The relation between cervical cancer and the human papillomavirus (HPV) has been established beyond doubt. The prevalence of HPV in cervical cancer of nearly 100%1 and the high sensitivity of HPV detection for high-grade cervical lesions,2 as well as the resulting high negative predictive value, would seem to have paved the way for the introduction of HPV detection in primary cervical cancer screening. The introduction of HPV detection in cervical cancer screening is halted, however, primarily by the high rate of false positivity in relation to morbidity, since the majority of women infected with HPV will not develop lesions. A single positive HPV test does not give a good specificity to detect high-grade lesions. The prevalence of HPV in cytologically normal smears is age-dependent, as has been shown by a number of studies.3, 4, 5, 6, 7, 8, 9, 10 Because of the frequency of asymptomatic HPV infections, as well as transient low-grade cervical lesions in young women, HPV testing as part of cervical cancer screening would be unwise before the age of 30 years.
In Belgium, as in many other countries, there is no nationwide population-based cervical cancer screening program. Women are free to have cervical smears taken, either by their general practitioner or their gynecologist, at an interval of their choice. This leads to overconsumption.11 HPV detection may be especially valuable to counteract overscreening of elderly women by performing a single HPV test in conjunction with cytology at a certain age (for instance, at 50 or 55 years). We believe that women who are HPV-negative and have a cytologically normal smear might be encouraged to refrain from further screening.
However, it was recently shown that after an initial decline in HPV prevalence, a higher prevalence after the age of 55 may be encountered.12, 13 Since a higher HPV prevalence in older women would negatively affect the applicability of HPV testing to reduce overscreening in elderly women, we have studied the HPV prevalence in a group of elderly women, screened either by their gynecologist or their general practitioner.
MATERIAL AND METHODS
Study group
Between January 2001 and August 2002, cytology material from consecutive women of 50 years of age or older attending the Department of Gynecology at the University Hospital Antwerp for regular cervical cancer screening (n = 1,504) and consecutive women of 50 years of age or older attending their general practitioner (n = 432) was included in this study. The study was performed anonymously; the only data available were age at sampling and result of cytology. The study protocol was approved by the medical ethical board of the Antwerp University (A01-005).
Cervical cytology
Liquid-based cytology (autocyte) was used for cytologic evaluation. The specimens were cytomorphologically classified according to the modified Bethesda system14 into normal, atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial neoplasia (LSIL), high-grade squamous intraepithelial neoplasia (HSIL) and cancer. Cytologic analysis of all slides was performed at the Department of Pathology, University Hospital Antwerp.
Sample preparation and HPV detection
The residual material after preparation of thin-layer cytology slides was used for HPV detection. The suspension was centrifuged for 5′ at 1,207g. Cells were resuspended in 0.5 ml TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) and frozen at −80°C. After thawing, 100 μl of the suspension was taken, boiled for 10′ and centrifuged (5′, 12,557g). Isolation of DNA was confirmed by β-globin PCR using primers PC03/04.15 In β-globin-negative cases, the remaining material was incubated overnight at 37°C in the presence of 0.2 μg/ml proteinase K. After boiling (10′) and centrifugation, the β-globin PCR was repeated.
Material positive for the β-globin PCR was subjected to the GP5+/6+ HPV PCR.16 Detection of PCR products was performed in an enzyme immunoassay (EIA) format as described by Jacobs et al.17 After detection of HPV with a high-risk HPV probe cocktail, typing analysis was performed for high-risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68. The cutoff value for HPV positivity was calculated as the mean plus 3 times the standard deviation (SD) of all samples in the plate. Outliers were excluded and the mean + 3 SD was recalculated. This process was repeated until no further outliers were found. The final mean + 3 SD was taken as the cutoff value.18
Statistical analysis
The chi-square test was used to compare the HPV prevalence according to cytologic result. Linear trend analysis was performed to detect differences in HPV prevalence with age. The Fisher exact test was used to compare single versus multiple infections, as well as HPV 16 versus other HPV infections, in the different age groups.
RESULTS
DNA could not be isolated from material of 29 out of 1,936 women (1.5%), as witnessed by negative β-globin PCR. Abnormal cytology was detected in 37 of the 1,907 remaining women (1.9%), and the prevalence of HPV was found to be significantly higher in women with abnormal cytology compared to women with normal cytology (p < 0.001). HPV positivity was detected in 68 of 1,870 women with normal cytology (3.6%), 5 of 19 women with ASCUS (26.3%), 3 of 15 women with LSIL (20%) and 3 of 3 women with HSIL (100%).
After dividing the women into 5 age groups, i.e., aged 50–54 (n = 701), aged 55–59 (n = 492), aged 60–64 (n = 321), aged 65–69 (n = 171) and aged 70 years or above (n = 222), the relation between age and HPV prevalence was studied (Fig. 1). Results are shown for all women and for women with normal cytology alone. After an initial decline, a higher prevalence can be seen with increasing age, which is even stronger in women with normal cytology. This trend, however, is not statistically significant (p = 0.60).
Prevalence of HPV in different age groups. White bars, all women; black bars, women with normal cytology alone.
Type-specific HPV prevalence was also investigated (Table I). Of the 14 high-risk HPV types included in the PCR-EIA format, 2 types (HPV 52 and 68) were never encountered, whereas 2 other types (HPV 33 and 51) were only present in multiple infections. HPV type 16 was by far the most prevalent HPV type in single infections followed by HPV types 18 and 31. These same types were also the most prevalent when multiple infections were included, with the addition of HPV type 35. There is a decrease in the prevalence of multiple infections with age as shown in Figure 2, paralleled by an increase in single infections, especially of HPV type 16 in the eldest-age group. However, neither the decrease in multiple infections (p = 0.14) nor the increase in single infections (p = 0.10) is statistically significant. Furthermore, multiple infections occurred equally frequently in women with normal cytology and in women with abnormal cytology (35% vs. 44%; p = NS).
Table I. Genotype specific prevalence of HPV
HPV type
Age group
50–54 (n = 701)
55–59 (n = 492)
60–64 (n = 321)
65–69 (n = 171)
70+ (n = 222)
Total (n = 1907)
S
M
S
M
S
M
S
M
S
M
S
M
S, single infections; M, multiple infections (double or triple).
a
HPV type could not be determined due to lack of material.
Prevalence of single and multiple infections and HPV 16-related infections in different age groups. White bars, single infections; black bars, multiple infections; grey bars, infections (both single and multiple) with HPV 16.
These data showed that 0.26% of the women needed immediate colposcopy because of ASCUS cytology in combination with the presence of high-risk HPV, and 5.7% of the women needed a follow-up smear after 12 months, either because of a β-globin-negative result or HPV-negative ASCUS or HPV-positive normal cytology. The large majority of 94.0% (99% confidence interval = 92.6–95.4%) of the women were both cytologically normal and HPV-negative and therefore would need no further screening.
DISCUSSION
Overscreening for cervical cancer in Flanders, Belgium, is quite common.11 A possibility to control one aspect of overscreening, screening up to old age, is the determination of a safe age to stop screening. The inclusion of HPV detection as an adjunct to cytology at this age provides further safety and may make cessation of screening more acceptable for elderly women if they are properly informed of the strong negative predictive value of the absence of HPV in the cervix. If screening could be stopped at age 55 or even 50 on the basis of HPV testing, this would save an important amount of money, which could be directed toward inclusion of women (of all ages) who are at present not screened, or toward other cancer screening programs, such as breast cancer.
The decrease in HPV prevalence with age has been reported in several studies,3, 4, 5, 6, 7, 8, 9, 10 but this prevalence has not been thoroughly studied in the elderly population. Here, we report a slight, nonsignificant increase in the high-risk HPV prevalence with age in a population of 1,907 women of 50 years and older.
Although detection of HPV was performed on residual material after preparation of thin-layer cytology slides, only 29 out of 1,936 (1.5%) samples were β-globin negative, which compares well with 3.1% found in a study using separate brushes for cytology and HPV detection.19 The HPV prevalence in our study is generally low (3.6%), but after an initial decline, the prevalence increased from the age of 55 to peak in the age group of 70 years and above, although this trend did not reach statistical significance. The existence of a second peak in the HPV prevalence has been reported in 2 recent studies.12, 13 In the Costa Rican study,12 the second peak was found earlier and was almost entirely due to an increase in low-risk HPV types. Although a decrease in immunocompetence with increasing age is part of a natural process, this decrease is not comparable to immunosuppression. It is therefore highly unlikely that low-risk HPV types play an important role in the development of cervical lesions in elderly women. In the Canadian study,13 a limited number of low-risk HPV types (6, 11, 42 and 53) were included, but the second peak was entirely due to high-risk HPV types, as it was in our study, where we only studied high-risk types. Furthermore, all of these studies are cross-sectional studies rather than cohort studies and consequently cannot rule out differences between the women in the different age groups, for instance, in exposure to HPV. Although the studies are consistent with an increase in HPV prevalence with older age, only a cohort study can actually prove this increase as a consequence of age.
There is no consistent difference in the prevalence of specific HPV types with age. The presence of multiple infections, however, is much more pronounced in younger-age groups. The prevalence of HPV type 16 in the eldest-age group was 2.25%, which was the highest prevalence of all age groups. It has been shown previously that women with multiple infections are at higher risk for the development of cervical lesions.20 The presence of HPV type 16 per se has also been shown to convey a higher risk of cervical lesions.12, 20, 21 The results of our study are somewhat puzzling in this aspect, since on the one hand they seem to suggest a lower risk because fewer multiple infections are detected, while on the other hand more HPV type 16 infections are detected, which suggests a higher risk. However, persistence of infection is even more predictive of the development of cervical lesions.22, 23, 24 To our knowledge, follow-up data on HPV infections are not yet available for elderly women, and hence the clearance rate of infection in these women is unknown.
The initial hypothesis of our study was that cervical cancer screening could be stopped in HPV-negative women with normal cytology at or around the age of 50. Up to 25% of the cases of cervical cancer occur in women over the age of 65. However, a long episode of persistent HPV infection will precede the development of cervical lesions and hence these women should be detected in a screening system combining cytology and HPV detection. A few studies have shown that in a country with a nationwide cervical cancer screening program, there is a rationale for withdrawing low-risk women from screening.25, 26, 27 In a country without population-based screening, where women are not bound to one general practitioner, and where no continuous patient record accompanies the woman upon switching healthcare provider, withdrawal of low-risk women necessarily needs to be based on the last smear. In that case, the addition of HPV testing to cytology at the time point of the final smear will provide optimal information to distinguish between low- and high-risk women. The data obtained in this study show that, even in the presence of a slight increase in the HPV prevalence in elderly women, approximately 94% of the elderly women could potentially be withdrawn from the cervical cancer screening. It would have been most interesting to know how many of the 82 women who needed follow-up would eventually have been referred for colposcopy. However, this could not be investigated within the format of the present cross-sectional study. In general, data concerning (re)infection of elderly women with HPV are lacking at present. We are currently performing a follow-up study to determine the frequency of (re)infection as well as the course of an HPV infection in elderly women.
Acknowledgements
The authors acknowledge the generous gift of plasmid clones by Dr. E.M. de Villiers (Deutsches Krebsforschungszentrum, Heidelberg, Germany), Dr. A.T. Lorincz (DiGene Diagnostics, Gaithersburg, MA), Dr. Y. Matsukura (National Institute of Health, Tokyo, Japan) and Dr. G. Orth (Institut Pasteur Paris, France). They are grateful to all patients, gynecologists and general practitioners who participated in this study, as well as to Ms. Aziza el Hachimi for sample logistics and Dr. John Sellors (PATH, Seattle, WA) for careful review of the article.