Importance: Cervical cancer can be prevented with early detection and treatment of precancerous lesions that are caused primarily by infection with high-risk strains of human papillomavirus (hrHPV). Current guidelines for screening in the United States focus on cytology screening with the Papanicolaou (Pap) test, with hrHPV cotesting as an option for women ages 30 to 65 years that allows for longer rescreening intervals. Evidence from large trials evaluating screening programs involving primary hrHPV testing (hrHPV alone as the initial test) and cotesting may inform new screening strategies. Evidence supporting cytology screening is well established, so this review evaluated screening with hrHPV testing alone (i.e., primary hrHPV testing) or as cotesting with cytology compared to cytology alone to address whether these forms of screening provide better protection from cervical cancer and allow for longer rescreening intervals. Rates of cervical cancer are very low among routinely screened women in the United States, but not all women are routinely screened, and there are significant racial/ethnic disparities in morbidity and mortality from cervical cancer.
Objective: To systematically review the benefits and harms of screening for cervical cancer using hrHPV testing as the screening strategy (with or without cytology).
Data Sources: MEDLINE, PubMed, PsychINFO, and Cochrane Collaboration Registry of Controlled Trials, and the Education Resources Information Center from January 2011 through February 15, 2017.
Study Selection: English-language trials of benefits or harms of screening for cervical cancer using hrHPV testing as the screening strategy (with or without cytology) in women age 21 years or older. Cohort studies were also considered for inclusion to evaluate harms and screening performance in large, representative primary care populations and in underscreened women.
Data Extraction and Synthesis: Two investigators independently reviewed abstracts and full-text articles, and then extracted data from fair- and good-quality trials and cohort studies. Results were qualitatively synthesized.
Main Outcomes and Measures: Cervical cancer mortality, invasive cervical cancer (ICC) incidence, early detection of disease (i.e., cervical intraepithelial neoplasia [CIN] 3+), rates of false-positive and false-negative screening, colposcopy and biopsy rates, quality of life and other harms.
Results: We included eight randomized, controlled trials (RCTs) (n=410,556), five cohort studies (n=402,609), and one individual participant data (IPD) meta-analysis (n=176,464). Trials were heterogeneous with regard to type of cytology (conventional vs. liquid-based cytology), type of hrHPV test (DNA PCR enzyme immunoassay vs. Hybrid Capture 2), screening interval (2 to 5 years), followup protocols for abnormal results, number of screening rounds (1 or 2), and consistency of screening protocols between rounds. Two fair-quality trials and one good-quality trial evaluated primary hrHPV screening (hrHPV testing alone) compared with cytology alone; two good- and two fair-quality trials compared hrHPV cotesting with cytology alone. The evidence was generally consistent across four trials with variable protocols and hrHPV test types demonstrating that primary hrHPV testing increased detection of CIN3+ in the initial round of screening (relative risk [RR] range, 1.61 [95% CI, 1.09 to 2.37] to 7.46 [95% CI, 1.02 to 54.66]). Only one trial of primary hrHPV testing, where all women with a positive hrHPV test were referred to colposcopy, had complete results from two rounds of screening (at Round 2 screening all women received cytology testing). In that study, CIN3+ detection in Round 1 was 3-fold higher in the hrHPV testing group. In the second screening round, CIN3+ detection was significantly lower among women in the intervention group (RR, 0.22 [95% CI, 0.08 to 0.58]), and cumulative detection over both screening rounds was 1.8-fold higher. Results of a large, single-arm, fair-quality cohort study of primary hrHPV testing at 3-year intervals were consistent with trial findings: CIN3+ detection in the second screening round was significantly lower compared to the first round (RR, 0.14 [95% CI, 0.06 to 0.32]). Among four trials of hrHPV cotesting, the first round CIN3+ detection was higher in the intervention group in two trials (though not significant) and equal in two trials. Cumulative CIN3+ detection over two rounds of screening ranged from 0.3 to 1.6 percent across studies. The relative risk for cumulative CIN3+ detection ranged from 0.91 to 1.13; none were significantly different from one. Long-term followup (13-year) in one trial showed similar results. ICC incidence was very rare. An IPD meta-analysis pooled data from five heterogeneous trials (including primary hrHPV testing and cotesting). A total of 107 cases of ICC among 176,464 women were identified in the trials, with a pooled RR of 0.60 (95% CI, 0.40 to 0.89) over one or two rounds of hrHPV screening compared to cytology alone and 5 to 12 years of followup data. Each of these trials included different patient populations and screening test protocols, adding uncertainty to interpretation of pooled findings. Evidence on subgroups was limited to age and a single cohort study focused on previously inadequately or unscreened women. Women younger than age 35 years had consistently higher rates of hrHPV positivity and of CIN3+. Outcomes of hrHPV primary testing or cotesting by age were not notably different from the results of the overall study populations. A small cohort study of cotesting among 1,832 Spanish women not screened in the previous 5 years found 9 cases of CIN3+; of these, 3 cases of CIN3+ were detected by hrHPV testing but not by cytology alone. The included trials did not report on potential adverse consequences of the screening tests, diagnostic procedures, or treatments and associated harms. Screening test positivity, false-positive rates (FPRs) for CIN2+ detection, and colposcopy referrals tended to be higher in the intervention groups of the trials, particularly at Round 1 screening. FPRs were higher in the intervention arm of one completed primary hrHPV trial and less discrepant in the other trial reporting test performance. In hrHPV cotesting trials, test positivity in the intervention group ranged from 7 to 22 percent of screened women, and was approximately 2- to 3-fold higher than in the control group arm. FPRs were also consistently higher in the intervention group at Round 1 for the three cotesting trials reporting on this outcome, ranging from 6 to 20 percent, and nearly 2- to 3-fold higher than control group rates. Two cotesting trials reported test performance data from Round 2 screening; the FPR was similar between arms in one trial, but 2 times higher in the intervention arm in another. Four hrHPV primary screening trials and two cotesting trials reported referrals to colposcopy. Rates of referral to colposcopy in the control groups ranged from 1 to 3 percent. Two primary hrHPV testing trials had more referrals among women in the intervention arms versus control group at Round 1 of screening (8% and 6% vs. 3%). Two other trials of primary hrHPV testing had similar rates of referral in both trial arms. Two hrHPV cotesting trials reported more referrals to colposcopy in the intervention group compared to the control group (11% vs. 3% and 7% vs. 5%). Round 2 colposcopy referral rates, reported only in one cotesting trial, were similar between treatment groups (IG, 3% vs. CG, 2%). Biopsy rates were reported in the IPD meta-analysis; the pooled estimate had very high heterogeneity, largely explained by the 2-fold difference in biopsy rates between intervention and control arms in the two trials that referred all hrHPV+ women to colposcopy. Biopsy rates were similar between arms for the other trials. Data were too sparse to draw conclusions regarding the risk of missed cases of cervical cancer (false negatives) for different screening strategies, given very few cases of ICC detected. Limited evidence on psychological harms from one cross-sectional study (n=428) and a substudy of one cotesting trial (n=2,508) suggested that women receiving hrHPV positive test results experienced increased anxiety and distress, and reduced satisfaction with sexual partnerships.
Conclusions and Relevance: Eight large randomized trials, four of primary hrHPV testing and four of hrHPV cotesting, contributed to the evidence comparing use of hrHPV testing as part of cervical cancer screening with cytology alone for detection of CIN3+. All trials were conducted in the context of organized screening programs, with heterogeneous screening strategies and followup protocols. Interpretation of trial findings was limited by the fact that after one round of screening, only one trial conducted further screening applying the originally assigned strategies in the control and intervention arms. In all other trials, both arms received the same test at Round 2 (either cytology alone or hrHPV cotesting). Primary hrHPV testing increased detection of CIN3+ in the initial round of screening by as much as 2- to 3-fold. Only the trial of primary hrHPV testing, where all women with a positive hrHPV test were referred to colposcopy, had results from two rounds of screening. In that study, CIN3+ detection in Round 1 was 3-fold higher in the primary hrHPV testing arm, and cumulative detection was 1.8-fold higher after the second round of screening. Evidence was mixed in cotesting trials. No trial showed a significant increase in CIN 3+ detection in Round 1 for cotesting. In two of four trials, CIN3+ detection was lower in Round 2 in the hrHPV cotesting arm and higher in the cytology alone arm. Cumulative CIN3+ detection was similar between intervention and control study arms in all trials. Because no trial sustained the intervention and control group protocols beyond two screening rounds, evidence comparing the long-term outcomes of primary hrHPV testing or cotesting with cytology was lacking. Data to compare potential harms of different screening strategies were similarly limited, and none of the included trials or observational studies reported on harms of the screening test or treatments. False-positive rates and referrals to colposcopy were in some trials 2- to 3-fold higher with hrHPV-based screening strategies relative to cytology alone in the first screening round, and evidence was lacking to determine whether these differences might persist over multiple screening rounds. Risks of missed ICC were very low regardless of the screening strategy used. An IPD meta-analysis suggested a lower rate of ICC with hrHPV screening strategies, but this analysis pooled data from trials with distinctly different screening strategies and hrHPV test types, adding uncertainty to interpretation of the findings. In most trials and in a large U.S.-based observational study, women younger than age 30 to 35 years had higher rates of hrHPV positivity and CIN3+, accompanied by higher rates of colposcopy. No completed studies compared different screening intervals. All of the RCTs on hrHPV screening were conducted in countries with organized screening programs, which are not available to most women in the United States. Rigorous comparative research is needed in U.S. screening settings to examine longer screening intervals, long-term outcomes, and to identify effective strategies for outreach and screening of poorly screened and unscreened women. The higher sensitivity of hrHPV testing in a single round may have potential to improve outcomes in this high-risk population.