Main findings

In 2017, 108,585 Danish women aged 69 years or older had cervical samples tested for HPV, and 4.1% were HPV-positive. By far the majority of the HPV-positive women had one or more follow-up tests taken, and about two-thirds had at least one histology result. Conization was performed in 1% of screened women, corresponding to 24% of HPV-positive women. Cervical cancer was detected in 37 women, and CIN3 in 300 women, which was a CIN3+ detection rate of 0.31% in screened women and of 7.5% in HPV-positive women. More than three women underwent conization per detected CIN3+ case. Proportions of screened women testing HPV-positive, undergoing histology or conization varied little below age 79 but were lower in women above this age. CIN3+ detection rates were similar across age.

While the CIN3+ detection rates varied little across the five Danish regions, there was a considerable variation in proportions of women undergoing histology and/or conization. As a consequence, there was approximately a two-fold variation in the numbers of biopsies/conizations per severe histology across regions.

Other studies

A Swedish study including 1,051 women aged 60–89 years found that 4.1% were HPV- positive [12]. For HPV-positive women, testing was repeated after 3 months, and if negative again at 12 months. Overall, 30 (3%) women had two positive HPV-tests within one year, and 30% of those positive on first test were negative on both repeat-tests. In another Swedish study, where women initially HPV-negative were retested after a mean interval of 3.2 years, 2.6% were HPV-positive, of whom 52% were positive again on repeat-test after an average of 2.5 months [13]. In both studies, women with two positive HPV-tests had a high prevalence of CIN on histology, although mainly CIN1, but few had detectable cytological changes. Comparable results were obtained in another study where HPV-test was performed on self-collected samples [14]. When retesting HPV-positive or HPV-negative women, these studies thus demonstrated that test results in elderly women vary over even short time spans, although it is unknown to which extent this is due to cleared infections, reinfection, reactivation of latent HPV-infection or difficulty in obtaining representative samples from elderly women. In these Swedish studies, fewer severe histology lesions were found than in ours. This may be related to sample size, selection of studied women, to screening histories of studied women, or to the historically lower cervical cancer incidence in Sweden than in Denmark [15].

A Danish GRADE-based evidence review on management of elderly, HPV-positive women concluded that there is insufficient evidence on how to ensure reduction in cervical cancer risk while minimizing overtreatment [16]. Surveillance over time is complicated by the difficulty in obtaining representative samples from the cervix due to the age-related changes, and the Danish guidelines therefore recommend considering conization in women where the transformation zone cannot be visualized and/or biopsies have been unrepresentative.

The Horizon study compared screening by cytology to different HPV assays in Denmark [17]. For women aged 30–65 years screened with Cobas HPV test (which was also used in most regions for the one-time HPV screening of older women), the study found 129 biopsies and 37 CIN3+ cases among 465 women positive on Cobas, i.e. 8% of HPV positive women had CIN3+, and 3.5 biopsies were performed per CIN3+ case diagnosed.

A woman’s risk of cervical cancer in old age is related to her screening history [18,19], and at the population level, the cervical cancer incidence in a specific age-group depends on the screening history of the birth cohorts making up that age-group at that point in time [8]. The Danish data showed a steady decrease in HPV-prevalence after the age of 35 years [9]. 82% of cervical cancer cases diagnosed in women aged above 60 years occurred in never or insufficiently screened women, and the average age of non- or insufficiently screened cancer patients was 76 years, the same age as cancer incidence peaks in old age [9,20]. However, the Danish Pathology Register is only complete after 1998, and the screening history of older women is therefore underreported in the register. Consequently, studies show an increase in proportion screened over time, reflecting not only increased screening but also increased completeness of the pathology register [21].

A cohort study of more than 500,000 Swedish women found that screening between age 61 and 65 years was beneficial for women with an insufficient screening history, while there was no significant reduction in cervical cancer risk up to age 80 years in women adequately screened with normal test results at age 51–60 [18].

Strengths and limitations

The availability of the national pathology register and unique identifiers ensured complete follow-up. However, data on deaths and emigrations were not available, and we were therefore unable to censor these women in the analysis. At the population level, we know that the birth cohorts targeted by the one-time HPV-screening were insufficiently screened, but individual data on the screening history, previous histology findings or therapeutic procedures were not available, as the pathology register data are incomplete before 1998.

For some women, conization was performed after a biopsy had revealed a severe lesion, but for other women the procedure was diagnostic, when a representative biopsy could not be obtained. For some women, a conization may have been performed even in case of a non-severe/no lesion to avoid lengthy follow-up with repeated cervical sampling. Some women may still be in follow-up with HPV test and/or biopsy and may still be conizised after the period covered in this study. Conization rates could thus be slightly higher than reported here, as a decision on end of follow-up is not coded.

Women whose HPV screening test was negative were not retested, so intermittent HPV-positivity, as in the Swedish studies referenced above, would not have been detected, and the prevalence of histological lesions in women testing HPV negative was unknown. This limits the ability of the study to fully estimate the cancer preventive effect of a single HPV test in these age groups.

Part of the variation between regions could theoretically stem from differences in age distribution. The aggregate RKKP data did not allow for age adjustment, but there is little difference across regions in the age distribution of women in the studied age-range [22]. We consider it unlikely that age specific participation in screening should vary across regions, and consequently we consider it unlikely that the observed regional differences in histology and conization rate derived from difference in age distributions.

Clinical implications

The European guidelines for cervical cancer screening recommend that women exit cervical cancer screening at age 60–65 years provided they have had a recent negative test [23]. The US preventive services task force recommends against screening of women older than 65 years if they have had adequate prior screening and are not at high risk for cervical cancer [24].

A possible extension of existing screening programs beyond age 65 years has been subject of both scientific and political considerations in many countries. In Denmark, the decision was to target in incompletely screened birth cohorts with the one-time HPV screening described here. Our results suggest that, for these cohorts, HPV screening can potentially prevent some of the morbidity and mortality from cervical cancer, as severe lesions (CIN3+) were detected in 7.5% of HPV-positive women, including 37 women with cervical cancer. However, not all CIN3+ cases progress to cancer, and some of the 37 detected cervical cancer cases might not have become symptomatic in the women’s remaining lifetime. Whether this screening intervention translates into fewer cervical cancer cases and deaths will be visible only after a longer follow-up.

The results may not, however, be generalizable to birth cohorts after 1948 who have had a systematic screening offer throughout their adult life. It is also important to note that, although women older than 65 years constitute an increasing proportion of cervical cancer cases, the incidence in older women has declined over the past decades [15,25], reflecting the progressively lower risk in each birth cohort [8]. The age-specific mortality has also declined over time [15]. With primary HPV screening being implemented in many countries, the risk in elderly women may be even lower in the future.

An alternative to a general extension of screening age might be to consider more personalized approaches, factoring in each woman´s age, screening history, comorbidities, general health and expected remaining lifespan [26] but more evidence is needed to estimate effects of both general and personalized strategies. In Denmark, complete cervical cancer screening results are available from the Danish Pathology Register for birth cohorts aging in the future, thus enabling studies comparing cancer risk after screening age for women who did or did not meet exit criteria, including women whose exit test was an HPV test.

In all screening initiatives, it is important to consider the potential harms in addition to the benefits. Although conization is generally considered a minor procedure, the risk of bleeding and complications like perforations may be elevated in older women, and age-related anatomical changes may render the procedure more difficult. In addition, the therapeutic effect of conization is not well documented for women over 60 years [16]. On the other hand, obtaining representative biopsies in older women is not always possible and therefore conization may be necessary for providing a diagnosis and can be preferable to avoid lengthy follow-up with repeated cervical sampling [16]. As shown in this study, the choice of management strategy has an important impact on the number of biopsies/conizations per findings of CIN2+ or CIN3+, and conceivably also on the number of women recalled for repeated tests in the absence of conization. Although differences in study design do not allow for direct comparison with the Horizon study, the results do suggest that the yield of CIN3+ compared to the number having histology results is less favorable for older women, with 8.3 histologies per CIN3+ in our study vs. 3.5 in Horizon.