Overview of Ancillary Techniques in Cervical Cytology

Cervical cytology is one of the most well-established and successful cancer screening tests utilized today. Following the introduction of the cervical Papanicolaou (Pap) smear into clinical practice in the 1940s, there has been a clear reduction in the incidence and mortality associated with cervical carcinoma. Screening has resulted in not only increased detection of early stage invasive cancer but also detection and subsequent treatment of precancerous lesions. Liquid-based cytology (LBC) has largely replaced the conventional Pap smear due to its improved sensitivity and specificity for detecting intraepithelial neoplasia, as well as for the potential of automated screening [1-3].

The vast majority of cervical cancers are caused by persistent infection with human papillomavirus (HPV) [4]. It is estimated that only less than 1% of all cervical squamous cell carcinomas are non-HPV associated, while the reported rate of non-HPV associated endocervical adenocarcinomas ranges from 7% to 40% [4-6]. Large-scale epidemiological studies have demonstrated heterogeneity within the HPV family, leading to the distinction between low-risk (most commonly 6 and 11) and high-risk (most commonly 16, 18, 31, 33, and 45) serotypes, based on their ability to promote malignant transformation. It is now well-established that persistent infection with high-risk HPV (hr-HPV) is a significant risk factor for the development of precancerous and cancerous lesions of the cervix, while low-risk serotypes are generally associated with genital warts and nonmalignant lesions [4, 7]. Although the overall lifetime risk of hr-HPV infection is high, only a small proportion of these infections lead to cervical carcinogenesis. More than 90% of infections are cleared by the immune system within 24 months [8, 9]. Persistent infections can develop into either “productive” infections, cytologically and histology recognized as low-grade squamous intraepithelial lesion (LSIL) or low-grade cervical intraepithelial neoplasia (CIN) 1, or “transformative” infections, recognized as high-grade squamous intraepithelial lesion (HSIL) or high-grade CIN2 and 3. Productive infections show morphologic (low-grade) dysplastic features but lack the altered viral gene expression characteristic of transformative infections necessary for carcinogenesis [8].

In order to improve the sensitivity of cervical cytology testing, several ancillary techniques have been developed largely based on the strong link between cervical neoplasia and HPV infection. Incorporation of these techniques has allowed for more accurate triage and management of those patients at highest risk of having high-grade lesions of the cervix.

The proven causative link between hr-HPV and cervical cancer has led to the incorporation of hr-HPV testing into many cervical cancer screening protocols. In several countries, cytology-based cervical cancer screening has been replaced or is on the verge of being replaced, by primary hr-HPV testing. High-volume longitudinal studies have established the superior sensitivity of hr-HPV testing as a screening method compared to Pap cytology alone for detection of cancer and high-grade precursor lesions. Primary hr-HPV testing has a reported sensitivity of 90–95% for ≥CIN 3 lesions, compared to 30–87% for cytology [10-13]. The Canadian Cervical Cancer Screening Trial (CCCaST), the first North American randomized control trial to compare the screening performance of hr-HPV testing versus cytology, not only proved the greater sensitivity of hr-HPV testing but also showed the screening interval for hr-HPV-negative women could be safely extended from 3 to 5 years [13]. As opposed to cytology alone, a negative hr-HPV test has been shown to provide greater reassurance that a high-grade premalignant lesion is absent, a finding which has been confirmed by a number of large international clinical trials [14-19].

The Kaiser Permanente Northern California (KPNC) cervical screening program represents one of the largest and longest experience with routine hr-HPV testing in North America, with over 1 million women (ages 30–64) enrolled in the study between 2003 and 2012 [16, 17]. The study aimed to determine whether a negative screening hr-HPV could provide the same reassurance of the absence of a high-grade squamous lesion as a negative co-test (Pap cytology and hr-HPV testing) or negative Pap cytology alone. Women were screened with concurrent Pap cytology and hr-HPV testing, with a screening interval of 3 years used if hr-HPV was negative and/or Pap cytology was negative. Risk analysis revealed a lower 3-year risk of developing ≥CIN3 with a negative hr-HPV alone result (CIN3- 0.07%; cancer- 0.011%) than the 3-year risk following a negative Pap cytology alone test (CIN3- 0.19%; cancer- 0.02%) and the 5-year risk following a negative co-test result (CIN3- 0.11%; cancer- 0.014%). These results provided evidence that hr-HPV co-testing every 5 years is at least as safe as cytology alone every 3 years. A 5-year screening interval using a primary hr-HPV screening paradigm, with cytology examination used only as a reflex test for positive results, would be projected to reduce the number of Pap cytology tests performed by up to 50%. Important, HPV testing missed far fewer cervical adenocarcinomas and adenocarcinoma in situ compared to Pap cytology (11.4% vs. 47.4% and 18.7% vs. 51.1%, respectively), adding to the benefits of hr-HPV screening.

The incidence and mortality associated with cervical cancer have a variable global distribution, due in part to the lack of cohesive national screening programs and limited resources in some countries. The Latin American Screening study (LAMS study), a large European Commission-funded multicenter screening trial with over 12,000 women enrolled between 2002 and 2003, aimed in part to evaluate the feasibility of different available screening tools in a medium-to-low resource setting, including hr-HPV, in women of varying risks for cervical cancer at 4 different regional clinics in Brazil and Argentina [20]. A subgroup of women with normal baseline cytology and negative or positive hr-HPV testing at the time of enrollment were prospectively followed for 2 years to assess for the subsequent development of cervical abnormalities [21]. hr-HPV-positive women at baseline had a significantly higher incident rate of LSIL (IR = 3.5%, RR = 1.4 [95% CI: 1.1–1.7]) and HSIL (IR = 0.7%, RR = 1.5 [95% CI: 1.4–1.7]) at 12-month follow-up, with an increase in the incidence rate of HSIL to 2.1% seen at 24 months (RR = 1.7 [95% CI: 1.5–1.9]). No HSIL was observed among hr-HPV-negative women at 24 months. In another follow-up study published in 2006, hr-HPV testing performed on a cohort of 4,284 women revealed an overall high HPV prevalence of 17.1%, which decreased with age (33.9% in women younger than 20 years old to 11% in women older than 41 years old) [22]. The detection of hr-HPV increased in parallel with progressive cytologic abnormalities, and in agreement with other studies, hr-HPV testing was found to be highly sensitive for high-grade cervical lesions (≥96.5% for CIN3+). The overall specificity of detecting histologically confirmed cervical lesions was 85%. Subsequent data from the combined LAMS and New Independent States of the Former Soviet Union (NIS) cohorts (>15,000 women) demonstrated hr-HPV testing alone was the overall most sensitive screening test, with a slight improvement seen in women ≥35 years (79% [95% CI: 71.2–86.6%] in women <35 years vs. 85.1% [95% CI: 78.4–91.8%] in women ≥35 years), but the least specific stand-alone test regardless of age group [23]. hr-HPV testing combined with Pap smear (LSIL+) demonstrated the highest specificity in women ≥35 years (98.7% [95% CI: 98.3–99%]) but was similar to that of Pap smear alone (98.2% [95% CI: 97.8–98.6%]). In women <35 years, Pap smear (LSIL+) was more sensitive as a stand-alone test than when combined with hr-HPV testing, with only a minimal gain in specificity with a combined testing algorithm. The overall lack of significant improvement in specificity with the addition of hr-HPV testing is an especially important consideration in countries with limited economic resources, which may impart unacceptably high incremental costs for confirmatory testing.

Although large global studies have largely agreed that hr-HPV testing lacks the specificity of cytology screening, extending the screening interval after a negative HPV test may improve the efficiency of screening programs. Follow-up data from the POBASCAM trial (Population Based Screening Study Amsterdam), which began enrollment in the Netherlands in 1999, has shown that stratifying women based on previous screening results is a feasible approach to improve testing efficacy. In a follow-up study, 18,448 women with a negative HPV result at the second round of study-related screening (5 years after enrollment) were risk-stratified according to screening results at the time of study enrollment. The 14-year CIN3+ risk among HPV-negative women was 0.48% (0.43% when prior HPV test was negative) but was significantly higher when the prior HPV test was positive (2.36%). This risk was not increased if the prior cytology result was positive. Given that a prior HPV-positive test is associated with an increased future risk of cervical neoplasia, even if the current HPV result is negative, these data suggest that HPV results from multiple screening rounds should be considered when implementing screening intervals [24].

Several hr-HPV testing platforms are available for cervical cytology specimens. hr-HPV can be identified by either viral DNA or mRNA detection (Table 1). A summary of key characteristics of each testing platform can be found in a recent review by Salazar et al. [25]. The molecular basis of hr-HPV testing provides highly reproducible and objective results. Additional advantages include the ability to achieve high-throughput testing and the ability to be applied to self-sampled material (brush- and lavage-based), which may aid in improved screening participation of nonresponders [26, 27]. One important limitation of hr-HPV testing, however, is the inability to distinguish between transient and persistent (clinically relevant) infections. The high sensitivity and lower specificity of a stand-alone hr-HPV-based screening paradigm would therefore lead to a higher number of women undergoing unnecessary colposcopy. To mitigate this, several triaging strategies for hr-HPV-positive women have been suggested, including HPV genotyping, dual p16/Ki-67 staining, and methylation assays (Table 2).

Cervical infections with HPV 16 and HPV 18 confer the highest risk of developing cancerous lesions. HPV 16 is associated with 50–60% of all cervical cancers, while HPV 18 is found in 10–15% of cervical cancer cases [28-30]. hr-HPV genotypes 16, 18, and 45 together account for up to 94% of all HPV-associated cervical adenocarcinomas (most common with classic subtype), with recent studies now suggesting that HPV 16 is likely the predominant genotype [6, 28]. Persistence of HPV infection for 1–2 years is a strong predictor of developing a high-grade intraepithelial lesion or worse (≥CIN 2) [31]. Several prospective studies have shown that women with HPV 16 or HPV 18 infections have a significantly higher risk of developing high-grade cervical lesions within a shorter follow-up period than women infected with other HPV genotypes [32, 33], which in part may be due to the ability of these genotypes to integrate earlier into the human genome [34]. Kjaer et al. [31] performed one of the first large, population-based studies to determine the long-term absolute risk of developing high-grade CIN with persistent infection with individual hr-HPV genotypes. More than 25% of HPV 16-positive women with normal cytologic findings at the beginning of the study developed ≥CIN 3 within 12 years, representing a five-fold increased risk compared to other hr-HPV types. The added benefit of HPV 16/18 genotyping was confirmed by the HPV FOCAL trial: a large randomized control trial aimed at establishing the efficacy of hr-HPV screening with LBC triage compared to LBC screening with hr-HPV testing triage in women with atypical squamous cells of undetermined significance (ASC-US) [35]. hr-HPV-positive women with abnormal LBC (ASC-US or worse) were referred for immediate colposcopy, while women with normal cytology underwent repeat LBC with concurrent hr-HPV testing. Genotyping was performed on all hr-HPV-positive baseline samples. Ninety-four percent of ≥CIN 2 lesions detected during the screening period were hr-HPV-positive at baseline. Moreover, 17% of cases identified as ≥CIN 2 were HPV 16- or HPV 18-positive at baseline with initial normal cytology, highlighting the benefit of immediate referral for colposcopy in women with these genotypes.

The ATHENA study (Addressing THE Need for Advanced HPV diagnostics) is one of the largest clinical trials to date aimed at studying the efficacy of HPV 16/18 genotyping as triage of hr-HPV-positive women compared to LBC for cervical cancer screening [36, 37]. In a subgroup analysis (women age ≥25 years old), the detection of HPV 16, HPV 18, or both in hr-HPV-positive women was at least equivalent to ASC-US or worse on LBC evaluation for triage to immediate colposcopy (comparable sensitivity and positive predictive value). Among women who had colposcopy, hr-HPV testing with genotyping was more sensitive than LBC for detection of ≥CIN 3. A cost-effectiveness analysis study performed by Huh et al. [38] demonstrated hr-HPV testing with HPV 16/18 genotyping and reflex cytology was not only superior in cost-effectiveness compared to other screening modalities but also led to earlier diagnosis of ≥CIN 3 at initial screening and overall decreased cancer incidence.

The long-term impact of HPV vaccination on hr-HPV prevalence and cervical disease is unknown. The Compass pilot randomized trial in Australia, which was one of the first countries to establish a national HPV vaccination program in 2007, reported increased detection of high-grade cervical lesions with primary HPV screening and genotyping in a population with a high HPV vaccination rate in women ≤33 years [39]. The increased detection rate of CIN2+ with 5-yearly HPV-based partial genotype screening (HPV 16/18, other) compared to 2.5-yearly LBC screening is consistent with trials conducted in non-vaccinated populations.

Another large retrospective study comparing the performance of genotyping versus cytology triage of hr-HPV-positive women demonstrated significant differences, with PPV for ≥CIN2 of 20.9% for cytology diagnoses ≥ASC-US, as compared to 31.8% for HPV 16+, 27.8% for HPV 18+, 30.8% for HPV 16+ and/or HPV 18+, and lastly 16.6% for HPV-negative genotype results [40]. Similarly, as part of the FRIDA study group in Mexico, a study comparing HPV 16/18 genotyping with LBC in the study population demonstrated improved sensitivity (58.3% vs. 42.9%), particularly with HPV 16/18 genotyping followed by reflex LBC (86.6%) [41]. These findings further support the utility of genotyping as part of the triage of hr-HPV-positive women.

Taken together, these studies suggest that risk estimates for specific genotypes may be utilized for triage of hr-HPV-positive women to help predict which patients are at highest risk of developing (pre)cancerous lesions and warrant immediate referral for colposcopy. Currently, the guidelines of the American Society for Colposcopy and Cervical Pathology (ASCCP) include HPV 16 and HPV 18 genotyping for women 30 years or older who are hr-HPV positive and have a normal cytology evaluation; if positive for HPV 16 or HPV 18, colposcopy is recommended [42]. The detection of HPV 16/18 may provide a more objective risk measurement than LBC alone, with the added benefit of faster throughput of results, and can be performed on self-sampled material. The Cobas 4800 HPV test and the BD Onclarity HPV assay both provide specific genotyping for HPV 16/18 as part of the testing platforms; genotyping for HPV 16/18 may be added as a separate test for the Cervista HPV HR and Aptima HPV assays. The management of hr-HPV types other than HPV 16/18 remains to be determined.

Our detailed knowledge of the pathogenesis of HPV infection in cervical cancer has made p16INK4a (p16) immunohistochemistry (IHC) an effective indirect method for detecting transformative HPV infections. p16, a cyclin-dependent kinase inhibitor involved in normal cell cycle regulation, is overexpressed when an HPV infection is established in the squamous basal cell layer through the E7/retinoblastoma pathway. Antibody detection of p16 overexpression in cervical cells, evident by accumulation of the p16 protein in the cell nucleus and cytoplasm, is a sensitive surrogate marker for transcriptionally active hr-HPV. p16 expression, however, is not limited to only dysplastic cells; expression of p16 can be seen in normal cervical squamous metaplastic cells, endocervical glandular cells with metaplastic changes (particularly tubo-endometrioid metaplasia), and endometrial cells. The addition of Ki-67 staining, a nuclear cell proliferative marker, to p16 improves the ability to distinguish normal from abnormal p16 expression. Simultaneous IHC expression of p16 and Ki-67 within a cell can be regarded as a marker of cell cycle dysregulation, indicating a transformative hr-HPV infection, whereas singular expression of either is interpreted as physiologic [43]. p16/Ki-67 IHC may be applied to unstained slides or cell block sections prepared from residual LBC samples.

The triage of hr-HPV-positive women with p16/Ki-67 dual staining has been shown to have a high sensitivity and increased specificity compared to cytology evaluation alone [44-46]. In 2015, Wentzensen et al. [45] published the results of a large population-based study evaluating the performance of p16/Ki-67 dual staining for detection of cervical precursor lesions in hr-HPV-positive women undergoing screening at Kaiser Permanente Northern California. The dual stain was shown to have similar sensitivity (83% vs. 77%) and higher specificity (59% vs. 50%) compared to cytology alone for detection of ≥CIN 2 lesions. The cumulative 5-year risk of ≥CIN2 in that population was subsequently analyzed in a separate study and it was shown that positive dual-stain results were associated with a significantly higher cumulative risk as compared to abnormal cytology results. Conversely, women with negative dual staining had a significantly lower cumulative risk as compared to women with normal cytology results, and their risk for ≥CIN2 remained below the colposcopy referral threshold for all 5 subsequent years [47]. A retrospective sub-study of patients from the ATHENA trial found dual-staining cytology to be significantly more sensitive than cytology alone (74.9% vs. 51.9%; p < 0.0001) with comparable specificity (74.1% vs. 75%) [48]. Conversely, Zhang et al. [49] demonstrated significantly increased specificity of dual staining (68.33%) as compared to cytology (38.33%) and hr-HPV testing (21.67%) for the detection of ≥CIN2 lesions (p < 0.05), while having similar sensitivity as compared to hr-HPV testing. The results of a large prospective European study comparing p16/Ki-67 dual-stained cytology with Hybrid Capture 2 (HC2) HPV testing demonstrated comparable sensitivity for CIN2+ and ASCUS, with higher specificity in both ASCUS and LSIL cases. Positive predictive values for CIN2+ were substantially higher for dual-stained cytology as compared to HC2 HPV testing [50]. These findings demonstrate p16/Ki-67 dual staining cytology could provide a more reproducible and less subjective alternative to Pap cytology for detection of ≥CIN 3 lesions, although long-term studies have yet to be performed. A model-based study from Belgium demonstrated the cost-effectiveness of this triage approach [51], and the possibility of automation through digital interpretation of p16 staining has already been described in paraffin-embedded tissue with CIN [52]. Similarly, p16/Ki-67 dual staining cytology was clinically validated as part of the IMPACT trial, a prospective observational study in the USA. Their results demonstrated significantly higher sensitivity and specificity for the detection of CIN2 or worse in HPV-positive women, as compared with HPV 16/18 genotyping in combination with cytology triage, and higher sensitivity but lower specificity when compared to cytology alone [53]. Interestingly, a study evaluating the clinical performance of LBC, p16/Ki-67 dual staining, and HPV 16/18 genotyping as triage strategies in HPV-positive women, demonstrated that the triage modality with the highest sensitivity and lowest negative predictive value for the detection of ≥CIN2, was a two-step triage including HPV 16/18 genotyping followed by LBC of hr-HPV non-16/18 + women [54].

The commercial availability of a dual staining kit (CINtec®PLUS Cytology; Roche Diagnostics, Indianapolis, IN, USA) as well as the relatively low technical difficulty as compared to molecular-based testing, would make this a reasonable option in places of limited financial and technical resources. CINtec®PLUS Cytology is the first FDA-approved biomarker test that can be used for the triage of hr-HPV-positive women in the US (received approval in 2020) and is intended to be used in combination with the Cobas® HPV test, as part of either primary screening or co-testing paradigms. Limitations of the dual stain include the need for adequate specimen cellularity, limited throughput of results, and the lack of applicability to self-sampled specimens.

DNA methylation is an enzyme-induced chemical modification of cytosine-guanine dinucleotide-rich areas (CpG islands) in human gene promoter regions, leading to normal epigenetic functional changes to the genome. Aberrant gene promoter methylation is one important mechanism for loss of gene function during carcinogenesis, often occurring as an early event and frequently present in precursor lesions of various types of cancers [55, 56]. The role of abnormal methylation patterns of both viral and host DNA in the multistep pathogenesis of cervical cancer has been extensively studied, with the goal of identifying specific epigenetic events that could predict carcinogenesis.

Although the HPV genome lacks classical CpG islands, high density CpG sites do exist, suggesting a possible functional role [57]. The exact means by which targeted HPV DNA methylation occurs, however, remains largely unknown. Methylation may represent: (1) a mechanism by which HPV switches from a productive to transformative infection, (2) a result of host defense mechanism for preventing viral replication and silencing transcription, or (3) a method of evading host immune recognition. Moreover, methylation of the HPV DNA genome has been shown to exhibit type-specific variation within the viral life cycle, leading to reported inconsistencies seen in earlier studies. Only until recently, when the link between methylation of specific combinations of CpGs within late regions of the HPV genome and cervical disease was made, has there been a consistent and reliable correlate between disease severity and HPV DNA methylation levels, with several studies suggesting that late region methylation may be indicative of viral genome integration [58, 59]. Specific patterns of HPV late gene methylation may, in turn, be predictive of the likelihood of persistence versus eventual clearance of infection, providing possible prognostic value to triaging of hr-HPV-positive women. Genome sequence methylation has been the most comprehensively studied in HPV 16, in which hypermethylation of the L1, L2, E2, and E4 regions has been shown to be associated with an increased risk of ≥CIN 2 lesions and persistent infections, with a reported sensitivity and specificity of approximately 91% and 60%, respectively [59, 60].

Evaluation of host DNA methylation markers has shown similar promising results, with a trend toward increased methylation seen with progressive pathological changes [61, 62]. Methylation of the host cell genome is a common epigenetic event in cervical carcinogenesis. Numerous studies have focused on the role of abnormal promoter methylation of tumor suppressor genes known to be frequently associated with cervical carcinomas, although establishing marker panels that are both highly sensitive and specific for cervical neoplasias has proved challenging. The bi-marker panel CADM1/MAL was one of the first to show a relationship between methylation levels and degree of cervical dysplasia and duration of hr-HPV infection, conferring a sensitivity of approximately 70–89% for CIN 3 or worse, when combined with cytology evaluation [63-65]. Although the increased sensitivity correlated with a decreased mean time to diagnosis, this is at the expense of decreased specificity, resulting in more colposcopy referrals. Combined MAL/miR-124-2 has shown similar results [66]. Methylation levels of FAM19A4 have been shown to correspond to CIN lesion volume [67]. The use of the bi-marker panel FAM19A4/miR-124-2 in cervical self-samples resulted in a sensitivity of approximately 70% and a specificity of 68–76% for detecting ≥CIN 3 lesions [68]. Analysis of a DNA methylation HPV triage classifier that includes both host target region EPB41L3 and viral late gene regions of HPV 16, 18, 31, and 33 demonstrated better overall sensitivity and specificity than genotyping in the identification of ≥CIN2 lesions in hr-HPV-positive women [69].

DNA methylation assays may serve as a novel risk stratification tool and alternative triage method for women who are hr-HPV positive. Advantages include objective and reproducible results, with a potential for high-throughput. The assays can be easily performed on a variety of cervical samples, including LBC specimens and histologic specimens. Adding HPV genotyping to this triage method has been shown in limited studies to increase specificity, and portends the potential of a full molecular screening paradigm [70, 71]. However, the efficacy of DNA methylation assays, either as a stand-alone test or combined with other triage methods, has yet to be established with large population studies. Further evaluation is needed to validate the integration of methylation assays into clinical practice.

hr-HPV testing is an important tool in cervical cancer screening. The best means by which it should be incorporated into clinical practice, whether as an adjuvant test or stand-alone replacement of Pap cytology, continues to be debated. In many parts of the world, it has already replaced cytology as the primary screening method and in that context the need for triaging hr-HPV-positive women to avoid high numbers of unnecessary colposcopy referrals is apparent. Currently, Pap cytology, HPV 16/18 genotyping, and CINtec®PLUS Cytology are the only FDA-approved triaging methods in the USA. Although studies have demonstrated the utility of a variety of other immunohistochemical and molecular techniques for the triage of hr-HPV-positive women, most data available is relatively limited and further studies are still required to determine the routine applicability and utility of these tests. In some countries, many with established HPV vaccination programs, these ancillary techniques are already been used. However, the impact of the use of these strategies in HPV-vaccinated and non-vaccinated populations is still being determined. Even in low-to-middle development index countries, with strained health care systems and significant issues related to patient access to care, it has been shown that hr-HPV testing performance is comparable between provider-collected and patient self-collected samples. Evidently, many of these ancillary techniques may play an important role in the future of cervical cancer prevention across the globe, particularly in places of difficult access [72-74].

Ricardo R. Lastra has performed consultancy work for ArsenalBio, Inc. Julieta E. Barroeta and Andrea D. Olivas have no conflicts of interest to declare.

The authors have no conflicts of interest to disclose.

None.

Dr. Olivas: preparation of manuscript draft and review of references; Dr. Barroeta, Dr. Lastra: review of manuscript draft, preparation of final manuscript, and review of references.

This is a review article that does not include new data.