Strategies for screening and early detection of anal cancers: A narrative and systematic review and meta-analysis of cytology, HPV testing, and other biomarkers

Abstract

Anal cancer incidence and mortality have been increasing over the past decade. While the incidence in the general population remains low, it is much higher in certain subgroups including those living with human immunodeficiency virus (HIV), and men who have sex with men (MSM). Approximately 90% of anal squamous cell cancers are caused by infection with carcinogenic human papillomavirus (HPV). Given the common etiology between anal and cervical carcinogenesis, screening for anal cancer has been proposed in certain high-risk populations using strategies adapted from cervical cancer prevention. In this review, we discuss important differences in anal and cervical cancer regarding the populations at risk, disease natural history, and clinical procedures and outcomes that need to be considered when evaluating strategies for anal cancer screening. We also performed a systematic review and meta-analysis of the performance of anal cytology, anal HPV testing, and various biomarkers for detection of anal precancers and cancers. We summarize the implications of these performance estimates in the context of risk-based screening and management of anal precancers and we highlight important research gaps that need to be addressed to fully understand the benefits and harms of anal cancer screening.

I. Strategies for anal cancer screening and early detection

Anal cancer epidemiology

Anal cancer is rare in the general population, with approximately 8,580 new cases and 1,160 deaths estimated for 2018 in the United States. Incidence and mortality in the United States have been increasing about 2.2% and 2.9% per year over the past decade, respectively[1]. Anal cancer is most commonly diagnosed in individuals aged 55-64 years, and is slightly more common in women compared with men (2.1 vs. 1.5 per 100,000, respectively). While anal cancer incidence is too low for general population screening, there are specific well-defined subgroups with much higher incidence, including those living with HIV, men who have sex with men (MSM), women with previous HPV-related disease, and immunosuppressed populations (non-HIV related). HIV-positive MSM have the highest risk of developing cancer, with incidence rates ranging from 77 to 137 per 100,000[2, 3], exceeding incidence rates of cervical cancer in some countries without screening[4]. The excess burden of anal cancer in these populations is in part attributed to the higher prevalence of human papillomavirus (HPV). HPV vaccination has shown good efficacy for primary anal cancer prevention[5, 6]; however, the vaccine is only recommended for men up to 27 years of age, and uptake has been limited[7]. Given the high incidence of anal cancer in certain high-risk groups such as HIV-positive MSM, there has been interest in developing screening and surveillance strategies targeting these populations. The common etiology between anal and cervical carcinogenesis has prompted evaluations of cervical cancer screening and prevention approaches in anal cancer. However, there are important differences in the populations at risk, disease natural history, and clinical procedures and outcomes that need to be considered when evaluating strategies that have been introduced for cervical cancer prevention.

Natural History of Cervical Cancers in Comparison to Anal Cancer

A functional progression model that has been established for cervical cancer includes three major steps (Figure 1A): HPV infection in the cervical transformation zone, followed by persistence and progression to precancer, and invasion to cancer.[8, 9] Almost all cervical cancers are caused by carcinogenic HPV infections. HPV infections are common in sexually active women, but most newly acquired HPV infections become undetectable after few months. A small subset of infections persists and may progress to cervical precancers (cervical intraepithelial neoplasia grade 3, CIN3). Without intervention, about a third of cervical precancers will ultimately invade after 20-30 years.[10]

Figure 1. Disease model and population prevalence of HPV infection and precancer.

Cervical cancer arises through three major steps: HPV infection, followed by persistence and progression to precancer, and invasion to cancer (A). The peak prevalence of HPV infection follows initiation of sexual activity. The secondary peak of precancer occurs several years later, depending on screening, whereas the peak or plateau of cancer occurs about 30 years after the initial HPV infection peak (B). Anal cancer develops through characteristic epithelial precursors like cervical cancer, suggesting that the cervical cancer progression model may similarly apply to anal cancer. In contrast to cervical HPV infection, anal HPV prevalence in all MSM (C) and in HIV-positive MSM (D) is elevated across most age groups. Similarly, anal precancers do not show a characteristic age peak. Green line = HPV prevalence; blue line = precancer prevalence. The age-prevalence curve for cervix was adapted from Schiffman & Wentzensen, 2013.[8] The age-prevalence curves for the anal region are summarized from several sources.[90–94]

Approximately 90% of anal squamous cell carcinomas (SCC) are attributed to infection with high risk human papillomavirus (HPV).[11] The anus, like the cervix, has a transformation zone where squamous epithelium of the anus transitions to columnar epithelium of the rectum which is susceptible to HPV infection and subsequent transformation.[12] Anal cancer develops through characteristic epithelial precursors like cervical cancer, suggesting that the cervical cancer progression model may similarly apply to anal cancer. High risk HPV infection alone is not sufficient for anal carcinogenesis, but determinants of HPV persistence and progression to precancer are not well understood. High-grade AIN (AIN2/3) is considered the precursor to invasive anal SCC. Very few studies have evaluated the progression rate of AIN2/3 to anal cancer, but it has been suggested that progression from anal precancer to cancer may be less common compared to cervical precancers.[13] However, there is a lot of uncertainty around the invasion estimates for both cancer sites.[14, 15]

Prevalence of Cervical and Anal HPV Infection and Precancer in the Population

The individual steps of the cervical cancer progression model have very characteristic age distributions in the population (Figure 1B). HPV infections peak around age 20, shortly after initiation of sexual activity. Cervical precancers peak about 10 years later. The highest prevalence of invasive cancers in the population occurs about 30 years after the initial infection peak, at around age 35 to 55 years in unscreened populations, indicating that the development of cancer from an initial infection typically takes decades.[8, 9] In stark contrast to that, anal HPV prevalence in MSM (Figure 1C) and in HIV-positive MSM (Figure 1D) is high across most age groups.[16, 17] Similarly, anal precancers do not show a characteristic age peak (Figure 1C+D) and diagnosis of anal cancer peaks from about age 45 to 64 years.[18] Among those with normal anal cytology, HPV prevalence is approximately 42% and 59% in HIV-negative and HIV-positive women, respectively, and 57% and 76% in HIV-negative and HIV-positive men, respectively.[19] In contrast, in HIV-negative men who have sex with women, HPV prevalence is much lower (6%). The prevalence of HPV infection increases with worsening cytology diagnosis, reaching over 90% in HSIL and nearly 100% in anal cancer among HIV-positive women and men, respectively.[19] These different patterns of cervical and anal HPV natural history may reflect both biological (e.g. less effective immune control, particularly among HIV-positive MSM) and behavioral factors (e.g. high levels of acquisition in older age groups, less screening and removal of earlier precancers).[20]

HPV Genotype Distribution in Cervical and Anal Disease

The risk of cervical cancer varies strongly across different HPV genotypes. Most are caused by HPV16, followed by HPV18 which together account for about 75% of cervical cancers.[21] The next 5 carcinogenic types (that are included in the nonavalent vaccine)[22] account for an additional 15% while the remaining types cause only very small proportions of cervical cancers (Figure 2). The distribution of HPV genotypes varies strongly by cervical disease stage; the proportion of HPV16 and 18 increases substantially with higher grade of disease.[23]

Figure 2. HPV genotype attribution in cervical and anal precancers and cancers by HIV status.

The proportion of high-grade lesions and cancers attributed to HPV16 (blue), HPV18 (orange), and combined HPV31/33/45/52/58 (grey) are shown by anatomical site and HIV status, with anal separately shown for women and MSM. HPV prevalence data for cervical and anal lesions was adapted from several large reviews. [19, 21, 23, 95–97]

Recently, a large systematic review and meta-analysis reported type-specific HPV prevalence across different anal diagnoses, by sex and HIV status.[19] HPV16 specific prevalence increased with the severity of anal cytology diagnosis in all subgroups, with a large proportion of cancers (~85%) caused by HPV16 (Figure 2). In contrast, the prevalence of HPV18 and HPV types 31/33/45/52/58 increased from normal to HSIL cytology, but decreased in anal cancers in all subgroups. Approximately 85% of HIV-negative anal cancers were positive for HPV16, the majority of which were single infections. In contrast, only 70% of HIV-positive anal cancers were positive for HPV16, with about one-third of those positive for multiple infections with other high-risk HPV types.[19] It has been hypothesized that HPV16 is more capable of evading host immune control than other carcinogenic types. In subjects with impaired immune function, other genotypes may persist longer and cause precancers or cancers more often compared to immune-competent individuals.[24–26] [27] This could explain a lower relative proportion of HPV16-positive cancers in HIV-positive compared to HIV-negative men and women (Figure 2).

Cervical and Anal Cancer Screening Approaches

Cervical cancer screening is highly efficacious because it detects cervical precancers that can be treated to prevent invasive cancers. At the population level, this is accomplished by identifying a small subset of all screened women who are at increased risk of cervical cancer and need additional workup and possibly treatment.[28]Cervical cytology-based screening has led to dramatic decreases of cervical cancer incidence and mortality over the past decades in many developed countries, and remains the most widely used screening approach.[29] Recently, HPV testing has been introduced in several countries, in addition to, or replacing cervical cytology as primary screening test.[29] HPV testing provides great reassurance against developing cervical precancer and cancer in women who test negative, allowing extended screening intervals.[30] HPV-positive women are at increased risk of cervical precancer, however most infections are transient and do not require further evaluation. Since it is not feasible to refer all HPV-positive women to colposcopy, additional tests (triage tests) are used to identify women at highest risk of cervical precancer.[28]

Currently, there are no randomized clinical trials documenting the effectiveness of an anal screening program to reduce anal cancer incidence, morbidity, or mortality.[31] Previous observational studies have evaluated anal cytology and HPV testing for detection of anal precancers. HPV testing is recommended for cervical cancer screening in women 25 years and older because HPV prevalence decreases and precancer prevalence increases, making HPV screening efficient to reassure the vast majority of women of a low risk of precancer and cancer.[32] In contrast, HPV prevalence is much higher in HIV-positive MSM, and it does not decrease at older age, resulting in high test positivity and low specificity.[33] As a result, much fewer patients can be reassured of a low risk of anal precancer and cancer. We do not know at what HPV prevalence primary HPV testing would be efficient, but the benefit is greatly reduced when most men test positive. It is also not clear how long a negative anal HPV test provides reassurance against anal precancer and cancer.

Management of women who test positive in cervical cancer screening relies on colposcopy as the diagnostic gold standard. It has been demonstrated that colposcopy is subjective and that detection of disease strongly depends on how many biopsies are taken.[34] Recently, colposcopy guidelines were established to improve reproducibility and accuracy of colposcopy and biopsy in the US.[35, 36] Similar to cervical colposcopy, high-resolution anoscopy (HRA) involves visualization of the anal canal and perianus using a colposcope with acetic acid and Lugol iodine to identify anal lesions.[37] Anatomical differences between the anus and the cervix including a larger surface area and anal mucosal folds that can hide lesions, make HRA more challenging than colposcopy.[38] A higher prevalence of coexisting pathology (e.g., condyloma, hemorrhoids, treatment-related scar tissue) may also obscure visualization during HRA. Attempts have been made to standardize HRA by the International Anal Neoplasia Society[39], but training and techniques vary across different providers from diverse specialties including infectious disease, HIV and sexual health, colorectal surgery, gynecology, family medicine, and dermatology.

Treatment of Cervical and Anal Precancers

Any public health or clinical intervention like population screening or surveillance of subjects at high risk of disease needs to weigh the benefits of the intervention against the harms.[14] The benefit of cervical cancer screening comes at the cost of treating many cancer precursors that would not progress to invasive cancer. However, since the treatment, excision of the cervical transformation zone, is relatively benign, overtreatment is widely accepted. Obstetric complications are the most important adverse event, but they depend on the depth of excision and the risk can be safely reduced in most women by performing less deep excisions.[40, 41]

An important difference to cervix is that excisions in the anal canal are associated with more adverse outcomes that can have long-term complications and affect anal sphincter functions.[42–44] The decision to treat AIN2/3 and the type of treatment depends on factors such as the size and the location of the lesion. Unlike cervical precancer, smaller lesions tend to be treated more often than larger lesions, owing to the ease of treating these lesions with topical therapy or excision. In contrast, surgical resection is usually required for larger lesions, which can involve substantial morbidity including pain, anal stenosis, and incontinence and a high likelihood of recurrence.[45] Patients with larger lesions are usually closely monitored every 4 to 6 months with repeat HRA to identify signs of progression to early stage anal cancer. Treatment of intra-anal lesions is more difficult than of lesions located on the perianal surface, and therefore lesions located within the anal canal have a higher threshold for treatment, particularly if they are large; circumferential intra-anal lesions are generally not treated.[45] These anatomical characteristics can vary by HIV status; large lesions tend to be more common in HIV-positive patients and have a higher rate of recurrence compared with HIV-negative patients.[46, 47]

II. Systematic review of cytology, HPV testing and other biomarkers for detection of anal precancer

The availability of various screening and triage markers for cervical precancers has prompted evaluation of these assays for detection of anal precancer and cancer in various populations. In the following sections, we describe systematic reviews and meta-analyses of anal cytology, anal HPV testing, p16 or p16/Ki-67 dual stain, and HPV mRNA testing for detection of anal precancers and cancers. The aims of this systematic review and meta-analysis were to assess the diagnostic accuracy (sensitivity and specificity) of the performance of these biomarkers for anal cancer screening and to determine whether these estimates vary by HIV status.

Methods

We conducted this systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Supplementary Figure 1). We included original studies with primary data reporting the performance of screening tests and biomarkers for detection of anal precancers and cancer. Our search was limited to English-language, peer-reviewed studies published before February, 2018. We searched the MEDLINE database via PubMed and EMBASE using search terms listed in the Supplemental Material. Titles and abstracts of identified articles were independently screened for inclusion and full-text versions of eligible articles were reviewed by MC and NW to determine eligibility. We reviewed the reference lists of articles identified in the primary search for additional relevant studies. Only studies evaluating anal precancer/cancer as the main outcome(s) were included. Additionally, we excluded studies that did not assess screening performance, and those with significant verification bias, where receipt or interpretation of the diagnostic test (e.g., high-resolution anoscopy and/or biopsy) was dependent on the results of a prior clinical test (e.g., anal cytology). In the case of sequential or multiple publications where there was a possibility of overlapping data, only data from the most recent publication were included. While we did not restrict our search to specific populations, most included studies were conducted among MSM and HIV-positive MSM, and thus we were not powered to analyze performance estimates in women.

To estimate diagnostic accuracy for anal precancers and cancers, we extracted information on the number of true positives, false negatives, false positives, and true negatives according to each test. Calculations of summary estimates required data from a minimum of four studies and were carried out using a random effects bivariate normal model for the sensitivity and specificity between studies, fit with the Stata metandi and gllam packages.[48–50] If the study reported separate performance estimates for more than one population (e.g., HIV+ MSM and HIV- MSM), we estimated sensitivity and specificity values for each reported population combined in the overall analysis and then separately in analyses stratified by HIV status. Sensitivity and specificity estimates for each study were plotted in a summary receiver operator characteristic (ROC) plot, where each point on the curve represents a different study, weighted according to sample size.

Results

Performance of anal cytology for detection of anal precancer

Anal cytology was first studied for detection of anal dysplasia in the 1990s and anal cytology was first included in the Bethesda cytology atlas in 2001.[51, 52] In the Bethesda system, anal cytology findings are classified similar to cervical cytology: atypical squamous cells of undetermined significance (ASC-US), atypical squamous cells cannot exclude high-grade, low-grade squamous intraepithelial lesion (LSIL) and high-grade squamous intraepithelial lesion (HSIL).[53]

A total of 14 studies reported on the performance of cytology for detection of anal precancers and cancers, including 881 with AIN2+ and 3,706 with <AIN2.[33, 54–65]. The distribution of abnormal cytology results in all 14 studies was 12% ASC-US, 18% LSIL, and 5% HSIL, with prevalence varying significantly by HIV status (Supplemental Figure 2). Among 9 studies with available data, the pooled proportion of inadequate cytology results was 3% (95% CI, 2-5%) (Supplemental Figure 3). Overall, the pooled sensitivity and specificity of ASC-US+ cytology for detection of AIN2+ was 77.3% (95% CI, 64.9-86.3%) and 55.5% (95% CI, 45.5-65.2%), respectively (Figure 3). For LSIL+ and HSIL+ cytology, the pooled sensitivity and specificity were 60.7% (95% CI, 41.6-77.0%) and 74.2% (95% CI, 61.6-83.8%) and 26.9% (95% CI, 10.0-52.4%) and 94.7% (95% CI, 85.4-98.2%), respectively. Among the 10 studies restricted to HIV-positive MSM, the pooled sensitivity and specificity of ASCUS+ was 80.8% (95% CI, 68.7-89.0%) and 54.0% (95% CI, 42.1-65.5%), respectively. For LSIL+ and HSIL+ cytology, the pooled sensitivity and specificity were 75.0% (95% CI, 55.4-87.8%) and 68.2% (95% CI, 50.5-86.8%) and 40.0% (95% CI, 15.1-71.5%) and 92.5% (95% CI, 76.0-97.8%), respectively. In the 4 studies restricted to HIV-negative MSM, the pooled sensitivity and specificity of ASCUS+ was 43.5% (95% CI, 14.2-78.3%) and 75.9% (95% CI, 61.0-86.4%), respectively. For LSIL+ and HSIL+ cytology, the pooled sensitivity and specificity were 23.6% (95% CI, 6.0-60.1%) and 88.1% (95% CI, 74.4-95.0%) and 6.0% (95% CI, 0.2-63.5%) and 98.3% (95% CI, 78.4-99.9%), respectively.

Figure 3. Clinical performance of anal cytology (ASC-US threshold, AIN2+ outcomes).

Meta-analysis plot of the diagnostic accuracy of anal cytology overall and in studies restricted to HIV-positive MSM. Three studies were split by HIV status (see Supplementary Table 1). The summary value for sensitivity and specificity (red square) is plotted with a 95% confidence region (dashed orange line). Individual studies are shown as hollow circles, with their size determined by the number of participants. The 95% prediction region (green dashed line) represents potential values of sensitivity and specificity that might be observed in a future study by describing the full extent of the uncertainty of the summary points, reflecting the between-study heterogeneity.

Performance of HPV testing for detection of anal precancers

Several HPV assays have been evaluated in anal samples from subjects at increased risk of anal cancer. These include signal amplification (hybridization-based) and target amplification (PCR-based) assays.[66] Some of these assays have received regulatory approval for use in cervical cancer screening, but there are currently no FDA-approved tests for anal cancer screening or surveillance. Large studies evaluating HPV DNA assays for cervical cancer screening have shown that the performance of these tests is very similar[67–72], allowing to pool data across these tests.

A total of 9 studies reported on the performance of any HR-HPV test for detection of anal precancers and cancers, including 671 with AIN2+ and 2,154 with <AIN2[33, 56, 57, 59, 61, 63, 64, 73, 74]. Overall, the pooled sensitivity and specificity of HR-HPV for detection of AIN2+ was 91.3% (95% CI, 78.9-96.7%) and 33.1% (95% CI, 22.2-46.3%), respectively (Figure 4). Among the 8 studies restricted to HIV-positive MSM, the pooled sensitivity and specificity was 95.4% (95% CI, 84.6-98.7%) and 23.8% (95% CI, 16.3-33.4%), respectively. We lacked a sufficient number of studies to evaluate the performance of HR-HPV testing in other populations.

Figure 4. Clinical performance of HPV testing (AIN2+ outcomes).

Meta-analysis plot of the diagnostic accuracy of HPV testing overall and in studies restricted to HIV-positive MSM. Two studies were split by HIV status (see Supplementary Table 1). The summary value for sensitivity and specificity (red square) is plotted with a 95% confidence region (dashed orange line). Individual studies are shown as hollow circles, with their size determined by the number of participants. The 95% prediction region (green dashed line) represents potential values of sensitivity and specificity that might be observed in a future study by describing the full extent of the uncertainty of the summary points, reflecting the between-study heterogeneity.

Genotyping for HPV16/18 has the potential to improve specificity of HPV testing for anal precancer detection. A total of 5 studies reported on the performance of HPV16/18 genotyping for detection of anal precancers and cancers, including 510 with AIN2+ and 1,562 with <AIN2. [59, 63, 64, 74, 75] Overall, the pooled sensitivity and specificity of HPV16/18 for detection of AIN2+ was 39.9% (95% CI, 22.4-60.5%) and 74.3% (95% CI, 67.3-80.1%), respectively. Among the 4 studies restricted to HIV-positive MSM, the pooled sensitivity and specificity was 41.3% (95% CI, 22.2-63.3%) and 68.5% (95% CI, 63.5-73.0%), respectively. We lacked a sufficient number of studies to evaluate the performance of HPV16/18 genotyping in other populations.

Other biomarkers

There are several promising biomarkers currently being evaluated for cervical cancer screening and triage that may have applications in screening and/or surveillance for anal cancer. Of these, only HPV E6 /E7 mRNA testing and immunostaining for p16 or p16/Ki-67 dual staining had a sufficient number of studies to be included in our review.

HPV E6/E7 mRNA

Four studies evaluated the performance of E6/E7 mRNA for detection of anal precancer and cancer, including a mix of HIV-positive and HIV-negative MSM (Supplementary Table 1).[59, 64, 75, 76] The pooled sensitivity and specificity of E6/E7 mRNA was 74.3% (95% CI, 68.3-79.6%) and 65.5% (95% CI, 58.5-71.9%), respectively (Figure 5). Performance estimates were similar across all studies.

Figure 5. Clinical performance of biomarkers (AIN2+ outcomes).

Meta-analysis plot of the diagnostic accuracy of HPV E6/E7 mRNA and p16 or p16/Ki-67 staining. One study was split by HIV status (see Supplementary Table 1). The summary value for sensitivity and specificity (red square) is plotted with a 95% confidence region (dashed orange line). Individual studies are shown as hollow circles, with their size determined by the number of participants. The 95% prediction region (green dashed line) represents potential values of sensitivity and specificity that might be observed in a future study by describing the full extent of the uncertainty of the summary points, reflecting the between-study heterogeneity. Studies that used p16/Ki-67 are marked with a red asterisk.

p16 or p16/Ki-67 Immunostaining

Four studies evaluated the performance of p16 staining (n=2) or p16/Ki-67 dual staining (n=2) for detection of anal precancer and cancer, including a mix of HIV-positive and HIV-negative MSM (Supplementary Table 1).[59, 64, 75, 77] The pooled sensitivity and specificity of p16/p16-Ki-67 was 56.6% (95% CI, 27.9-81.5%) and 62.3% (47.8-74.9%), respectively, with substantial heterogeneity observed between studies (Figure 5). There were only limited data to assess differences between p16 and p16/Ki-67 dual staining, but it appeared that the p16/Ki-67 dual stain is more accurate than p16 alone for detection of anal precancer and cancer.

Methylation

DNA methylation testing of both host and viral genes has been strongly associated with cervical precancer and cancer, and is a promising method for triage of HPV-positive women.[78–81] Similar findings have been reported for anal precancers and cancers [82]; however, there have been no studies assessing the clinical performance of methylation for anal cancer screening.

III. Application of biomarkers for screening, early detection and management of anal precancers

Detection of anal precancers may be possible with similar tools that have been established for cervical cancer screening. We systematically evaluated the performance of anal cytology, HPV testing, and biomarkers for detection of AIN2/3 and cancer. In table 1, we put these performance estimates into the context of anal cancer screening in HIV-positive MSM. Our systematic review and meta-analysis demonstrates that the performance of anal cytology differs from cervical cytology in a screening population, both with regard to sensitivity and specificity. This is due to the higher burden of HPV infection and higher degree of disease severity, particularly in HIV-positive MSM. Performance characteristics of a diagnostic test (i.e., cytology) are often thought to be portable between populations with different disease prevalence. However, the combination of a subjective diagnostic test (cytology) with an imperfect reference standard (HRA) can affect performance estimates between populations.[83] A higher disease prevalence may improve the sensitivity of anal cytology for detecting disease, but can reduce specificity, similar to cytology performance observed for triage of HPV-positive women (Figure 6). The performance of cytology in other high-risk populations, such as HIV-negative MSM, is highly variable. To inform clinical practice, more studies in this population are needed, and when possible, results should be reported within strata of HIV-positive and HIV-negative MSM.

TABLE 1.

Clinical Performance of Cytology, Human Papillomavirus Testing, and E6/E7 Messenger RNA in a Hypothetical Population of 10,000 Human Immunodeficiency Virus-Positive Men Who Have Sex With Men

Prevalence of AIN2/3 = 25%a	Test Sensitivity, %b	Test Specificity, %b	PPV, %	cNPV, %	Referral, %	No. Referred per Case
Primary screening
Cytology	81	53	36.5	10.7	55.5	2.2
HR-HPV	95	24	29.4	6.5	80.8	3.2
HPV16/18	41	69	31.0	22.2	33.5	1.3
E6/E7 mRNA	74	66	42.0	11.6	44.0	1.8
Screening and triage
Cytology and HR-HPV	77	64	55.5	17.2	27.1	1.4
Cytology or HPV16/18	89	37	31.8	9.3	69.8	2.8
Cytology and E6/E7 mRNA	58	85	68.4	21.6	17.1	0.9

Abbreviations: AIN2/3, anal intraepithelial neoplasia grade 2 or 3; cNPV, complement of the negative predictive value; HPV16/18, human papillomavirus types 16 and 18; HR-HPV, high-risk human papillomavirus; mRNA, messenger RNA; MSM, men who have sex with men; PPV, positive predictive value.

^aThe analysis assumes a population size of 10,000, and prevalence estimates assume a 25% prevalence of AIN2/3. Estimates of sensitivity and specificity are derived from studies that were restricted to HIV-positive MSM, with the exception of E6/E7 mRNA.

^bNet sensitivity and specificity were calculated for sequential testing with cytology followed by HPV16/18 or E6/E7 mRNA.

Figure 6. Risk-based management in cervical and anal cancer screening.

The absolute risk of precancer from 0 to 1 is shown on the y-axis. For cervical cancer screening (A), established risk thresholds are shown that are used to manage women with abnormal screening, triage, and histology results. Adaptations of this approach are shown for HIV-negative MSM (B) and HIV-positive MSM (C). Importantly, there are currently no established risk thresholds for anal cancer screening and management.

Given the high prevalence of HPV in HIV-positive MSM, the role of HPV testing in primary screening for anal cancer may be limited, as too many men would test positive and require additional workup. Indeed, our analysis of the performance of HPV testing indicated high sensitivity, but low specificity for HPV testing in studies overall, and especially in those restricted to HIV-positive MSM. Triage of cytology with HPV (or vice versa) improves specificity with a marginal decrease in sensitivity compared to cytology alone (Table 1). While HPV16/18 genotyping is substantially more specific, its sensitivity is insufficient for a standalone test and based on limited data, it does not seem to add much sensitivity at the cost of specificity in combination with cytology (a currently recommended strategy for cervical cancer screening) (Table 1).

Several other biomarkers are currently being evaluated for triage of HPV-positive women in cervical cancer screening that may better differentiate between transforming and benign HPV infections, including p16/Ki-67 dual stain, host and viral DNA methylation, and HPV E6/E7 mRNA.[28, 79] For anal cancer screening, these biomarkers may have applications as a primary screening tool, given the high background of HPV infections. Although some studies have evaluated p16/Ki-67 dual staining and DNA methylation in anal cancer screening, data are very limited, and we were not able to systematically evaluate the performance of methylation markers. The summary estimates for p16 or p16/Ki-67 dual stain are highly heterogeneous, related to differences in antibodies used, staining protocols, and interpretation. Our review indicates that E6/E7 mRNA shows promising performance, with greater specificity albeit lower sensitivity for detecting anal precancer compared with HPV DNA testing. However, it has been reported that the improved specificity of RNA compared to DNA testing can be largely attributed to the limited number of HPV genotypes (HPV16, 18, 31, 33, and 45) included in two of the reported HPV E6/E7 mRNA tests.[84] This suggests that extended genotyping or type restriction of HPV assays could improve the clinical performance of HPV DNA testing for anal cancer screening, but more research with current assays is needed.

Risk based management of cervical and anal precancers

It is important to consider how various tests may be implemented for anal cancer screening. Again, cervical cancer screening can serve as a model and contrast. With HPV testing, cytology and other available tests, we can now predict risk of cervical precancer and cancer with unprecedented precision. Recent screening and management guidelines efforts have adapted a risk-based approach to screening, management and treatment, that focuses on using risk rather than assay-specific results to guide clinical management.[85, 86] The approach follows the principle of equal management of equal risks, meaning that independent of how a certain risk level is arrived at, the management is the same. Screening, management, and treatment decisions are risk-based, and established clinical decision thresholds are used for the different management steps. The higher the risk of precancer, as a surrogate of cancer, the more likely treatment is going to be performed. The principle is shown in Figure 6A: The risk of cervical cancer in the general population is low, but increases at each step when additional tests are performed that provide risk stratification.[28] We show an adaption of that model to anal cancer screening in HIV-negative (Figure 6B) and HIV-positive MSM (Figure 6C). Importantly, anal cancer screening is not population-based, but focuses on certain high-risk groups. The disease prevalence in these populations is higher than the population prevalence of cervical precancer. Therefore, anal cancer screening starts at a higher risk level and because of that, it is more challenging to reassure patients of a low risk of precancer. HRA is the current standard for surveillance of HIV-positive MSM, but there are no established management thresholds for anal cancer screening.[39] Another important difference is that a higher risk of precancer does not necessarily lead to more likely treatment decisions, due to the morbidity associated with treatment of large anal precancers.[31] HPV DNA testing, which plays an increasingly important role in many cervical cancer screening programs world-wide, may be less important for anal cancer high risk groups such as HIV-positive MSM, because HPV status has limited influence on risk assessment. While a negative test may still provide good reassurance, this will only benefit a small group of patients, since the HPV prevalence is so high. However, novel biomarkers such as HPV mRNA, dual stain, HPV or host methylation could have important roles for screening and management in HIV-negative and HIV-positive MSM.[64, 75]

IV. Conclusions

In considering the available evidence, it is clear that more research is needed to address several important knowledge gaps that would inform the implementation of anal cancer screening. Most of the literature on anal cancer natural history and screening has been focused on HIV-positive MSM, given the high burden of disease in this population. Future studies should aim to better understand anal cancer natural history in other high-risk populations in order to determine if and how screening should be extended to these groups. The Study of the Prevention of Anal Cancer (SPANC), an ongoing prospective cohort study of HIV-positive and HIV-negative MSM based in Australia evaluating anal cytology, HPV genotyping, and other biomarkers for anal cancer screening, will help to better understand the influence of HIV infection on the clinical performance of these different strategies.[87] The lack of studies evaluating anal cancer screening in HIV-positive women and those with previous HPV-associated high-grade disease underscores the need for more research in these populations.[88]

Importantly, longitudinal studies evaluating the cumulative risk of anal precancer and cancer following testing with cytology, HPV, and/or other biomarkers are needed to establish optimal screening intervals. In cervical cancer screening, a negative cytology result does not have a strong, long-term negative predictive value because of its lower sensitivity and must be repeated often. On the other hand, HPV testing provides long-term reassurance against future risk of precancer if the test is negative, allowing for extended intervals between screens.[30] For a population at high-risk for anal cancer, such as HIV-positive MSM, a screening test should have high sensitivity in order to provide adequate reassurance that those testing negative will not develop anal precancer or cancer.[89]

A better understanding of the risk of progression of HPV infection to anal precancer to cancer and the influence of HIV on these estimates would help to inform screening intervals, ages to start and stop screening, as well as decisions about treatment versus expectant management.[31] Defining clinical action thresholds for anal cancer screening and management will require more research regarding the possible benefits (eg, the prevention of cancer) versus potential harms (eg, complications related to screening and treatment), which may vary substantially between individual patients. The Anal Cancer/HSIL Outcomes Research [ANCHOR] study is an ongoing trial designed to evaluate the treatment of anal precancers versus watchful waiting and will address some of the major unresolved questions related to the progression of HSILs and clinical outcomes.[31]

In summary, there are promising opportunities on the horizon that may improve screening and surveillance of populations at high risk of anal cancer. Established assays and approaches from cervical cancer screening give us a head start, but we need to carefully evaluate them in the context of anal cancer natural history. Ultimately, important data gaps need to be filled so that we can evaluate the benefits and harms of anal cancer screening approaches in various populations.