Cervical cancer risk for women undergoing concurrent testing for human papillomavirus and cervical cytology: a population-based study in routine clinical practice

Abstract

Background: Concurrent testing for human papillomavirus (HPV) and cervical cytology (co-testing) is an approved alternative to cytology alone in women aged 30 years and older. We aimed to assess the safety in routine clinical practice of 3-year screening intervals for women testing negative for HPV with normal cytology and to assess if co-testing can identify women at high risk of cervical cancer or cervical intraepithelial neoplasia grade 3 (CIN3) or worse over 5 years.

Methods: We assessed the 5-year cumulative incidence, starting in 2003-05, of cervical cancer and CIN3 or worse for 331,818 women aged 30 years and older who enrolled in co-testing at Kaiser Permanente Northern California (Berkeley, CA, USA) and had adequate enrolment co-test results. Follow-up continued until Dec 31, 2009. We defined cumulative incidence to include prevalence at enrolment and incidence after enrolment. Prevalence at enrolment was defined as the ratio of women diagnosed with each outcome on the biopsy visit immediately after their enrolment screening visit to the total enrolled women. At screening visits only HPV test and Pap smear samples were collected, and at biopsy visits colposcopically directed biopsies were taken. To estimate post-enrolment incidence, we used Weibull survival models.

Findings: In 315,061 women negative by HPV testing, the 5-year cumulative incidence of cancer was 3.8 per 100,000 women per year, slightly higher than for the 306,969 who were both negative by HPV and Pap testing (3.2 per 100,000), and half the cancer risk of the 319,177 who were negative by Pap testing (7.5 per 100,000). 313,465 (99.5%) women negative by HPV testing had either normal cytology or equivocal abnormalities. Abnormal cytology greatly increased cumulative incidence of CIN3 or worse over 5 years for the 16,757 positive by HPV testing (12.1%vs 5.9%; p<0.0001). By contrast, although statistically significant, abnormal cytology did not increase 5-year risk of CIN3 or worse for women negative by HPV testing to a substantial level (0.86%vs 0.16%; p=0.004). 12,208 (73%) of the women positive by HPV testing had no cytological abnormality, and these women had 258 (35%) of 747 CIN3 or adenocarcinoma in situ, [corrected] 25 (29%) of 87 cancers, and 17 (63%) of 27 adenocarcinomas.

Interpretation: For women aged 30 years and older in routine clinical practice who are negative by co-testing (both HPV and cytology), 3-year screening intervals were safe because a single negative test for HPV was sufficient to reassure against cervical cancer over 5 years. Incorporating HPV testing with cytology also resulted in earlier identification of women at high risk of cervical cancer, especially adenocarcinoma. Testing for HPV without adjunctive cytology might be sufficiently sensitive for primary screening for cervical cancer.

Funding: Intramural Research Program of the US National Cancer Institute/NIH/DHHS, and the American Cancer Society.

Figure 1. Value of HPV testing versus Pap smears (Panel A) and Value added by Pap smears to HPV testing (Panel B)

NOTE: Data are shown on cumulative incidence of CIN3+ by enrollment HPV test versus Pap smear. “Pap−“ means Pap-negative (NILM). “Pap+” means any abnormality (non-NILM). In Panel A, HPV test results are in blue, Pap smear test results are in black, positive test results are solid lines, and negative test results are dashed lines. Prevalent CIN3+ is plotted at time 0 (enrollment) and incident CIN3+ is plotted from that point. The HPV test more clearly separated high-risk women from low-risk women than the Pap smear because both (1) HPV+ women at enrollment had higher CIN3+ risk than the Pap+ women after 3 years (5.0% vs. 3.8%, p=0.046) and 5-years (7.6% vs. 4.7%, p=0.001), and (2) HPV− women at enrollment had lower CIN3+ risk than Pap− women at enrollment after 3-years (0.063% vs. 0.17%, p=0.001) and after 5-years (0.17% vs. 0.36%, p=0.02). In Panel B, HPV+ is in blue, HPV− is in black, Pap+ is solid and Pap− is dashed. Pap+ strongly modified risks for the HPV+ at three-years (10.0% vs. 3.1%, p<0.001) and five-years (12.1% vs. 5.9%, p<0.001), but not for the HPV− either at three-years (0.52% vs. 0.047%, p<0.001) or five-years (0.86% vs. 0.16%, p<0.001), although risks are statistically distinguishable. Comparing to Panel A, also being Pap-did not reduce CIN3+ risk from just being HPV− either at three-years (0.047% vs. 0.063%, p=0.6) or five-years (0.16% vs. 0.17%, p=0.8). See supplemental table 1 for 95% confidence intervals for all risk estimates. The number of women in each group are: HPV-positive: 16,757; HPV-negative: 315,061; Pap-positive: 12,641; Pap-negative: 319,177; HPV-positive/Pap-positive: 4,549; HPV-positive/Pap-negative: 12,208; HPV-negative/Pap-positive: 8,092; HPV-negative/Pap-negative: 306,969.

Figure 2. Five-year cumulative risks of CIN2+ (Panel A), CIN3+ (Panel B), and Cervical Cancer (Panel C) by enrollment HPV test and finely-categorized enrollment Pap smears

NOTE: Prevalent disease is plotted at time 0 (enrollment) and incident disease is plotted from that point. Although only 0.25% of women, women with ASC-H/HSIL/SCC Pap smears had by far the highest 3-year risks of CIN2+ (59%), CIN3+ (28%), and cancer (4.1%). Versus women with LSIL, women with HPV+/ASC-US had higher 3-year risk of CIN2+ (21% vs. 14%, p=0.01) and CIN3+ (6.4% vs. 3.2%, p=0.03). The majority (73%) of enrollment HPV+ women were HPV+/Pap−, and although they had by far the lowest disease risks at enrollment of any HPV+ group (CIN2+: 0.36%, CIN3+: 0.16%, Cancer: 0.04%), they accrued 5-year risks post-enrollment (CIN2+: 13%, CIN3+: 5.9%, Cancer: 0.5%) akin to women with ASC-H/HSIL/SCC (CIN2+: 14%, CIN3+: 6.1%, Cancer: 1.7%) or with HPV+/ASC-US (CIN2+: 13%, CIN3+: 4.5%, Cancer: 0%) (p=0.8 vs. ASC-H/HSIL/SCC, p=0.7 vs. HPV+/ASC-US). HPV+/Pap− women had higher 5-year cancer risk than HPV−/Pap+ women (0.54% vs. 0.16%, p=0.2). HPV−/ASC-US and HPV−/Pap− women had the smallest, and nearly indistinguishable, 5-year risks of CIN2+ (1.3% vs. 0.54%, p=0.07) and CIN3+ (0.54% vs. 0.16%, p=0.08). The 5-year cervical cancer risk in the enrollment HPV−/Pap− was 0.016% (3.2 per 100,000 women per year), only slightly smaller than the 0.019% (3.8/100,000/year) risk in the enrollment HPV− overall (p=0.8). By comparison, the 5-year cervical cancer risk in the Pap− was 0.037% (7.5/100,000/year) (p=0.3 vs. HPV− cancer risk). On the per 100,000 women per year scale, the 5-year cancer risks are: HPV−/Pap−: 3.2; HPV−/Pap+: 32; HPV+/Pap−: 108; HPV+/Pap+: 180. See supplemental table 1 for 95% confidence intervals for all risk estimates. The number of women in each group are: ASC-H/HSIL/SCC: 833; HPV+/ASC-US: 2,021; AGUS/NOS: 764; HPV-positive/Pap-negative: 12,208; LSIL: 2,527; HPV-negative/ASC-US: 6,496; HPV-negative/Pap-negative: 306,969.

Figure 3. Cumulative incidence of CIN3+ after second visit among 195,975 women with an enrollment HPV-negative/Pap-negative co-test, by second HPV test and second Pap test separately (Panel A) and jointly (Panel B)

NOTE: Data are shown on cumulative incidence of CIN3+ based on HPV test and Pap smear at the second screen, among women HPV−/Pap− at enrollment, for HPV test and Pap smear separately (Panel A), and jointly (Panel B). “Pap−“ means Pap-negative (NILM). “Pap+” means any abnormality (non-NILM). Prevalent CIN3+ is plotted at time 0 (second visit) and incident CIN3+ is plotted from that point. The y-axis goes to 12%, the maximum risk observed based on enrollment co-tests (Figure 1), to emphasize that CIN3+ risks for women HPV+ or Pap+ after enrollment HPV−/Pap− are markedly decreased from women HPV+ or Pap+ at enrollment. In Panel A, (1) women HPV+ at their return visit had higher 3-year CIN3+ risk than women Pap+ at their return visit (3.0% vs. 1.3%, p=0.2) and (2) women HPV− at their return visit had half the 3-year CIN3+ risk of women Pap− at their return visit (0.082% vs. 0.15%, p=0.3). Neither is statistically significant since there were only 102 CIN3+ at or after the second screening visit among women HPV−/Pap− at enrollment. However, comparing to enrollment co-test CIN3+ risks in Figure 1, the CIN3+ risks for women HPV+ or Pap+ have decreased markedly. In particular, the second HPV test (or Pap smear) does not separate 3-year CIN3+ risks as clearly as at enrollment (HPV+: 5.0% vs. 3.0%, p= 0.09; Pap+: 3.8% vs. 1.3%, p=0.04). Furthermore, comparing to Figure 1, the second HPV− test did not reduce 3-year CIN3+ risk accrued by the first HPV− result (0.082% vs. 0.063%, p=0.6); similarly, the second Pap− did not reduce the 3-year CIN3+ risk accrued by the first Pap− result (0.15% vs. 0.17%, p=0.8). In Panel B, being Pap+ modified risks more for the HPV+ (5.1% vs. 1.8%, p=0.4) than for the HPV− (0.19% vs. 0.079%, p=0.4), although neither is statistically significant. Comparing to figure 2, the 3-year risks were greater for HPV+/Pap+, HPV+/Pap−, HPV−/Pap+ at enrollment than at the second co-test (10.0% vs 5.1%, p=0.3; 3.1% vs. 1.8%, p=0.3; 0.52% vs. 0.19%, p=0.4), although none were statistically significant. However, comparing to Figure 2, risks were similar for HPV−/Pap− women at enrollment vs. women HPV−/Pap− again at their return visit (0.047% vs. 0.079%, p=0.5). Finally, being Pap− did not reduce 3-year CIN3+ risk from just being HPV− at second co-test after enrollment HPV−/Pap− (0.079% vs. 0.082%, p=0.9). See supplemental table 3 for 95% confidence intervals for all risk estimates. The number of women in each group are: HPV-positive: 5,381; HPV-negative: 190,594; Pap-positive: 8,453; Pap− negative: 187,522; HPV-positive/Pap-positive: 2,056; HPV-positive/Pap-negative: 3,325; HPV-negative/Pap-positive: 6,397; HPV-negative/Pap-negative: 184,197.