.

STUDY DESIGN

We included 1711 women (18–25 years) with carcinogenic human papillomavirus infection and 311 women of similar age who underwent treatment for cervical precancer and who participated in a community-based trial of the AS04-adjuvanted HPV-16/18 virus-like particle vaccine. Participants were randomized (human papillomavirus or hepatitis A vaccine) and offered 3 vaccinations over 6 months. Follow-up included annual visits (more frequently if clinically indicated), referral to colposcopy of high-grade and persistent low-grade lesions, treatment by loop electrosurgical excisional procedure when clinically indicated, and cytologic and virologic follow-up after treatment. Among women with human papillomavirus infection at the time of vaccination, we considered type-specific viral clearance, and development of cytologic (squamous intraepithelial lesions) and histologic (cervical intraepithelial neoplasia) lesions. Among treated women, we considered single-time and persistent human papillomavirus infection, squamous intraepithelial lesions, and cervical intraepithelial neoplasia 2+. Outcomes associated with infections absent before treatment also were evaluated. Infection-level analyses were performed and vaccine efficacy estimated.

RESULTS

Median follow-up was 56.7 months (women with human papillomavirus infection) and 27.3 months (treated women). There was no evidence of vaccine efficacy to increase clearance of human papillomavirus infections or decrease incidence of cytologic/histologic abnormalities associated with human papillomavirus types present at enrollment. Vaccine efficacy for human papillomavirus 16/18 clearance and against human papillomavirus 16/18 progression from infection to cervical intraepithelial neoplasia 2+ were −5.4% (95% confidence interval −19,10) and 0.3% (95% confidence interval −69,41), respectively. Among treated women, 34.1% had oncogenic infection and 1.6% had cervical intraepithelial neoplasia 2+ detected after treatment, respectively, and of these 69.8% and 20.0% were the result of new infections. We observed no significant effect of vaccination on rates of infection/lesions after treatment. Vaccine efficacy estimates for human papillomavirus 16/18 associated persistent infection and cervical intraepithelial neoplasia 2+ after treatment were 34.7% (95% confidence interval −131, 82) and −211% (95% confidence interval −2901, 68), respectively. We observed evidence for a partial and nonsignificant protective effect of vaccination against new infections absent before treatment. For incident human papillomavirus 16/18, human papillomavirus 31/33/45, and oncogenic human papillomavirus infections post-treatment, vaccine efficacy estimates were 57.9% (95% confidence interval −44, 88), 72.9% (95% confidence interval 29, 90), and 36.7% (95% confidence interval 1.5, 59), respectively.

CONCLUSION

We find no evidence for a vaccine effect on the fate of detectable human papillomavirus infections. We show that vaccination does not protect against infections/lesions after treatment. Evaluation of vaccine protection against new infections and resultant lesions warrants further consideration in future studies.

Abstract 2586: Evaluation of risk-based colposcopy in the ALTS trial

Objective: Women referred to colposcopy have a wide spectrum of underlying risk of precancer that could influence colposcopic practice and management. We sought to evaluate important risk-strata based on HPV status, cytology, and colposcopic impression in women undergoing colposcopy.

Methods: Among participants in the ASCUS-LSIL Triage Study, we stratified women from the immediate colposcopy and HPV triage study arms based on enrollment HPV typing, study cytology, and colposcopic impression. In each group, absolute risk of CIN3+ at enrollment colposcopy and cumulative risk over 2-year follow-up were estimated. Immediate CIN3+ risk of LSIL and HSIL cytology were used as thresholds for colposcopy referral and immediate treatment, respectively.

Results: We observed substantial differences in risk of precancer across strata. Women with HSIL cytology, who were HPV16-positive, and high-grade colpscopic impression had a 71% baseline and 78% 2-year risk of CIN3+, respectively. HPV16-positive women with HSIL cytology and low-grade impression, and women with high-grade impression and either HSIL cytology or HPV16 infection, also showed significant CIN3+ risk exceeding the risk of HSIL. In contrast, women with normal colposcopy, &lt;HSIL cytology, and absence of HPV16 infection had the lowest baseline CIN3+ risk (3.2%), below the colposcopy referral threshold in ALTS.

Conclusion: Risk assessment at colposcopy can inform colposcopy-biopsy practice and guide management. In the low-risk group, biopsy yields little additional disease and a normal colposcopy may confer higher reassurance that CIN3+ is not present; while in the highest risk groups, immediate treatment without biopsy confirmation could be considered according to current guidelines. Analyses evaluating the benefit of multiple biopsies in these risk strata are underway and will be presented at the meeting.

Citation Format: Angela H. Liu, Mark Schiffman, Julia C. Gage, Philip E. Castle, Nicolas Wentzensen. Evaluation of risk-based colposcopy in the ALTS trial. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2586.

Abstract 2579: Longitudinal predictors of adherence to cervical cancer cotesting guidelines

Introduction: Although clinical guidelines recommend cervical cancer screening using human papillomavirus (HPV) combined with Papanicolaou (Pap) testing for average-risk women, little is known about the longitudinal adoption of cotesting in clinical care, or if interval extension resistance is evident in practice. Using data from Kaiser Permanente Northern California (KPNC), which switched from annual Pap to 3-year interval cotesting in 2003, we examined predictors of cotesting guideline adherence.

Methods: We included all female patients aged 30-64 years who underwent baseline cotesting between 2003-2007 and completed at least one additional screening within 5.5 years. We excluded patients with a documented history of &gt;CIN2, or positive HPV test at baseline or subsequent screenings (leaving 335,822 patients for analysis). We categorized interval length between cotesting into 3 categories: early (&lt;2.5 years), adherent (2.5-3.5 years), and late (3.5-5.5 years). We used multinominal logistic regression models to examine the association (adjusted odds ratio [aOR]) between interval length from baseline cotesting to first follow-up cotesting (Interval 1) and the following predictors assessed at baseline: year of first cotesting, age, previous hysterectomy, and previous low-grade abnormal pap (&lt;CIN2). We also examined predictors of persistent early screening in women classified as early screeners in both Interval 1 and Interval 2 (defined as time between first and second follow-up cotesting).

Results: Compared with the earliest cohort of women (baseline cotesting in 2003), the 2007 cohort was 66% and 52% less likely of being screened early (aOR = 0.34, 95% CI: 0.33, 0.35) or late (aOR = 0.48, 95% CI: 0.46, 0.50). However, among women classified as persistent early screeners, we found no clear trend in the relationship between year of initial cotesting and interval length. The strongest predictor of persistent early screening was older age, whereby women aged 60-64 years had the greatest likelihood of being a persistent early screener (aOR = 2.1, 95% CI: 1.9, 2.3) versus women aged 30-34 years. Despite subsequent negative screens, women reporting a previous low-grade abnormal Pap at baseline had a greater likelihood of being a persistent early screener compared with women who did not (aOR = 1.8, 95% CI: 1.6, 2.0).

Conclusions: Overall, women who underwent baseline cotesting in more recent years were significantly more likely to adhere to the recommended interval length than women in earlier years, indicating an increase in adoption of cotesting guidelines over time. However, there remain subgroups, including older women, which appear less likely to follow extended cotesting intervals. Further research is needed to understand these HPV screening practices in other healthcare settings and identify additional (potentially modifiable) patient and provider factors that are associated with screening practices.

Citation Format: Katharine A. Rendle, Mark Schiffman, Li C. Cheung, Walter K. Kinney, Barbara Fetterman, Nancy E. Poitras, Philip E. Castle. Longitudinal predictors of adherence to cervical cancer cotesting guidelines. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2579.

Variant-specific persistence of infections with human papillomavirus Types 31, 33, 45, 56 and 58 and risk of cervical intraepithelial neoplasia

In our previous study of the etiologic role of oncogenic human papillomavirus (HPV) types other than HPV16 and 18, we observed a significantly higher risk of cervical intraepithelial neoplasia Grades 2-3 (CIN2/3) associated with certain lineages of HPV types 31/33/45/56/58 [called high-risk (HR) variants] compared with non-HR variants. This study was to examine whether these intra-type variants differ in persistence of the infection and persistence-associated risk of CIN2/3. Study subjects were women who had any of HPV types 31/33/45/56/58 newly detected during a 2-year follow-up with 6-month intervals. For each type, the first positive sample was used for variant characterization. The association of reverting-to-negativity with group of the variants and CIN2/3 with length of positivity was assessed using discrete Cox regression and logistic regression, respectively. Of the 598 newly detected, type-specific HPV infections, 312 became undetectable during follow-up. Infections with HR, compared with non-HR, variants were marginally more likely to become negative [adjusted hazard ratio = 1.3; 95% confidence interval (CI), 0.9-1.8]. The adjusted odds ratio associating with the development of CIN2/3 was 3.0 (95% CI, 1.2-7.4) for persistent infections with HR variants for 6 months and 10.0 (95% CI, 3.8-38.0) for persistent infections with HR variants for 12-18 months as compared with the first positive detection of HR variants. Among women with non-HR variants, there were no appreciable differences in risk of CIN2/3 by length of positivity. Findings suggest that the lineage-associated risk of CIN2/3 was not mediated through a prolonged persistent infection, but oncogenic heterogeneity of the variants.

Comparison of Colposcopic Impression Based on Live Colposcopy and Evaluation of Static Digital Images

OBJECTIVE

The aim of the study was to evaluate the agreement and compare diagnostic accuracy of colposcopic impressions from live colposcopy versus evaluation of static digital images.

MATERIALS AND METHODS

Live impressions and corresponding static images obtained during colposcopy of 690 women were independently compared. Diagnostic accuracy was calculated for colposcopic impressions from both methods, varying hypothetical thresholds for colposcopically directed cervical biopsies (acetowhitening or worse, low grade or worse, high grade or worse). Stratified analyses investigated the impact of referral cytology, human papillomavirus 16 infection, and age on colposcopic impression.

RESULTS

Overall agreement between live and static colposcopic visualization was 43.0% (κ = 0.20; 95% CI = 0.14-0.26) over normal, acetowhitening, low-grade, and high-grade impressions. Classification of acetowhitening or worse impressions showed the highest agreement (92.2%; κ = 0.39; 95% CI = 0.21-0.57); both methods achieved more than 95% sensitivity for CIN 2+. Agreement between live and static colposcopic visualization was 69.3% for rating low-grade or worse impressions (κ = 0.23; 95% CI = 0.14-0.33) and 71% when rating high-grade impressions (κ = 0.33; 95% CI = 0.24-0.42). Live colposcopic impressions were more likely to be rated low grade or worse (p < .01; odds ratio = 3.5; 95% CI = 2.4-5.0), yielding higher sensitivity for CIN 2+ at this threshold than static image assessment (95.4% vs 79.8%, p < .01). Overall, colposcopic impressions were more likely rated high grade on live assessment among women referred with high-grade cytology (odds ratio = 3.3; 95% CI = 1.8-6.4), significantly improving the sensitivity for CIN 2+ (66.3% vs 48.5%, p < .01).

CONCLUSIONS

Colposcopic impressions of acetowhitening or worse are highly sensitive for identifying cervical precancers and reproducible on static image-based pattern recognition.

Use of primary high-risk human papillomavirus testing for cervical cancer screening: interim clinical guidance

In 2011, the American Cancer Society, the American Society for Colposcopy and Cervical Pathology, and the American Society for Clinical Pathology updated screening guidelines for the early detection of cervical cancer and its precursors. Recommended screening strategies were cytology or cotesting (cytology in combination with high-risk HPV (hrHPV) testing). These guidelines also addressed the use of hrHPV testing alone as a primary screening approach, which was not recommended for use at that time. There is now a growing body of evidence for screening with primary hrHPV testing, including a prospective US-based registration study. Thirteen experts including representatives from the Society of Gynecologic Oncology, American Society for Colposcopy and Cervical Pathology, American College of Obstetricians and Gynecologists, American Cancer Society, American Society of Cytopathology, College of American Pathologists, and the American Society for Clinical Pathology, convened to provide interim guidance for primary hrHPV screening. This guidance panel was specifically triggered by an application to the FDA for a currently marketed HPV test to be labeled for the additional indication of primary cervical cancer screening. Guidance was based on literature review and review of data from the FDA registration study, supplemented by expert opinion. This document aims to provide information for health care providers who are interested in primary hrHPV testing and an overview of the potential advantages and disadvantages of this strategy for screening as well as to highlight areas in need of further investigation.

Risk assessment to guide cervical screening strategies in a large Chinese population

Three different cervical screening methods [cytology, human papillomavirus(HPV) testing and visual inspection with acetic acid(VIA)] are being considered in China for the national cervical screening program. Comparing risks of CIN3 and cervical cancer (CIN3+) for different results can inform test choice and management guidelines. We evaluated the immediate risk of CIN3+ for different screening results generated from individual and combined tests. We compared tests using a novel statistic designed for this purpose called Mean Risk Stratification (MRS), in a pooled analysis of 17 cross sectional population-based studies of 30,371 Chinese women screened with all 3 methods and diagnosed by colposcopically-directed biopsies. The 3 tests combined powerfully distinguished CIN3+ risk; triple-negative screening conferred a risk of 0.01%, while HPV-positive HSIL+ that was VIA-positive yielded a risk of 57.8%. Among the three screening tests, HPV status most strongly stratified CIN3+ risk. Among HPV-positive women, cytology was the more useful second test. In HPV-negative women, the immediate risks of CIN3+ ranged from 0.01% (negative cytology), 0.00% (ASC-US), 1.1% (LSIL), to 6.6 (HSIL+). In HPV-positive women, the CIN3+ risks were 0.9% (negative cytology), 3.6% (ASC-US), 6.3% (LSIL) and 38.5% (HSIL+). VIA results did not meaningful stratify CIN3+ risk among HPV-negative women with negative or ASC-US cytology; however, positive VIA substantially elevated CIN3+ risk for all other, more positive combinations of HPV and cytology compared with a negative VIA. Because all 3 screening tests had independent value in defining risk of CIN3+, different combinations can be optimized as pragmatic strategies in different resource settings.