According to the World Health Organization (WHO), 33 countries have implemented HPV vaccination as part of their national immunization program.41
Generally, vaccination is provided to girls, and it is therefore recommended to monitor the impact of vaccination on HPV prevalence among female adolescents.15
In August 2012, we identified four publications that reported HPV prevalence in urine samples from asymptomatic female adolescents. HPV prevalence has been strongly associated with age, being nearly nonexistent in preadolescents, gradually increasing with sexual activity among adolescents, and generally peaking around 25 years of age.42
However, regional differences in HPV prevalence do exist.43
Therefore, the first step when monitoring HPV prevalence is to establish the baseline HPV prevalence in the relevant age groups. To date, only Hussain et al and O’Leary et al reported baseline HPV prevalence based on urine samples from a large unvaccinated adolescent population.24
The Hussain et al25
study from India used self-collected urine of the cervix and that care should be taken when samples from healthy children attending public school and achieved a 57.3% participation rate. O’Leary et al24
analyzed urine samples from 11–18-year-old school and college males (1121) and females (1341) in Scotland 4 months before vaccination was introduced in the national immunization schedule in 2008. A limitation of the study was that the estimated response rate for providing a urine sample was as low as 14%. While the low response rate can introduce bias and lead to erroneous estimates of the overall HPV prevalence, it was not directly related to the sampling method and probably refects a general challenge to achieve high response rates in this age group. This, in turn, could partially explain why only few studies measuring HPV prevalence in urine samples from female adolescents in the general population have been performed and published.
The five studies that reported HPV prevalence by lesion severity showed a similar association for paired urine and cervical samples, with higher HPV prevalence in the most severe lesions. There were major variations in HPV prevalence in urine samples across the studies. This is to some degree related to regional differences in HPV prevalence,43
the age distribution of the different study populations,43
and the setting in which the women were recruited. There were also differences in sampling procedures and HPV detection methods, including the number of types detected by a given assay. It is therefore not possible, as Vorsters et al17
pointed out, to perform a meta-analysis on the present urine-based HPV prevalence studies.
Detection of human genomic DNA is commonly used to control for the adequacy of samples for HPV detection. Studies including female populations showed a high detection rate of human genomic DNA in urine samples (83%–100%), while male populations showed a larger range of detection rates (30%–100%). In an HPV monitoring setting, a low human genomic DNA detection rate would lead to reduced coverage and create a concern of bias in HPV estimates. In general, studies on female populations in this review indicated that high detection rates of human genomic DNA are feasible.
In cervical screening the main focus is to detect HPV or cervical abnormalities at the individual level, while population-based HPV prevalence is used more in a monitoring or epidemiological setting. Therefore, although prevalence in urine samples was lower than in cervical samples in the studies included in this review, monitoring by regular urine measurements over time may still be a useful way of identifying shifts in HPV prevalence due to imparted immunity against vaccine HPV types. However, the differences in HPV concordance of paired urine and cervical samples illustrates that HPV positivity in urine should be interpreted independently of the cervix and that care should be taken when inferring that a similar change is taking place in the cervix. On the other hand HPV detection in urine samples could be considered an independent measurement of the impact of HPV vaccination but it would have only limited public health interest. Furthermore, we observed that HPV negativity in the cervix commonly predicted an HPV-negative result in the urine as well, while HPV positivity in the cervix less commonly predicted HPV positivity in the urine. Although Daponte et al showed an increased concordance with increased lesion severity for any HPV type, other studies like Alameda et al,26
Rymark et al,33
and Gupta et al30
showed a relatively high concordance, even in populations where HPV prevalence is low. The variability of HPV16/18-specific concordance, the types included in both of the available HPV vaccines, further exemplifies the uncertainty of using urine samples to estimate future changes in the incidence of cervical lesions.
The most comprehensive monitoring of changes in HPV prevalence would be carried out by establishing baseline HPV prevalence before measuring any impact of vaccination as well as regular measurements of HPV prevalence in both vaccinated and unvaccinated females and males. The age group (or groups) and sample size to include in HPV monitoring should be carefully selected to assure there is enough statistical power to identify changes in overall HPV prevalence as well as HPV type-specific changes. Models suggest that there will be a significant reduction in the prevalence of vaccine HPV types in males in the future because after vaccination, fewer girls will transmit HPV to their male partners.44
Monitoring HPV prevalence in males could therefore be a near-term end point that could also help to estimate the effect of herd immunity. However, monitoring HPV prevalence in males presents several challenges. The differences in HPV prevalence across different urogenital sites illustrates that no single site repeatedly shows the highest HPV DNA detection rate and that urine in particular has a relatively low HPV DNA detection rate compared to other sites. In addition, male urine generally has a lower detection rate for human genomic DNA than samples from other urogenital sites. With lower detection rates for human genomic DNA, a larger sample size would be needed to detect changes with the same power as other urogenital sites. Based on these aspects, other anatomical sites seem more favorable for males.
Monitoring changes in HPV prevalence requires regular prevalence measurements over many years. A protocol with sufficient detail on technical and practical issues that influence HPV detection is therefore necessary to ensure comparability between these measurements. This includes, among other issues, urine sampling procedure, handling of samples, extraction of DNA, and assay used for HPV genotyping.17
In addition it might be useful to store an extra aliquot of extracted DNA from each regular measurement to be able to perform HPV genotyping on all DNA collected from urine samples over many years. This would also allow for using any novel genotyping technology that may have developed during the monitoring period. Information on more aspects of HPV monitoring can also be found in the Human Papillomavirus Laboratory Manual issued in 2009 by the WHO HPV Laboratory Network (WHO HPV LabNet).45
This manual covers guidance on specimen collection and handling for HPV testing, with the aim to assist in establishing the laboratory support required for implementation and monitoring of HPV vaccination programs. Several of the WHO HPV LabNet members are actively undertaking studies of HPV detection in urine, and a leading role for the WHO HPV LabNet in further standardizing and optimizing the technology for HPV detection in urine seems appropriate.18
This is the first review that focuses solely on the use of urine to monitor changes in HPV prevalence in an asymptomatic population. The major shortcoming of this review is that, to date, there are few studies on the topic. We have therefore included studies from symptomatic populations and older populations that used urine for purposes other than monitoring, although these are not comparable to asymptomatic adolescents in all aspects. In addition, the studies highlighted in the present review as well as in the reviews of Vorsters et al17
and Seghal et al,21
showed that the large variability in sampling and genotyping methodology make direct comparisons of data, like concordance, inaccurate.17
Assuming a future reduction in overall HPV prevalence and vaccine HPV type-specific prevalence and using urine testing as a monitoring method, care should be taken when interpreting the data. Indeed the data may not necessarily mimic the true HPV distribution in the cervix nor estimate the expected reductions in cervical cancer and high-grade lesions, as indicated by variable vaccine HPV type-specific (HPV16/18) concordance between paired urine and cervical samples. There is great scientific and political interest in monitoring the early effects of HPV vaccination in the general population. However, monitoring HPV prevalence as an early measurement of vaccine impact is only possible in a few countries as substantial financial and human resources are needed as well as a 5–10-year commitment in order to demonstrate results.16