Organized cervical cancer screening reduces cervical cancer incidence. Cytology-based screening programs were introduced several decades ago in many high-income countries. High-risk human papillomavirus (hrHPV) causes the overwhelming majority of cervical cancer cases, and hrHPV-based primary screening has been shown to be more sensitive in detecting high-grade squamous intraepithelial lesions (HSILs) and more effective in preventing cervical cancer than traditional cytology-based screening, ^, increasingly becoming the preferred screening modality. The introduction of hrHPV-based screening increases colposcopy referrals during the initial years of implementation, ^, due to higher sensitivity but lower specificity than cytology-based screening for detecting HSIL. ^,

Colposcopy is the gold standard follow-up investigation after a positive screening test. Sensitivity of colposcopic impression for detecting HSIL in patients with low-grade cytology has been reported to be 23% to 26%. ^, Classical colposcopic findings described by International Federation for Cervical Pathology and Colposcopy and clinical scoring systems, such as Swedescore, have been described and developed in clinical settings, where colposcopy is typically performed on patients with abnormal cytology. ^, The prevalence of HSIL in hrHPV-positive patients with cytology negative for intraepithelial lesion or malignancy (NILM) has been estimated to range from 7% to 25%. ^,,,, The role of colposcopy in an hrHPV-screened population has been discussed already over a decade ago, but little research has been conducted on the performance of colposcopy in persistently hrHPV-positive patients with NILM cytology, that is, those who are not referred to colposcopy according to cytology-based screening. Therefore, it remains unclear whether the diagnostic performance of colposcopy and clinical scoring tools like the Swedescore remain consistent in this population.

Our objective was to study the performance of colposcopy and Swedescore as diagnostic tools, present colposcopic findings, and assess the prevalence of HSIL in a population of persistently hrHPV-positive patients with NILM or atypical squamous cells of undetermined significance (ASC-US) cytology.

The study was conducted in Finland at the Helsinki Women’s Hospital Gynecological Outpatient Clinic, the main reference center within the capital region with more than 5000 annual colposcopies. In 2019, cytology-based screening was replaced by hrHPV-based screening with cytology triage, conducted at 5-year intervals starting at age 30. Screening is also offered at age 25 with cytology only. Aptima HPV assay (Hologic, Inc) without genotyping was used for hrHPV testing. Cervical cytology was performed using the conventional method until 2021, when the hospital district transitioned to liquid-based cytology (ThinPrep; Hologic, Inc). Patients with a positive hrHPV test and NILM cytology were invited for follow-up testing within 1 to 2 years. If the follow-up test was positive for hrHPV with NILM or ASC-US cytology, these patients were referred for colposcopy within 6 months.

Inclusion criteria for this study were: (1) individuals living in Helsinki who attended the national screening program, (2) Two positive hrHPV tests with reflex cytology at least 1 year apart between 2019 and 2021, (3) first reflex cytology NILM and second NILM or ASC-US, and (4) attended colposcopy at the Helsinki Women’s Hospital Gynecological Outpatient Clinic. Exclusion criteria were: (1) neither a cervical colposcopic impression was recorded nor histological samples were taken from the cervix and (2) previous hysterectomy. The colposcopies were performed between July 2021 and February 2023.

All colposcopies were performed by senior colposcopists or trainees under their direct supervision. Transformation zone (TZ) type (as defined by International Agency for Research on Cancer terminology) and an overall clinical colposcopic impression (categories: normal, human papillomavirus [HPV] atypia, cervical intraepithelial neoplasia [CIN] grade 1 to 3, adenocarcinoma in situ [AIS], cancer, other) were recorded routinely and Swedescore was used for scoring and assessment of cervical lesions. The overall threshold for performing a cervical punch biopsy was a suspected low-grade squamous intraepithelial lesion and not limited to suspected HSILs. See-and-treat loop electrosurgical excision procedure (LEEP) was discouraged in cases of low-grade referral cytology. No national or institutional guidelines existed for diagnostic approaches in cases with NILM/ASC-US cytology when no cervical lesions were observed during colposcopy. Therefore, the decision to take random biopsies, perform endocervical curettage (ECC), or take no histological samples was left to the individual clinician’s discretion.

We collected data from the electronic medical records systems, including the colposcopy report of each patient, Swedescore, TZ type, colposcopic diagnosis, referral cytology, and cervical histopathology results. In cases where the clinician was undecided between 2 colposcopic diagnoses (eg, CIN 1–2), we used the more severe diagnosis for analysis. For statistical analysis, we divided colposcopic impression and histopathological results into 2 groups, which are referred to hereafter as: non-HSIL (for normal, low-grade squamous intraepithelial lesion, other non-neoplastic) and HSIL+ (for CIN2, CIN3, AIS, cancer, unspecified dysplasia). Stratified analyses were used to assess the effect of the collected explanatory variables.

For ordinal or non-normally distributed data, the Mann-Whitney U test was used to compare distributions between 2 groups, while the Kruskal-Wallis rank sum test was applied for 3 or more groups. The 2-sample test for equality of proportions was used for group comparisons with large enough group sizes, where normal approximation of the binomial distribution could be assumed. If this was not the case, Fisher’s exact test was used. For comparison of areas under the curve (AUCs) of 2 receiver operating characteristic curves, DeLong’s test was used. The optimal Swedescore cutoff was determined using the Youden index, and to assess the stability of the optimal cutoff, bootstrapped resampling (B=2000) was performed by repeatedly sampling the dataset with replacement and identifying the Youden-optimal threshold in each resample. P values of <.05 were considered statistically significant. All statistical tests used were 2-sided. R version 4.4.1 was used for statistical analysis.

This study was conducted with approval of the review board of the Helsinki and Uusimaa Hospital District (decision HUS/538/2024) and in accordance with ethical principles established by the Declaration of Helsinki. We did not seek medical ethics committee approval or individual informed consent, since neither is required for registry-based studies under Finnish legislation.

We identified 536 patients initially eligible for inclusion. Of these, 23 cases were excluded because of missing records, 15 for previous hysterectomy, and 5 because of incomplete records on colposcopic impression and histopathology. After exclusions, 493 patients were included in the analysis. The mean age was 43.7 years (standard deviation ±11.9; range, 32–68), with a median age of 38 years. Of these, 345/493 patients (70.0%) were aged less than 50 years, while 148/493 (30.0%) were aged more than 50 years.

Histopathological examination identified HSIL+ in 88/489 (18.0%; 95% confidence interval [CI], 14.7–21.7) patients who underwent cervical punch biopsy, LEEP, or ECC. Among these, there were 2 cases (2/489, 0.04%) of AIS. We found no cases of invasive carcinoma ( Table 1 ).

In the most recent triage cytology before colposcopy, 337 patients had NILM cytology and 156 had ASC-US cytology. The proportion of HSIL+ was significantly higher in cases with ASC-US compared to cases with NILM (27.3 vs 13.7%; difference 13.6%; 95% CI, 5.1–22.0; P <.001).

Of 493 total cases, 460 (93.3%) had documented Swedescores and underwent histological sampling. Cases with histopathologically confirmed HSIL+ exhibited significantly higher Swedescores across all subcategories compared to non-HSIL cases ( Figure 1 ). The distribution of Swedescores ( Figure 2 ) further demonstrated a positive association between higher scores and increased HSIL+ prevalence. HSIL+ prevalence and Swedescores were higher in patients with ASC-US cytology (mean Swedescore, 4.0±2.2; median, 4.0; interquartile range [IQR], 2.0) than in those with NILM cytology (mean, 3.6±2.1; median, 4.0 [IQR, 3.0]; P =.048). Among HSIL+ cases, age (less than 50 years vs more than 50 years) showed no significant association with Swedescore distributions, except for iodine staining ( Supplemental Table 1 A).

Frequency distribution of individual Swedescore variables in cases with data for both Swedescore and histopathology (n=460)

Differences in score distribution between HSIL and non-HSIL were statistically significant ( P <.05) for all variables (Mann-Whitney U test)

HSIL , high-grade squamous intraepithelial lesion.

Distribution of total Swedescores

(A) Whole cohort, n=460. (B) Distribution by most recent cytology: NILM cytology, n=314; ASC-US cytology, n=146

ASC-US , atypical squamous cells of undetermined significance; NILM , negative for intraepithelial lesion or malignancy.

In HSIL+ cases with Swedescore data (n=85/493), ASC-US cytology was associated with higher total Swedescores (mean, 5.9±1.8; median, 6.0 [IQR, 2.3]) compared to NILM cytology (mean, 4.7±1.7; median, 5.0 [IQR, 2.0]; P =.01). Significant differences in lesion size and iodine staining subcategories were observed between cytology groups ( Supplemental Table 1 B).

At Swedescore thresholds of ≥8 and ≥6, sensitivities for HSIL+ were 11% (9/85) and 47% (40/85), with specificities of 99% and 86%, respectively. Receiver operating characteristic analysis revealed moderate discriminative ability for HSIL+ in the overall cohort (AUC, 0.75; 95% CI, 0.69–0.80), although performance varied by cytology: ASC-US cases showed higher AUC (0.83; 95% CI, 0.76–0.90) than NILM cases (0.68; 95% CI, 0.60–0.76; P =.006) ( Figure 3 ). The optimal Swedescore cutoff for HSIL+ was 5 (>4.5) by the Youden index ( Tables 2–4 ), with a bootstrapped 95% CI of 4 to 6 (>3.5 to >5.5; B=2000 resamples).

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Asthma in pregnancy: contemporary management
2. Associations between self-reported mood-related symptoms and early contraceptive method discontinuation
3. Differences in epigenetic age acceleration according to a diagnosis of endometriosis/adenomyosis: the Norwegian Mother, Father and Child Cohort Study
4. The impact of first birth by cesarean delivery on subsequent reproductive outcomes—a population cohort study

Stay updated, free articles. Join our Telegram channel

Join