Discussion
In order to increase the overall proportion of women participating in HPV screening programs, efforts are focused on the incorporation of noninvasive self-sampling methods of vaginal and urine samples, as previously reported by our team.[38,39,53] Outside pregnancy, vaginal self-obtained samples and have been the most studied option and have been shown to be appropriate alternatives for physician-collected cervical sampling, in particular in underserved women.[33–37] Urine sample, as a noninvasive self-obtained sample is better accepted and preferred by women.[38,47] Urine HPV DNA is due to contamination with exfoliated HPV-infected cells from the lower genital tract. The high sensitivity of DNA amplification methods allows the use of samples away from the site of infection, while the increasing number of exfoliated epithelial cells increases the detection sensitivity. Owing to the close anatomical relationship of the lower urinary tract with the cervix and lower genital tract, it is considered that the urethral epithelium would also be vulnerable to HPV infections,[25] and it was suggested that the presence of HPV DNA may also be due to a HPV infection of the urinary tract.[50] In this recent review, it was concluded that HPV DNA detection in urine is feasible and may become a useful tool, but needs further improvement and standardization.[50]
The method of measurement is important and Sehgal et al. concluded that further research is needed to standardize and optimize the corresponding technology before recommending it.[25] There are many commercial automated broad-spectrum hr-HPV tests available for screening large populations for which our results could find application. Owing to financial restrictions, we used an inhouse method for type-specific hr-HPV testing that was developed and applied successfully previously.[38] In this previous study, we found that urine/cervix HPV detection sensitivity was higher for high-grade lesions and correlated with the number of epithelial cells present in urine samples, implying that it could be applicable in certain populations such as pregnant women who present more epithelial cells in urine. Indeed, this method has proven accurate in the urine of pregnant women and confirmed the feasibility of hr-HPV testing in pregnancy. However, it is time consuming and requires surcharge for laboratory personnel.
A recently published study using a PCR-based method, similar to ours but with another primer set, showed a very high sensitivity of 98.6% and a specificity of 97.4%, with a very high negative predictive value (97.4%; 95% CI: 87.7–99.9%) in first-voided urine versus cervix HPV DNA detection.[60] In addition, studies that have reported the prevalence of HPV during pregnancy were not conclusive. Some studies did not observe a higher HPV detection rate in pregnant compared with nonpregnant women,[61–67] whereas others actually showed higher prevalence of HPV in pregnancy.[68,69]
Reinforcing urine testing in hr-HPV detection as a reliable noninvasive method, we assessed self-sampling vagina and urine HPV detection in pregnant women, since antenatal visits represent an opportunity to screen for hr-HPV, and compared them with matched nonpregnant women. We chose pregnant women because for women with a low regular screening compliance whorepresent a significant proportion of the female population, this may be a unique opportunity to undertake a HPV test. Even if we cannot intervene during pregnancy, women would benefit from therapy in a postpartum visit, if needed, and perhaps would better accept a screening program afterwards, especially with an easily obtainable sample. We must not forget that some of these nonattendees would not present to the doctor until the cancer was symptomatic if they did not have such a screening during pregnancy and this could cause considerable public health implications. In the literature, only four studies using self-obtained vaginal samples for HPV detection included pregnant women. Three of them did not report a difference in performance between the pregnant and nonpregnant population comparing self-sampling with physician sampling,[67,70,71] while the fourth reported a higher hr-HPV prevalence in pregnant women.[68] To our knowledge, no prior studies have evaluated the feasibility and utility of urine samples for hr-HPV screening in pregnant women.
Urine HPV DNA is due to contamination with exfoliated epithelial cells. In approximately 10–20% of samples, urine either does not contain adequate cells or possibly has inhibitors, leading to low detection rates. Along this line of thought, we used first-void urine and our previously described PCR method for urine HPV detection.[38]
Furthermore, after spiking HPV-negative pregnancy urine specimens in our study with CaSki and HeLa cells, it was indicated that, at least for the purpose of our study, urine from pregnant women with normal urinalysis and urine microscopy (inclusion criteria) does not contain significant inhibitors and can be used for hr-HPV detection. Our findings in pregnant women, namely that 73% of urine samples from pregnant women contained two or more squamous epithelial cells (minimum to allow HPV detection[38]) compared with 52% for nonpregnant women and that the amount of epithelial cells correlated with HPV detection, indicates that screening in pregnancy might offer a potential advantage. Comparing HPV prevalence overall and for each hr-HPV type in urine and vaginal samples between pregnant and nonpregnant women demonstrated no significant differences in overall HPV proportions distributed across the two study populations (all p-values >0.05).
We did not examine cervical smears with cytology in pregnant women, as they are considered to give noninterpretable results in pregnancy. In addition, HPV testing has, during the last years, evolved into a well-accepted primary screening alternative in certain settings. We rather focused on the feasibility of self-collected urine and vaginal samples as an alternative screening approach in pregnant women for the detection of HPV infection. A limitation of our study is the very small number of detected infections.
HPV Self-sampling Results in Our Region Compared With the Rest of Greece
Our study took place before the introduction of HPV vaccines in Greece, the quadrivalent vaccine that protects primarily against HPV-16, -18, -6 and -11 (GARDASIL®; Merck Sharp & Dohme, NY, USA), and the bivalent vaccine that protects primarily against HPV-16 and -18 (CERVARIX®; GlaxoSmithKline, Brentford, UK). In a large, retrospective, cross-sectional worldwide study of HPV genotype attribution in invasive cervical cancer, HPV types 16, 18 and 45 were the three most common types in each histological group (squamous cell carcinoma, adenocarcinoma and adenosquamous cell carcinoma).[72] Therefore, we focused on hr-HPV-16 and -18, which are included in both vaccines, and HPV types 31 and 45, which are closely related phylogenetically to HPV-16 and HPV-18, respectively.
In addition, the present study provides some information about the infection rate of HPV in this particular region of Central Greece that was not included in a national study from Greece.[73] Our sample had a seemingly low overall hr-HPV detection rate (2.9%), which was similar to the largest screening population study in Greece conducted at approximately the same time interval,[73] and was consistent with the results of a recent meta-analysis,[74] indicating that Southern Europe is among the areas with the lowest HPV prevalence. With regard to the HPV types tested (16, 18, 31 and 45), the detection rate in the national study was 2.4%. This is consistent with our detection rate of 2.7% in pregnant and 3.2% in nonpregnant women.
Furthermore, in that study, the most common type detected, HPV-16, had a lower prevalence (1.4%) than described worldwide (2.5%), however, being similar to that described for Southern Europe (1.2%)[74] and similar to our pregnant (1.5%) and nonpregnant women (2%). Comparable detection rates were also observed for HPV-18, -31 and -45.
Acceptance & Feasibility of Self-sampling in Pregnancy & Its Influence on Future HPV Screening Plans
It has been shown that in different countries, population coverage in health infrastructural-based screening programs does not usually exceed 60%.[75] In developed countries, as in our study, methods that used self-collected samples have been shown to be more user-friendly (88% acceptance for self-collected vaginal sampling, 93% for vulvar sampling and 98% for urine sampling) in comparison with 79% acceptance for samples collected by physicians.[75] Our pregnant population confirmed this observation, since all agreed to supply the samples (100%). This might be of future clinical importance. Despite the lower detection rates of urine versus cervical samples,[38,39,53] a better acceptance of the urine-based program by women provides some compensation in terms of increased coverage. The 100% acceptance rate in pregnancy and the fact that women become available for hr-HPV screening create the need for the logistics of the cost:benefit ratios and the adoption of self-sampling to be worked out for these different settings, populations and HPV prevalences,[47] which was outside the aim of this pilot study.
Concordance of Vaginal & Urine Samples
As far as concordance between the two self-sampled specimens outside pregnancy is concerned, in a recent review, the concordance between paired cervical and urine samples ranged from 75 to 100% (or 41–93% by κ-agreement).[59] Cuschieri et al. who studied young females from a drop-in, sexual health service reported a κ-agreement of 0.60 for HPV DNA detection between urine and cervical samples, with a sensitivity of 90.5% and a specificity of 67.6%.[76]
In our pilot study, all vaginal HPV-positive samples from pregnant women were also positive in urine for the same HPV type (100% concordance) and there was no positive urine sample when the other sampling sites were negative, indicating that HPV in urine was due to an infection of the cervix and not the urinary tract. A lower concordance was observed in nonpregnant women. This is a pilot study and our sample size is too small. Our preliminary results should be confirmed in a larger trial, which is under development, in pregnant women compared with a matched nonpregnant population, in order to obtain firm conclusions. This is a novel finding for self-sampled HPV screening in pregnancy and might have clinical importance, since during routine low-risk antenatal care, patients are providing urine (urinary tract infection) or vaginal (streptococcal screening) samples at different times, and there have also been suggestions of chlamydial screening in pregnancy. A large cross-sectional study showed that urine sampling for Chlamydia trachomatis is equivalent to endocervical sampling in pregnancy.[77] hr-HPV and Chlamydia cotesting in urine might be a future cost-effective approach.
Future Virology. 2014;9(4):385-395. © 2014 Future Medicine Ltd.
