Objective
The purpose of this study was to evaluate the feasibility and effectiveness of Viral Testing Alone with Pap (Papanicolaou) Triage for Screening Cervical Cancer in Routine Practice (VASCAR) in a publicly funded university-affiliated hospital in Montreal, Canada.
Study Design
Women who are 30-65 years old are screened with the Hybrid Capture-2 assay. Women with negative results are retested at 3-year intervals; women with positive results are triaged with conventional cytologic methods. Women with Papanicolaou positive test results (≥atypical squamous cells of undetermined significance) are referred to colposcopy; women with Papanicolaou negative test results are retested with Hybrid Capture-2 assay and a Papanicolaou test in 1 year. Results were compared with a historic era (annual cytology with ≥atypical squamous cells of undetermined significance threshold for colposcopy referral) in the 3 years before VASCAR.
Results
VASCAR included 23,739 eligible women, among whom 1646 women (6.9%) tested positive for the human papillomavirus (HPV). Because of the need for subsequent sampling for cytologic testing, follow-up evaluation for cytologic triage was relatively poor; only 46% and 24% of HPV-positive women were Papanicolaou-triaged and underwent biopsy, respectively. Protocol violations occurred mainly in the early phases of implementation (12%). Detection of high-grade cervical intraepithelial neoplasia increased nearly 3-fold (rate ratio, 2.78; 95% confidence interval [CI], 2.1–3.7) during VASCAR, mostly because of a doubling in the rate of high-grade cervical intraepithelial neoplasia (34.0%; 95% CI, 21.2–48.8) compared with the historic cytology-only era (16.3%; 95% CI, 13.2–19.8). VASCAR reduced the median time to colposcopy from a positive screen from 11 months (95% CI, 10.48–11.50) to 3 months (95% CI, 2.64–3.80).
Conclusion
VASCAR is feasible; however, it requires cosampling for HPV and cytology and for continuous education of healthcare providers of the HPV-Papanicolaou triage protocol. Efficacy in disease detection and reduction in time to colposcopy referrals compared with the historic cytology era is encouraging but should be considered preliminary because of the small number of patients who were tested.
The overall performance of Papanicolaou cytologic test increasingly has been challenged, because several metaanalyses of the literature have shown that a Papanicolaou test fails to detect, on average, 50% of clinically significant precancers and cancers that are present at the time of testing. The interpretation of Papanicolaou smears is subjective, and cellular abnormalities may be missed because of the reader’s fatigue and distraction. Another drawback of the Papanicolaou test is the need for highly skilled cytotechnologists, the training of whom takes several years.
Because of the limitations of cytologic testing and the discovery that human papillomavirus (HPV) is the necessary cause of virtually all cervical cancers, viral tracing tests have been developed for screening for HPV infection and, by inference, its related complications such as cancer and precancer. Two dozen cross-sectional studies and, more recently, 9 multicountry randomized controlled trials have shown that HPV DNA screening tests, with and without concomitant cytologic testing, performed invariably better (20-60%) for the detection of cancer precursors than cytologic testing alone. American public health organizations recommend HPV and cytologic cotesting every 5 years in primary screening mode for women 30-65 years old with negative previous cotesting results. Furthermore, it has been proposed and documented that HPV-DNA testing alone in women who are ≥30 years old is likely to be a promising alternative method to cotesting for primary cervical cancer screening. A series of recently published randomized controlled trials in Finland and Canada clearly showed that testing women who are ≥30 years old first for high-risk HPV DNA and then triaging high-risk HPV positive women with cytologic testing provides better sensitivity and at least as high specificity as conventional cytologic testing, which considerably reduces the number of additional testing.
Although current guidelines for the detection of cervical cancer do not yet recommend screening by HPV-testing alone, primary screening with HPV followed by cytologic triage was judged to “be possible in selected or special clinical situations in a Western high resource setting.”
At the Jewish General Hospital (JGH), we have a major shortage in cytotechnologists, limited number of smears read by cytotechnologists by union contract (50 smears per 8-hour shift), and over-screening (eg, annual vs recommended 3-year intervals). This situation led us to switch cervical cancer screening from annual Papanicolaou cytologic testing to molecular HPV testing. As a pilot implementation project, the Viral Testing Alone with Papanicolaou Triage for Screening Cervical Cancer in Routine Practice (VASCAR) protocol was initiated in 2011. We report herein our first 3-years’ experience with VASCAR, the difficulties a new screening protocol can bring, and the results of the HPV testing with Papanicolaou triage strategy.
Materials and Methods
Protocol design
The primary goal of VASCAR was to evaluate the overall efficacy and safety of HPV testing as a primary screening modality, followed by cytologic triage of women who test HPV positive (hereinafter called HPV+/Papanicolaou triage). This approach was deployed in routine screening for cervical cancer and its precursors in women who were 30-65 years old over a 3-year period. VASCAR was implemented on March 1, 2011, to replace a policy of annual conventional cytologic screening with colposcopy referral based on the results. VASCAR was approved by the JGH’s Research and Ethics Committee in September 2011 and renewed in September 2013. According to the VASCAR protocol, women whose initial HPV test is negative are retested every 3 years (the screening interval that is considered to provide maximum acceptable safety). Women who test HPV-positive undergo Papanicolaou triage within 6 months. Those who are Papanicolaou-positive (≥atypical squamous cells of undetermined significance [ASCUS]) are referred for colposcopy. HPV-positive but Papanicolaou-negative women are tested for HPV and cytologic evidence (cotesting) in 12 months. If either test or both are positive, the patient is referred for colposcopy. All cervical samples are obtained from women who attend routine cervical cancer screening programs by attending physicians of the JGH and the affiliated medical clinics in the Montreal area. All the screening samples are sent to the JGH’s Pathology Laboratory.
To assess the performance of VASCAR’s new screening paradigm, we used an historic control group that included all cytologic and histologic results from all women who were 30-65 years old who were screened routinely with cytologic testing at the JGH 3 years before VASCAR. To restrict the assessment to routine screening, all women with cervical disease that had been documented either by cytologic or histologic testing or both during the historic era were excluded from the VASCAR policy period.
Samples
All cervical specimens for HPV testing were collected with the Digene sampling kit (Qiagen, Germantown, MD) that contains an exo/endocervical collection brush and a tube with specimen transport medium. All specimens were placed in a –70°C freezer within 5 days of reception and remained there until they were processed for HPV DNA. The samples were evaluated for the presence or absence of 13 high-risk HPV types with the use of the Hybrid Capture-2 (HC2) HPV-DNA assay (Qiagen) according to the manufacturer’s recommendations. All specimens were processed for testing within 1-5 days of reception. We routinely performed 264 tests per 8-hour shift per 1 Qiagen-certified medical technologist. Test results were issued within 5-10 working days together with written recommendations as to risk management.
Conventional cervical cytologic smears were read in a routine fashion by JGH cytotechnologists who used the Bethesda classification. The cytologic abnormality thresholds for colposcopy referral were (1) ASCUS or worse abnormality in the historic Papanicolaou-only era and (2) a Papanicolaou test that were classified ASCUS or worse after a positive HPV-test in the VASCAR-era. Colposcopy was performed in both the VASCAR and pre-VASCAR eras by 3 gynecologic oncologists, 4 gynecologists, and 1 gynecologic pathologist (A.F.) with a collective experience in colposcopy of 102 years.
At initial and follow-up colposcopic visits, biopsy specimens were taken from visible lesions of the cervix; if no lesions were visible, they were taken arbitrarily mostly at the 12 and 6 o’clock positions of the transformation zone. Endocervical curetting samples were collected in cases such as unsatisfactory colposcopy (squamocolumnar junction not visible), atypical glandular cells, and adenocarcinoma in situ. Papanicolaou and discrepancy between positive HPV/Papanicolaou yet negative exocervical/vaginal histologic results. The same protocol was used in both the historic and the VASCAR eras. Information as to the HPV/Papanicolaou results and colposcopic impressions were indicated on the pathology requisitions. We used the squamous intraepithelial lesion terminology with CIN grades 1, 2, or 3 as a modifier (ie, low-grade squamous intraepithelial lesion (LSIL; CIN1), high-grade squamous intraepithelial lesion (HSIL; CIN2), HSIL (CIN3) for the histologic diagnosis of precancerous lesions of the cervix). This terminology was used in the last edition of Blaustein’s Pathology of the Female Genital Tract . HSIL (CIN2 and 3) and cancer (squamous cell carcinoma or adenocarcinoma) were grouped together for the purpose of the calculation of lesion outcome rates for both the pre-VASCAR and VASCAR screening eras.
Statistical analyses
For the VASCAR-era, we analyzed only women who entered the study and had a sample for HPV DNA assay and results that were recorded between March 1, 2011, and Feb. 28, 2013. The present analysis refers only to the cross-sectional assessment of screening yield at enrolment (which includes procedures triggered by baseline screening results) and excludes follow-up data after triage procedures.
We used the Kaplan-Meier technique to estimate the cumulative distribution of times to colposcopy from an initial set of screening results that implied a colposcopy referral (see aforementioned cytologic thresholds). We calculated rates of colposcopy referral and rates of high-grade lesion detection (and corresponding 95% confidence intervals based on exact estimation) between the 2 screening eras after adjustment for incomplete follow-up data under the assumption that such rates were the same for those with complete and incomplete screening attendance and follow-up evaluation. All statistical analyses were performed with Stata software (version 12.1; STATACorp, College Station, TX) and WinPepi11.26 ( www.brixtonhealth.com ).
Results
Figure 1 shows a flow diagram of screening and diagnostic steps for all 28,939 women who were screened in the VASCAR era between March 1, 2011, and Feb. 28, 2013. After exclusions because of protocol-inappropriate age and previous disease (n = 746), 26,193 women who were 30-65 years old were eligible for analysis. Protocol violations are indicated in the Figure. A total of 2454 women (9.3%) had a Papanicolaou smear taken at their initial screening visit instead of an HPV test. Among the HPV-tested women as per protocol (n = 23,739), 22,093 women (93%) were HPV negative. Among these, 146 women (0.66%) had a smear taken (negative, 130 women; ASCUS, 12 women; LSIL, 4 women); 33 of them had a colposcopy (all negative), in violation of the protocol. Among 897 women (54.5%) with an HPV-positive test, 5 women underwent colposcopy before their triage Papanicolaou test result had been entered into the database (all negative), and 19 women underwent colposcopy despite having a negative triage Papanicolaou test (negative, 16 women; CIN1, 2 women; CIN2/3, 1 woman). In 11 women with a positive cytologic triage, Papanicolaou smears were taken rather than a referral for colposcopy. Overall, the VASCAR protocol was violated in 3414 of 28,939 screened women (11.7%). Among the 1646 women (6.9%) who were HPV positive, the cytologic triage was still pending for 897 women (54.5%) at the first round analysis closing date, either because a Papanicolaou triage visit had not yet been done or the Papanicolaou report was yet to be entered into the electronic medical record of the patient. An additional reason for a pending Papanicolaou triage result was loss to follow-up evaluation, because some women might have decided to be observed at a different clinic.
Of the 1646 HPV-positive women, 752 (45.7%) had their Papanicolaou smears taken and the results entered in the database. The Papanicolaou triage result was positive (≥ASCUS) for 210 of them (27.9%). Table 1 displays the breakdown by cytologic classification. A biopsy with or without endocervical curetting had been completed in 50 of the 210 HPV+/Papanicolaou triage women (23.8%). At the study closing date, 160 women were still due for colposcopy or had it performed elsewhere.
| Variable | VASCAR, n (%) | Historic era, n (%) |
|---|---|---|
| Women screened | 23,739 | 34,583 |
| Human papillomavirus tests performed | 23,739 | 0 |
| Negative | 22,093 (94.1) | NA |
| Positive | 1646 (6.9) | NA |
| Papanicolaou smears performed | 752 (45.7) | 34,583 (100) |
| Negative | 542 (72.1) | 33,463 (96.8) |
| Positive (≥atypia of squamous cells of undetermined significance) | 210 (27.9) | 1120 (3.2) |
| Atypia of squamous cells of undetermined significance | 121 | 839 |
| Low-grade squamous intraepithelial lesion | 67 | 202 |
| High-grade squamous intraepithelial lesion | 21 | 71 |
| Carcinoma | 1 | 8 |
| Repeat Papanicolaou smear after ≥ atypia of squamous cells of undetermined significance | 11 | 259 |
| Colposcopy with biopsy | 50 (25.0) | 503 (44.9) |
| Biopsy negative | 23 | 365 |
| Low-grade squamous intraepithelial lesion (CIN1) | 10 | 56 |
| High-grade squamous intraepithelial lesion (CIN2/3) | 16 | 76 |
| Carcinoma | 1 | 6 |
Table 1 also shows the screening yield for the historic control era, which included 34,583 age-restricted (30-65 years) and risk-eligible (no history of cervical disease) women. The Papanicolaou positivity rate (≥ASCUS) was 3.2% (n = 1120 women). Accordingly, a total of 503 women (44.9%) had colposcopy after an abnormal Papanicolaou test in the years preceding the VASCAR era.
There were 11,108 women from the historic era who were admitted into the analysis set for the VASCAR era because they were free of cervical abnormalities as per their negative Papanicolaou cytologic data. Therefore, nearly one-half of the women (46.8%) in the VASCAR era had been screened at least once in the previous 3-year period and had been considered lesion free. Yet, Table 2 shows that VASCAR brought a statistically significant 33% increase in the rate of colposcopy referrals consequent to the adoption of the HPV+/Papanicolaou triage approach. However, this relatively modest increase in colposcopy referrals led to a near 3-fold increase (rate ratio, 2.78; 95% confidence interval (CI), 2.1–3.7) in the rate of detection of histologically ascertained HSIL (CIN2-3) or cancer. Such an increase in lesion detection was due to a significant >2-fold increase (rate ratio, 2.09; 95% CI, 1.4–3.2) in the lesion detection rate per colposcopy performed.
