Mari Nygård

Beyond Women's Health: Long-Term Human Papillomavirus–Related Cancer Trends in Norway

Abstract Background Understanding the total burden of human papillomavirus (HPV)–related cancers is crucial for improving prevention strategies. While organized cervical cancer (CC) screening has been implemented for many years, other HPV-related cancers lacked screening programs. Primary prevention through HPV vaccination has been implemented through national programs, initially for girls and later for boys. To analyze changes in HPV-attributable cancer incidence for both men and women, we used data from the Cancer Registry of Norway (CRN). Methods This was an observational population-based study using high-quality data from the CRN. The proportion of cancers at each site attributed to HPV was calculated based on existing literature. We estimated HPV cancer incidence rates from 1990 to 2023 and forecasted incidences until 2038 for cervical squamous cell carcinoma (SCC) and adenocarcinoma, along with other HPV-related SCCs. Results Among men, HPV-attributable cancer incidence was rising, with male oropharyngeal SCC showing the fastest increase (annual percentage change [APC], 4.5; P < .01). Overall, the incidence of HPV-attributable cancers not prevented by screening steadily increased (APC, 2.8; P < .01), surpassing CC incidence and projected to continue rising until 2038. In women, CC remains the most common HPV cancer. However, after an increasing trend since 2004, cervical SCC incidence rates decreased 6% annually from 2018 to 2023 (95% confidence interval [CI], −9.9 to −1.9; P < .01). Conclusions The burden of HPV-related cancers beyond CC is increasing in Norway, whereas CC incidence is declining. Addressing the rising total burden of HPV-attributable cancers requires additional preventive measures.

Sociodemographic characteristics associated with cervical cancer screening participation by send‐to‐all and opt‐in HPV self‐sampling: Who benefits? Results from a randomized controlled trial among long‐term non‐attending women in Norway

AbstractWith the objective to investigate associations between sociodemographic characteristics and participation in interventions designed to increase participation in cervical cancer screening among under‐screened women, we randomized a random sample of 6000 women in Norway aged 35–69 years who had not attended cervical screening for ≥10 years to receive either (i) a reminder to attend regular screening (control), (ii) an offer to order a self‐sampling kit (opt‐in), or (iii) a self‐sampling kit unsolicited (send‐to‐all). We analyzed how sociodemographic characteristics were associated with screening participation within and between screening arms. In the send‐to‐all arm, increased screening participation ranged from 17.1% (95% confidence interval [95% CI] = 10.3% to 23.8%) to 30.0% (95% CI = 21.5% to 38.6%) between sociodemographic groups. In the opt‐in arm, we observed smaller, and at times, non‐significant increases within the range 0.7% (95% CI = −5.8% to 7.3%) to 19.1% (95% CI = 11.6% to 26.7%). In send‐to‐all versus control comparisons, there was greater increase in participation for women in the workforce versus not (6.1%, 95% CI = 1.6% to 10.6%), with higher versus lower income (7.6%, 95% CI = 2.2% to 13.1%), and with university versus primary education (8.5%, 95% CI = 2.4% to 14.6%). In opt‐in versus control comparisons, there was greater increase in participation for women in the workforce versus not (4.6%, 95% CI = 0.7% to 8.5%), with higher versus lower income (6.3%, 95% CI = 1.5% to 11.1%), but lower increase for Eastern European versus Norwegian background (−12.7%, 95% CI = −19.7% to −5.7%). Self‐sampling increased cervical screening participation across all sociodemographic levels, but inequalities in participation should be considered when introducing self‐sampling, especially with the goal to reach long‐term non‐attending women.

Real-world data on cervical cancer risk stratification by cytology and HPV genotype to inform the management of HPV-positive women in routine cervical screening

Abstract Background HPV16/18 detection may improve cervical cancer risk stratification and better guide which HPV-positive women warrant immediate colposcopy/biopsy. We estimated risks of cervical precancer and cancer by HPV genotype and cytology during the implementation phase of primary HPV testing in Norway. Methods A total of 3111 women, aged 34–69 years, testing HPV-positive at baseline and undergoing cytology testing from February 2015 to April 2018 had data available for analysis. Risk estimates with 95% confidence intervals (95%CIs) of cervical intraepithelial neoplasia grade 3 or more severe (CIN3+) were estimated for cytology results and HPV genotypes (HPV16, HPV18, and other high-risk HPV). Results CIN3+ risks were higher for HPV16/18 than other high-risk HPV genotypes. Among women with any cytologic abnormality [atypical squamous cells of undetermined significance or worse], immediate risks were 57.8% (95%CI = 53.0–62.6%) for HPV16, 40.2% (95%CI = 32.3–49.2%) for HPV18, and 31.4% (95%CI = 28.7–34.3%) for other high-risk HPV. Among those with normal cytology, CIN3+ risks were 19.9% (95%CI = 15.0–26.1%) for HPV16 positives, 10.8% (95%CI = 5.6–20.5%) for HPV18 positives, and 5.5% (95%CI = 4.2–7.1%) for other high-risk HPV. Conclusions The benefits and harms of managing women based on HPV positivity and cytology results can be better balanced by inclusion of HPV genotyping in screening and choosing more conservative management for other high-risk HPV compared to HPV16/18.

Hierarchical continuous-time inhomogeneous hidden Markov model for cancer screening with extensive followup data

Continuous-time hidden Markov models are an attractive approach for disease modeling because they are explainable and capable of handling both irregularly sampled, skewed and sparse data arising from real-world medical practice, in particular to screening data with extensive followup. Most applications in this context consider time-homogeneous models due to their relative computational simplicity. However, the time homogeneous assumption is too strong to accurately model the natural history of many diseases including cancer. Moreover, cancer risk across the population is not homogeneous either, since exposure to disease risk factors can vary considerably between individuals. This is important when analyzing longitudinal datasets and different birth cohorts. We model the heterogeneity of disease progression and regression using piece-wise constant intensity functions and model the heterogeneity of risks in the population using a latent mixture structure. Different submodels under the mixture structure employ the same types of Markov states reflecting disease progression and allowing both clinical interpretation and model parsimony. We also consider flexible observational models dealing with model over-dispersion in real data. An efficient, scalable Expectation-Maximization algorithm for inference is proposed with the theoretical guaranteed convergence property. We demonstrate our method’s superior performance compared to other state-of-the-art methods using synthetic data and a real-world cervical cancer screening dataset from the Cancer Registry of Norway. Moreover, we present two model-based risk stratification methods that identify the risk levels of individuals.

HPV self-sampling among long-term non-attenders to cervical cancer screening in Norway: a pragmatic randomised controlled trial

Abstract Background Cervical cancer screening participation is suboptimal in most settings. We assessed whether human papillomavirus (HPV) self-sampling may increase screening participation among long-term non-attenders in Norway. Methods A pragmatic randomised controlled trial with participation as the primary outcome was initiated in the national cervical screening programme in March 2019. A random sample of 6000 women aged 35–69 years who had not attended screening for at least 10 years were randomised 1:1:1 to receive either (i) a reminder to attend regular screening (control), (ii) an offer to order a self-sampling kit (opt-in) for HPV testing or (iii) a self-sampling kit unsolicited (send-to-all) for HPV testing. Results Total participation was 4.8%, 17.0% and 27.7% among control, opt-in and send-to-all ( P  < 0.0001; participation difference (%) send-to-all vs. control: 22.9 (95%CI: 20.7, 25.2); opt-in vs. control: 12.3 (95%CI: 10.3, 14.2); send-to-all vs. opt-in: 10.7 (95% CI: 8.0, 13.3)). High-risk HPV was detected in 11.5% of self-samples and 9.2% of clinician-collected samples ( P  = 0.40). Most women (92.5%) who returned a positive self-sample attended the clinic for triage testing. Of the 933 women screened, 33 (3.5%) had CIN2 + (1.1%, 3.7%, 3.8% among control, opt-in, and send-to-all, respectively), and 11 (1.2%) had cervical cancer (0%, 1.2%, 1.3% among control, opt-in, send-to-all, respectively). Conclusion Opt-in and send-to-all self-sampling increased screening participation among long-term, higher-risk non-attenders. Clinical trial registration ClinicalTrials.gov NCT03873376.

Randomized Implementation of a Primary Human Papillomavirus Testing–based Cervical Cancer Screening Protocol for Women 34 to 69 Years in Norway

Abstract Background: Cervical cancer screening programs are facing a programmatic shift where screening protocol based on human papillomavirus testing (HPV-Screening protocol) is replacing the liquid-based cytology (LBC-Screening protocol). For safe technology transfer within the nationwide screening programme in Norway, HPV-Screening protocol was implemented randomized to compare the real-world effectiveness of HPV-Screening protocol and LBC-Screening protocol at the first screening round. Methods: Among 302,295 women ages 34 to 69 years scheduled to attend screening from February 2015 to June 2017, 157,447 attended. A total of 77,207 were randomly allocated to the HPV-Screening protocol and 80,240 were allocated to the LBC-Screening protocol. All women were followed up for 18 months. Results: The HPV-Screening protocol resulted in a relative increase of 60% in the detection of cervical intraepithelial neoplasia (CIN) grade 2 or worse [risk ratio (RR) = 1.6, 95% confidence interval (CI) = 1.5–1.7], 40% in CIN grade 3 or worse (RR = 1.4, 95% CI = 1.3–1.6), 40% in cancer (RR = 1.4, 95% CI = 1.0–2.1), and 60% in colposcopy referrals (RR = 1.6, 95% CI = 1.5–1.6) compared with LBC-Screening. The performance of both protocols was age dependent, being more effective in women ages under 50 years. Conclusions: The HPV-Screening protocol was well accepted by women in Norway and detected more CIN2, CIN3, and cancers compared with the LBC-Screening protocol. Impact: A randomized implementation of the HPV-Screening protocol with real-world assessment enabled a gradual, quality assured, and safe technology transition. HPV-based screening protocol may further be improved by using HPV genotyping and age-specific referral algorithms.

Cross-population evaluation of cervical cancer risk prediction algorithms

Cervical cancer is a preventable disease, despite being one of the most common types of female cancers worldwide. Integrating existing programs for cervical cancer screening with personalized risk prediction algorithms can improve population-level cancer prevention by enabling more targeted screening and contrive preventive healthcare innovations. While algorithms developed for cervical cancer risk prediction have shown promising performance in internal validation on more homogeneous populations, their ability to generalize to external populations remains to be assessed. To address this gap, we perform a cross-population comparative study of personalized prediction algorithms for more personalized cervical cancer screening. Using data from the Norwegian and Estonian populations, the algorithms are validated on internal and external datasets to study their potential biases and limitations when applied to different populations. We evaluate the algorithms in predicting progression from low-grade precancerous cervical lesions, simulating a clinically relevant application of more personalized risk stratification. As expected, our numerical experiments show that algorithm performance varies depending on the population. However, some algorithms show strong generalization capacity across different data sources. Using Kaplan-Meier estimates, we demonstrate the strengths and limitations of the algorithms in detecting cancer progression over time by comparing to the trends observed from data. We assess their overall discrimination performance in personalized risk predictions by analyzing the accuracy and confidence in individual risk estimates. This study examines the effectiveness of personalized prediction algorithms across different populations. Our results demonstrate the potential for generalizing risk prediction algorithms to external populations. These findings highlight the importance of considering population diversity when developing risk prediction algorithms.

Self-sampling Among Long-term Non-attenders to Cervical Cancer Screening

The trial will evaluate whether human papillomavirus (HPV) self-sampling may increase cervical cancer screening attendance among under-screened women in Norway, how different ways of offering self-sampling and follow-up may affect attendance, and whether self-sampling may reduce inequities in attendance.