Invasive cervical cancers in the United States, Botswana and Kenya: HPV type distribution and health policy implications

An estimated 528,000 cases of invasive cervical cancer (ICC) occurred worldwide in 2012 [1]. More deaths occur in African women from cancer of the cervix than from any other malignancy. The highest incidence rates of ICC occur in Eastern and Southern Africa, with age standardized rates (ASR) of 42.7 and 31.5 per 100,000 women, respectively [1].

Many obstacles exist for sub-Saharan countries in the fight against cervical cancer, where this malignancy is diagnosed most often at ages 35 to 45 years, claiming the lives of women when they may be raising children and contributing to the economy of their communities [2]. All but the least progressed stages of cervical cancer have high case-fatality rates in this region of the world (estimated to be greater than 50 %) due to poor access to effective treatment modalities. Women with advanced cases have 2-year survival rates of less than 20 % [3], and the average life-years lost due to cervical cancer is greater than 18 [4]. Effective screening programs utilizing cervical cytology have decreased the number of cases of ICC in developed nations. However, screening is rarely performed in sub-Saharan Africa, where more than 50 % of global cervical cancer deaths occur.

Infection with oncogenic human papillomaviruses (HPV) is the cause of ICC [5]. Cervical cancer can be prevented by vaccination against HPV infection, but the distribution of HPV types in invasive cancers has not been fully characterized, especially in women infected with the human immunodeficiency virus (HIV) [6].

As sub-Saharan African countries consider implementing HPV vaccination, it is important to understand the range of HPV types responsible for ICC cases in a country. Surveillance and better understanding of local epidemiology must be encouraged before selecting the HPV vaccine of choice for a specific country. The object of the study was therefore to compare HPV typing data from a previously published study of ICC cases in the United States and Botswana [7] to additional cases identified in a second sub-Saharan country, Kenya, and to determine the HPV type distribution in invasive cervical cancers from HIV-infected women living in these countries.

It is important to understand differences in the HPV types causing ICC in different regions of the world, and differences in the HPV types causing ICC between women who are infected or not infected with HIV. This information is paramount in the fight against cervical cancer, especially in the era of safe and effective vaccines against HPV. It is entirely possible that HPV type distributions could differ in ICC cases between different countries in a large continent such as Africa. In addition, because there are several available methods for both collection of cancer cells/tissue and for HPV analysis, comparing studies can be difficult.

More country-specific data is needed before conclusions can be drawn. Two analyses of studies performed in African women suggested that HPV 16 was present in approximately half of ICCs, the lowest prevalence among the major regions of the world [12]. Our own previous study of ICCs obtained from women in Botswana (and included in this study) utilized thin section PCR of actual ICC tissue. DNA was directly extracted from sections of ICC and analyzed using the Roche Linear Array assay. The analysis demonstrated a prevalence of 42.7 % for HPV 16, compared to 87 % of ICCs from the U.S. [7].

We wanted to expand the analysis, using identical methods of tissue collection and HPV typing to a second sub-Saharan country, Kenya. The current study was therefore designed to compare and contrast HPV types causing ICC in the U.S., Botswana, and Kenya. Using a direct approach known as thin-section PCR, we found that the distribution of HPV types in ICC specimens from western Kenyan women was similar to that in cancers from the U.S. No statistically significant differences were found for individual HPV types or for groups of types. HPV 16 was the most frequently detected type in ICCs from women living in the U.S. (87.0 %) and from women living in Kenya (82.0 %). Together, HPV types 16 or 18 were detected in approximately 94 % of ICC specimens from both the U.S. and Kenya.

In contrast, the differences between HPV types in ICCs from women living in Kenya and Botswana are striking and likely not exclusively related to HIV infection prevalence. Fewer ICCs from women living in Botswana contained HPV 16. More non-HPV 16/18 oncogenic types were identified in ICCs from women living in Botswana compared to those from the U.S. or Kenya.

Other groups have also examined ICC cases from sub-Saharan African women, but in each case, different methods were utilized in both sampling and in HPV detection, compared to our study and to each other. Several other studies have included data on HIV status as well, but utilized serological HIV testing, as such information was available. For example, De Vuyst et al., conducted a study of Kenyan women with ICC [13]. Fifty-one HIV-infected women were matched by age to 153 HIV-uninfected women. Exfoliated cervical cells from women with ICC were collected and tested for HPV DNA using the PCR-based SPF10-INNO-LiPA system. The actual tumor was apparently not sampled in this study. Multiple-type infections were more frequent in HIV-infected (37.2 %) than in HIV-uninfected (13.7 %) women, but the overall distribution of HPV types was similar. HPV16 was detected in 41.2 % versus 43.8 % and HPV16 and/or 18 in 64.7 % versus 60.1 % of HIV-infected versus HIV-uninfected women, respectively.

Another study from this group was performed to determine associations of HPV infection and cervical intraepithelial neoplasia (CIN) among 498 HIV-infected Kenyan women who underwent HPV PCR-based testing, cytology, and cervical biopsy [14]. Overall, 68.7 % of women were HPV-positive, 52.6 % with HR-HPV, and 40.2 % had multiple type infections. The most frequent types in 113 CIN2/3 cases were HPV 16 (26.5 %), HPV 35 (19.5 %), and HPV 58 (12.4 %). CD4 count was negatively associated with detection of HR-HPV (P < 0.001) and CIN2/3 among women not receiving anti-retroviral treatment (P = 0.013). The authors concluded that the burden of HR-HPV and CIN2/3 was high in HIV-infected Kenyan women and was related to immunosuppression level. A third study from that group examined frozen tissue biopsies from women with ICC in Kenya and South Africa [15]. One hundred and six HIV-infected and 129 HIV-uninfected women with ICC were included. A fragment from the biopsy was tested for HPV DNA using GP5+/6 + -PCR assay Multiple HPV types were identified in 21.6 % of ICCs from HIV-infected women compared 3.3 % in HIV-uninfected women. HPV16 and/or 18 prevalence combined was similar in ICCS from HIV-infected (66.7 %) and HIV-uninfected women (69.1 %) No significant difference was found for other HPV types between the two groups. The authors concluded that current prophylactic HPV vaccines against HPV 16 and 18 may prevent similar proportions of cervical cancers in HIV-infected and HIV-uninfected Kenyan women.

A recent study by Maranga et al., examined HPV types in cervical smears and carcinomas from HIV-infected and HIV-uninfected Kenyan women [16]. Smear samples from women were analyzed, and those from HIV-infected women had a higher rate of detection of HR-HPV types 52, 58, 68, potential high risk 53/70, and low-risk 44/55 compared to HIV-uninfected women. HPV types 16/18 were detected in cells collected from ~46 % of HIV-infected women and ~46 % of HIV-positive women. HPV 45 was more common in cells collected from HIV-infected women. The authors concluded that HIV infection may alter the spectrum of HPV types found in cervical smears and in ICC.

HIV co-infection accelerates the natural history of cervical cancer. For Kenya and Botswana, we did not know the HIV status for the women. Using a thin-section PCR devised by our group, HIV was identified in ~32 to 34 % of ICC specimens from women living in Botswana or Kenyan, compared to ~4 % of ICCs from women living in Indiana. These numbers do not reflect the HIV prevalence of all women in these countries but illustrate the higher incidence of ICC in HIV-infected women. More studies of the thin-section PCR for HIV are needed to validate this potentially useful assay. Based on studies by other groups, we expected that ICCs from women who were HIV-infected may have a different HPV type distribution that from ICCs from HIV-uninfected women in these countries [17]. However, when ICC cases were stratified by HIV status, no significant differences were found for any HPV type or groups of types. The reason for this observation is not clear, and further study is needed.

The current study has implications for health policies related to prevention of cervical cancer. This study indicates that HPV types 16 and 18 are the primary types causing ICC in Indiana, U.S., and in western Kenya. However, our recent study in Botswana (and shown again in this manuscript) showed a different HPV type distribution. As indicated above, three HPV vaccines are currently available: the bivalent vaccine, the quadrivalent vaccine, and the nine-valent vaccine. All are expensive, costing approximately $360 in the U.S. for the series of three required doses. All three provide excellent protection against HPV 16 and 18. The nine-valent HPV vaccine includes VLPs present in the quadrivalent vaccine (HPV types 6, 11, 16, and 18) plus VLPs of five additional HR-HPV types (31, 33, 45, 52, and 58) [11].

When global data is examined, the five additional HR-HPV types in the nine-valent HPV vaccine account for approximately 12 % of all cervical cancers [11]. Combined with HPV 16 and HPV 18 estimates, the nine-valent HPV vaccine has the potential to protect against nine types responsible for greater than 90 % of cervical cancers. The U.S. Food and Drug Administration (FDA) recently approved the nine-valent vaccine for use. A cost-benefit analysis would be required to determine if the newer vaccine would be worth any additional cost, in terms of life-years gained compared to the other available HPV vaccines.

How important would the nine-valent HPV vaccine be in Botswana? Due to the large percentage of non-HPV 16/18 types identified in the ICC cases in Botswana, the bivalent and quadrivalent vaccines would theoretically provide less protection against ICC than it would for women in Kenya or the U.S. The nine-valent vaccine would theoretically provide a higher degree of protection against ICC among women living in Botswana.

There are several limitations to this study. First, only ICC cases from western Kenya were included and results cannot be generalized to the entire country of Kenya. Second, this study used paraffin-embedded cervical cancers, and as a result, the sensitivity for HPV detection could be less than if fresh cancer tissues had been available. Because the current study used archived specimens, fixation conditions could not be controlled and may have contributed to differences between countries in the frequencies of HPV detection [18]. Third, it is possible that the techniques used to recover DNA and detect/type HPV have differing capabilities for certain HPV types, thus potentially underrepresenting non-HPV 16/18 HPV types. However, the Roche Linear Array is the standard detection and typing method in epidemiologic studies of HPV, and in other studies has demonstrated the capacity to detect numerous HPV types in cervical cancers [19]. Fourth, only 85 to 89 % of the samples from African women were analyzed due to inadequate epithelium histologically, and the inability to identify HPV in roughly 20 % of the African specimens could shift the results and conclusions, given the small size of the study. Fifth, the proportion of HPV-negative cancers in Botswana (among those with adequate beta-globin DNA) is curious and could be related to differences in fixation and processing techniques as mentioned above. It is possible that HPV types in some of these ICCs are not detected in the Linear Array Assay. Sixth, previous studies detected HPV 16 in ~70 to 80 % of ICC cases, compared to >90 % in our study for the ICCs from the U.S. and Kenya. While it is possible that our methods overestimate HPV 16/18 prevalence in ICCs, there is no evidence to support this. Most previous studies used cervical swabs rather than thin-section PCR to capture tumor cells and to determine HPV types in the cancers but this could lead to an overestimation of the role of certain HPV types when compared to thin-section PCR performed directly on ICCs. Lastly, the HIV status of women in Kenya and Botswana was not known. Therefore, an indirect method a thin-section PCR was developed to determine the HIV status of these women. The sensitivity and specificity of this novel method has not yet been determined, but we verified the authenticity of HIV sequences in two ICC specimens.

Continued efforts must be made to improve utilization of the HPV vaccines, and with the help of the World Health Organization and other agencies, and hopefully a reduced price of the nine-valent HPV vaccine can be negotiated for countries in which non-HPV 16/18 types are frequently identified in careful studies of ICC. For now, every effort should be made to vaccinate young girls and adolescent women with any of the currently available HPV vaccines. This will help assure a cervical cancer-free future for Kenya, Botswana, and other sub-Saharan countries where cervical cancer screening is not widely available.