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

Data Characterization for Reliable AI in Medicine

Research in Artificial Intelligence (AI)-based medical computer vision algorithms bear promises to improve disease screening, diagnosis, and subsequently patient care. However, these algorithms are highly impacted by the characteristics of the underlying data. In this work, we discuss various data characteristics, namely Volume, Veracity, Validity, Variety, and Velocity, that impact the design, reliability, and evolution of machine learning in medical computer vision. Further, we discuss each characteristic and the recent works conducted in our research lab that informed our understanding of the impact of these characteristics on the design of medical decision-making algorithms and outcome reliability.

Moving towards a strategy to accelerate cervical cancer elimination in a high-burden city-Lessons learned from the Amazon city of Manaus, Brazil

The World Health Organization Call to Eliminate Cervical Cancer resonates in cities like Manaus, Brazil, where the burden is among the world’s highest. Manaus has offered free cytology-based screening since 1990 and HPV immunization since 2013, but the public system is constrained by many challenges and performance is not well-defined. We obtained cervical cancer prevention activities within Manaus public health records for 2019 to evaluate immunization and screening coverage, screening by region and neighborhood, and the annual Pink October screening campaign. We estimated that among girls and boys age 14–18, 85.9% and 64.9% had 1+ doses of HPV vaccine, higher than rates for age 9–13 (73.4% and 43.3%, respectively). Of the 90,209 cytology tests performed, 24.9% were outside the target age and the remaining 72,230 corresponded to 40.1% of the target population (one-third of women age 25–64). The East zone had highest screening coverage (49.1%), highest high-grade cytology rate (2.5%) and lowest estimated cancers (38.1/100,000) compared with the South zone (32.9%, 1.8% and 48.5/100,000, respectively). Largest neighborhoods had fewer per capita screening locations, resulting in lower coverage. During October, some clinics successfully achieved higher screening volumes and high-grade cytology rates (up to 15.4%). Although we found evidence of some follow-up within 10 months post-screening for 51/70 women (72.9%) with high-grade or worse cytology, only 18 had complete work-up confirmed. Manaus has successfully initiated HPV vaccination, forecasting substantial cervical cancer reductions by 2050. With concerted efforts during campaigns, some clinics improved screening coverage and reached high-risk women. Screening campaigns in community locations in high-risk neighborhoods using self-collected HPV testing can achieve widespread coverage. Simplifying triage and treatment with fewer visits closer to communities would greatly improve follow-up and program effectiveness. Achieving WHO Cervical Cancer Elimination goals in high-burden cities will require major reforms for screening and simpler follow-up and treatment.

The Natural History of Anal High-grade Squamous Intraepithelial Lesions in Gay and Bisexual Men

BACKGROUND

Gay and bisexual men (GBM) are disproportionately affected by anal cancer. Prevention is hindered by incomplete understanding of the natural history of its precursor, anal high-grade squamous intraepithelial lesions (HSIL).

METHODS

The Study of the Prevention of Anal Cancer, conducted between 2010 and 2018, enrolled human immunodeficiency virus (HIV)-positive and HIV-negative GBM aged ≥35 years. Anal cytology and high-resolution anoscopy (HRA) were performed at baseline and 3 annual visits. A composite HSIL diagnosis (cytology ± histology) was used. Cytological high-grade squamous intraepithelial lesions (cHSIL) incidence and clearance rates were calculated with 95% confidence intervals (CIs). Predictors were calculated using Cox regression with hazard ratios (HRs) and 95% CIs.

RESULTS

Among 617 men, 220 (35.7%) were HIV-positive, median age 49 years. And 124 incident cHSIL cases occurred over 1097.3 person-years (PY) follow-up (11.3, 95% CI 9.5-13.5 per 100 PY). Significant bivariate predictors of higher incidence included age <45 years (HR 1.64, 95% CI 1.11-2.41), HIV positivity (HR 1.43, 95% CI .99-2.06), prior SIL diagnosis (P-trend < .001) and human papillomavirus (HPV)16 (HR 3.39, 2.38-4.84). Over 695.3 PY follow-up, 153 HSIL cleared (clearance 22.0, 95% CI 18.8-25.8 per 100 PY). Predictors were age < 45 years (HR 1.52, 1.08-2.16), anal intraepithelial neoplasia (AIN)2 rather than AIN3 (HR 1.79, 1.29-2.49), smaller lesions (HR 1.62, 1.11-2.36) and no persistent HPV16 (HR 1.72, 1.23-2.41). There was 1 progression to cancer (incidence 0.224, 95% CI .006-1.25 per 100 PY).

CONCLUSION

These data strongly suggest that not all anal HSIL detected in screening requires treatment. Men with persistent HPV16 were less likely to clear HSIL and are more likely to benefit from effective HSIL treatments.

CLINICAL TRIALS REGISTRATION

Australia New Zealand Clinical Trials Registry (ANZCTR365383).

An Observational Study of Deep Learning and Automated Evaluation of Cervical Images for Cancer Screening

BACKGROUND

Human papillomavirus vaccination and cervical screening are lacking in most lower resource settings, where approximately 80% of more than 500 000 cancer cases occur annually. Visual inspection of the cervix following acetic acid application is practical but not reproducible or accurate. The objective of this study was to develop a "deep learning"-based visual evaluation algorithm that automatically recognizes cervical precancer/cancer.

METHODS

A population-based longitudinal cohort of 9406 women ages 18-94 years in Guanacaste, Costa Rica was followed for 7 years (1993-2000), incorporating multiple cervical screening methods and histopathologic confirmation of precancers. Tumor registry linkage identified cancers up to 18 years. Archived, digitized cervical images from screening, taken with a fixed-focus camera ("cervicography"), were used for training/validation of the deep learning-based algorithm. The resultant image prediction score (0-1) could be categorized to balance sensitivity and specificity for detection of precancer/cancer. All statistical tests were two-sided.

RESULTS

Automated visual evaluation of enrollment cervigrams identified cumulative precancer/cancer cases with greater accuracy (area under the curve [AUC] = 0.91, 95% confidence interval [CI] = 0.89 to 0.93) than original cervigram interpretation (AUC = 0.69, 95% CI = 0.63 to 0.74; P < .001) or conventional cytology (AUC = 0.71, 95% CI = 0.65 to 0.77; P < .001). A single visual screening round restricted to women at the prime screening ages of 25-49 years could identify 127 (55.7%) of 228 precancers (cervical intraepithelial neoplasia 2/cervical intraepithelial neoplasia 3/adenocarcinoma in situ [AIS]) diagnosed cumulatively in the entire adult population (ages 18-94 years) while referring 11.0% for management.

CONCLUSIONS

The results support consideration of automated visual evaluation of cervical images from contemporary digital cameras. If achieved, this might permit dissemination of effective point-of-care cervical screening.

High-resolution HLA allele and haplotype frequencies in majority and minority populations of Costa Rica and Nicaragua: Differential admixture proportions in neighboring countries

The HLA system shows the most extensive polymorphism in the human genome. Allelic and haplotypic frequencies of HLA genes vary dramatically across human populations. Due to a complex history of migration, populations in Latin America show a broad variety of admixture proportions, usually varying not only between countries, but also within countries. Knowledge of HLA allele and haplotype frequencies is essential for medical fields such as transplantation, but also serves as a means to assess genetic diversity and ancestry in human populations. Here, we have determined high-resolution HLA-A, -B, -C, and -DRB1 allele and haplotype frequencies in a sample of 713 healthy subjects from three Mestizo populations, one population of African descent, and Amerindians of five different groups from Costa Rica and Nicaragua and compared their profiles to a large set of indigenous populations from Iberia, Sub-Saharan Africa, and the Americas. Our results show a great degree of allelic and haplotypic diversity within and across these populations, with most extended haplotypes being private. Mestizo populations show alleles and haplotypes of putative European, Amerindian, and Sub-Saharan African origin, albeit with differential proportions. Despite some degree of gene flow, Amerindians and Afro-descendants show great similarity to other Amerindian and West African populations, respectively. This is the first comprehensive study reporting high-resolution HLA diversity in Central America, and its results will shed light into the genetic history of this region while also supporting the development of medical programs for organ and stem cell transplantation.