HPV triage of patients with ASCUS cervical pap smears

The pap smear 

In 1989 the Bethesda system was introduced to replace the previous Papanicolaou system. It established general diagnostic categories, introduced new and uniform descriptive terminology and diagnoses, and established specimen adequacy standards. However, it was poorly reproducible and had a high proportion of atypical squamous cells of undetermined significance (ASCUS)—as high as 10% in some areas. There was also significant confusion regarding treatment and follow-up of ASCUS specimens.

The Bethesda system was revised in 2001. The categories were changed to “atypical squamous cells of undetermined significance” (ASCUS) and “atypical squamous cells—favor high-grade lesion” (ASC-H). An ASCUS has cytologic changes suggestive of a lesion but lacks strict criteria for definitive interpretation. An ASC-H has changes suggestive of HSILand is highly predictive for HSIL lesions, but lacks criteria for definitive interpretation. This subcategory is expected to account for only 5% to 10% of all ASC Paps.

Despite efforts to improve classification and interpretation, there continue to be significant problems. Conventional Paps have a poor sensitivity for disease. A recent meta-analysis demonstrated a sensitivity for ≥LSIL lesions of only 58% with a specificity of 69% (3).

Liquid-based cytology has recently been introduced to improve the sensitivity and specificity of the Pap. This involves suspending cervical scrapings in a fluid medium for centrifugation and analysis. Sensitivity of a single thin-layer Pap of ASCUS for CIN3is 61% with a specificity of 82%, while 19% would be referred for colposcopy (4). How predictive is Pap screening? That depends on outcome settings (ASCUS, LSIL, HSIL), on repeat Paps, and the intervals between screenings.

The main limitation in screening is that it isn't done. A recent review found that 52% of cervical cancer patients had suboptimal screening (5), including 28% who had never had a Pap and 33% whose last Pap was more than 5 years prior to diagnosis. Only 34% of patients developed cervical cancer within three years of their last Pap. Kinney et al. found that in 60% of cervical cancer cases in a prepaid health care plan, patients had recent access to health care, but failed to have cervical cancer screening (2).

The Pap is only a screening tool. It does not yield a diagnosis; it only serves to identify women who require further testing, while avoiding unnecessary testing in low-risk women.

Colposcopy—the gold standard? 

Until recently the options available for women with ASCUS smears were either immediate colposcopy or repeat cytology. Immediate colposcopy, although expensive, has been considered the gold standard. However, this has been questioned lately. Even with experienced colposcopists, the false negative rate is as high as 40% to 50% (pending publication: ASCUS and Low-Grade Triage Study). A meta-analysis comparing colposcopic impression with histology findings demonstrated 89% accuracy, with 61% exact correlation to histology (6). The sensitivity of colposcopy has been noted as 87% to 99%, with a specificity of 26% to 87%.

Etiology—human papillomavirus 

There is a strong association between cervical cancer and specific high-risk human papillomavirus (HPV) types. More than 100 types of HPV have been identified and as many as 30 have been found to cause genital mucosal infection. These have been classified as high-risk types (16, 18, 45, 56) (7), and intermediate-risk types (31, 33, 35, 39, 51, 52, 55, 58, 59, 66, 68). (Low-risk types [6, 11, 26, 42, 43, 44, 54, 70, 73] are more associated with genital warts, LSIL, and recurrent respiratory papillomatosis.) There is substantial evidence that the persistence of high-risk HPV is a major risk factor for high-grade CIN (8, 9).

Epidemiological studies have since been supported by laboratory studies. The integration of HPV DNA into the cellular DNA disrupts the E2 regulatory gene of HPV, leading to increased expression of E6 and E7. These genes have the ability to inhibit p53 and Retinoblastoma gene function, inducing cellular immortalization (7). We now have a unique opportunity to screen for the specific etiologic factor that predisposes to cervical dysplasia and cancer.

HPV infection is a highly prevalent sexually transmitted disease, with as many as 50% of sexually active women having been infected with one or more HPV types in the past (10). Some 15% to 22% of women display evidence of current HPV infection and 50% to 75% of these (or about 10% of the population) are infected with high-risk HPV types (11, 12).Another study found a prevalence of all HPV types in 20 million women in the U.S. with an annual incidence rate of 5.5 million (13). In one study of a group of patients with no prior history of HSIL lesions, there was a prevalence of high-risk HPV types of 27% (14). In a study by Ho et al. looking at all HPV types, there was a 36-month incidence of 43% with a median duration of infection of 8 months (15). Persistence of HPV infection after one year was found to be 30%; after two years, 9%.

In the persistent group, there was a 50% chance of developing a high-grade cervical lesion. Once developing a mild cervical dysplasia from HPV infection, there is a 1% to 2% chance of progression to severe dysplasia (16), and once moderate dysplasia is present, there is a 16% chance of progression to severe dysplasia or cancer within two years (16, 17).

Despite the high incidence of infection with HPV, most HPV infections are transient and disappear within several months to two years (9, 18). There is a 56% chance of regression from mild dysplasia to normal and even a 44% to 53% chance of regression from moderate dysplasia to normal (16, 17). How can we distinguish women with HPV infection who are likely to progress to severe dysplasia and cancer, from those with a transient sexually-transmitted infection?

HPV testing 

For many years, expensive and cumbersome PCR techniqueswere the only way to identify HPV infection, but recent advances have ushered in affordable and easy HPV tests, such as the Hybrid Capture II (Digene Corp., Gaithersburg, MD). Monolayer liquid-based cytology has been introduced, reducing false-positive rates and allowing for HPV testing on remaining material, thus avoiding repeat clinical visits to evaluate borderline results.

HPV testing as triage for ASCUS paps 

Prior to HPV testing, the only options for management of women with ASCUS Paps were immediate colposcopy or repeat cytology with colposcopy at a determined threshold (such as repeat if ≥ASCUS, ≥LSIL, or ≥HSIL). Immediate colposcopy, while highly sensitive for detection of high grade dysplasia, was considered too costly. Ten to 15% of high grade dysplasias is found in the ASCUS population, while high-risk HPV is found in 40-60% of ASCUS patients (Fig 2).

View Large Image Download to PowerPoint

Figure 2. High Risk HPV by Hybrid Capture

HPV testing has the potential for halving the number of colposcopies performed for ASCUS Paps. However, as many as 83% of LSIL Paps and 95% of HSIL Paps are high-risk HPV-positive, which diminishes the utility of HPV testing in these groups as a triage strategy (19). ASC-H Paps have been found to be similar to LSIL Paps, with a large proportion of high-risk HPV positivity (unpublished data from ASCUS and Low-Grade Triage Study).

Hybrid capture HPV testing in ASCUS 

Several studies have evaluated hybrid capture HPV testing for ASCUS Paps (Table 1). The largest has been the prospective ASCUS and Low-Grade Triage Study (ALTS) (19), which assigned a group of women with either ASCUS or low-grade Paps into one of three arms: immediate colposcopy, repeat Pap every six months with an HSIL Pap as threshold for colposcopy, or reflex testing for high-risk HPV, with HPV-positive women going to colposcopy.

Table 1. Performance of HPV Hybrid Capture Testing in ASCUS Paps

	Study	#Pts	Pap	Age	Referral Rate	% Sens HSIL	% Spec HSIL	NPV HSIL	PPV HSIL
	Solomon (ALTS)21	1149	ASCUS	18+(27)	56	96	49	98.9	19.6
	Manos22	973	ASCUS	14–92(40)	39	89	64	98.8	15
	Morin27	360	ASCUS	18–50(?)	29	89.5	74	99.2	16
	Lonky28	278	ASCUS	––	46	82	59	96	21
	Fait29	226	ASCUS (× 2)	17–48(28)	24	86	97	93.4	90.5
	Shlay30	195	ASCUS	15–76(34)	31	93	74	99.3	23
	Ferris31,32	143	ASCUS	18+(27)	––	89	40	98.2	9.1
	Bergeron33	111	ASCUS	15–75(35)	43	83	62	97	21
	Lytwyn34	87	ASCUS/LSIL	16–50(30)	––	––	51	98	15
	Lin35	74	ASCUS	>50(62)	53	100	65	100	67
	Clavel36	23	ASCUS	15–72(37)	48	100	57	100	9
	Calculated Weighted Mean				41%	89%	61%	98.3%	23%

Several important issues are worth noting from this trial. First, thin-layer liquid-based cytology was used for cytological screening (Fig 3). This allowed the performance of reflex HPV tests on residual sample material without a return clinical visit. Second, although the trial was begun prior to release of the Bethesda 2001 classification, the results were confirmed even when the Paps were reclassified into the new system.

The LSIL arm of the trial was closed early because an interim analysis showed that 83% of women with an LSIL Pap would be referred to colposcopy due to a positive high-risk HPV test. Of ASCUS Paps, 15% of patients were found to have CIN2 or greater at time of immediate colposcopy, suggesting a significant rate of high-grade dysplasia in what was thought to be a benign group of patients.

The trial concluded that reflex HPV testing as a triage strategy is effective, with a 96% sensitivity for CIN3, while referring only 56% of ASCUS patients to colposcopy. Of those, 28% were CIN2 or greater on histological examination, thus improving the specificity of colposcopy in this setting. If colposcopy were only to be performed for HSIL or greater Pap (i.e. normalizing ASCUS and LSIL Paps) then there would only be a 44% sensitivity for CIN3+ with a 7% referral rate. Such results are not considered sensitive enough for the US population.

The previous standard, repeat cytology followed by colposcopy for ASCUS or greater, was found inferior to HPV testing. It provided an 85% sensitivity for CIN3+ with a 59% referral rate; thus it was less sensitive and had a higher referral rate than HPV testing. Considering the costs of clinical visits for repeat cytology versus the cheaper reflex-HPV testing, makes the old system even less appealing. (Long term follow-up on these patients is still pending.)

The second largest trial to date is one by Manos et al. (20). Of a cohort of 46,000 women, 995 were identified as having ASCUS on a conventional Pap, with an overall rate of 3.5% ASCUS, 0.9% LSIL, and 0.3% HSIL. This trial population had a mean age of 37 vs 27 for the ALTS trial. It excluded women with a history of CIN in the previous six months. All patients went to colposcopy where the colposcopist was blinded to the Pap result. Hybrid Capture II testing was used, as in the ALTS trial. Overall, 7% of ASCUS patients were found to have high-grade dysplasia (CIN2+) at histology and 39.5% of patients were high-risk HPV-positive. The sensitivity of HPV testing for high-grade dysplasia was 89%, specificity 64%, positive predictive value 15% and negative predictive value 98.8%. The referral rate to colposcopy was only 39%. The researchers concluded that repeat cytology was inferior, due to its lower sensitivity for high-grade dysplasia and equal or higher referral rates to colposcopy.

HPV testing as an adjunct to cytology in women over age 30 

Recently the FDA has approved the use of screening for high-risk HPV subtypes as an adjunct to routine Pap cytology. However, several concerns exist about the increased costs of this strategy as well as the risk of overtreatment. An estimated 10% to 20% of all women can be expected to have a transient, clinically insignificant HPV infection. However in women 30 years or older, this is reduced to 5% to15% (4, 14, 21). High-risk HPV testing in this population (Table 2) has a sensitivity for high-grade dysplasia of greater than 90% compared to Pap sensitivity of only 60% (22). This comes at a cost of reduced specificity, from 96% to 93%. When used in combination, Pap cytology and HPV testing have a sensitivity and specificity of 91%. A recent workshop including participants from the National Institutes of Health-National Cancer Institute, ASCCP, and the American Cancer Society, concluded that the use of adjunctive HPV testing for women over 30 was an acceptable strategy.

Table 2. Weighted performance of Pap and High-Risk HPV in Women >30 (adapted from Wright, TC et al24)

	n	% CIN2+	Sensitivity (%)			Specificity (%)
			Pap	HPV	Combination	Pap	HPV	Combination
	36142	1.00	63.9	91.4	95.9	96.0	93.1	90.9

Conclusion 

In summary, in a patient population younger than 30 years of age, HPV is more prevalent but the infections are more likely to be transient, resulting in more positive HPV tests with no dysplasia (false positives). Patients older than 30 are more likely to be truly positive, with HPV infection that correlates with dysplasia. Use of HPV testing as a triage strategy for ASCUS Paps is effective and potentially economical (Table 3), especially with thin-layer liquid cytology. However, if HPV testing is used as primary screening, screening intervals should be lengthened for cost-effectiveness.

Based on available data, the American Society for Colposcopy and Cervical Pathology has developed the following recommendations:

•ASCUS

○Repeat cytology, immediate colposcopy, or HPV testing are all acceptable strategies.

○Hybrid capture HPV testing after ASCUS Paps has sensitivity for CIN2+ of 83-100% (higher than single repeat cytology) and negative predictive values of 98% or greater.

○Reflex HPV testing of ASCUS Paps spares 40-60% of women colposcopy and is the preferred management strategy when liquid based cytology is used.

○ASCUS/HPV negative patients may be followed up with repeat cytology at 12 months.

○ASCUS/HPV positive patients with nCIN on colposcopy should have repeat cytology at 6 and 12 months.

○If repeat cytology strategy is used, it should be performed at 4-6 month intervals, until two consecutive negative Paps are obtained, with colposcopy for any repeat ASCUS or greater cytology.

○If immediate colposcopy is used, patients with no CIN on colposcopy should have repeat cytology at 12 months.

○Cervical excision procedures should not be used to treat women with ASC in the absence of biopsy-confirmed CIN.

•ASC-H

○Referral to immediate colposcopy for any ASC-H regardless of whether conventional or liquid-based cytology is used.

•ASCUS in postmenopausal women

○Treatment of atrophic changes with intravaginal estrogen with repeat cytology one week after treatment is completed.

○If subsequent cytology is ASCUS or greater, colposcopy is indicated.

•ASCUS in immunosuppressed women

○Referral to colposcopy for cytology of ASCUS or greater, regardless of HIV viral load or anti-retroviral therapy.

•ASCUS in pregancy

○Manage same as nonpregnant women.

Table 3. Cost Effectiveness of Management Strategies for ASCUS Paps- Biennial Screening with Liquid-Based Cytology

	Strategy	Avg Lifetime Costs ($)	Absolute Reduction in Cancer Incidence (%)	$ Per Year Life Saved (CE Ratio)
	Ignore ASCUS	1423	84.03	13,700
	Reflex HPV Testing	1712	90.41	44,400
	Repeat Cytology	1820	90.15	Dominant*
	Immediate Colpo	1867	90.54	905,300

CE: Cost Effectiveness

* Dominant- Strategy is more costly and less effective than other strategies.

View Large Image Download to PowerPoint

Figure 3.