Study design and subjects

This study was a comparative frequency-matched study conducted among women ≥18 years attending the HIV or medical outpatient clinic of the Cape Coast Teaching Hospital (CCTH) in Ghana. A systematic sampling of every 5th woman was used starting with a random sampling of the first woman. Women who were ineligible (previous total hysterectomy, pregnant or menstruating on that day), had the opportunity passed on to the next until a total of 10 women per day were selected. After recruitment, the study protocol was explained to each woman and written informed consent was then obtained.

Enrolled women then had their HIV status confirmed and then they answered a questionnaire in either English or the local language (Fante). This questionnaire gathered socio demographic characteristics of the women and then for HIV positive women, the HIV specific characteristics including use of ART, nadir CD4+ T-lymphocytes count, and WHO clinical staging were also obtained (Additional file 1).

Sample collection

Blood was drawn to perform HIV serology using the First Response Kit (Premier Medical Corporation Limited, India) to detect HIV-1 and HIV-2 antibodies, confirmed with the OraQuick kit (OraSure Technologies, USA), as per national guidelines. Counseling based on the results was done for all women and women found HIV-1 seropositive but not already in care were referred to the ART treatment center at CCTH. Participants underwent gynaecological examination with speculum, during which cervical swabs were collected from the ecto/endocervix targeting the squamo-columnar junction using a DNA PAP™ Cervical Sampler™ and transported in the Swab Specimen Collection Kit (Qiagen, Gaithersburg, MD). Cervical smears were taken for cytology with a cervical brush and alcohol-fixed at the clinic. All women found to have LSIL+ cytology were referred to the gynaecologist for further evaluation and management.

HPV DNA detection

HPV genotyping was performed using the recently developed Anyplex™ II HPV28 assay (Seegene, Seoul, Korea). The assay detects 28 HPV genotypes including 19 h-HPV types, of which 13 are considered carcinogenic (HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68) and 6 possible carcinogenic (HPV26, 53, 66, 69, 73, 82), and 9 low-risk HPV types (HPV6, 11, 40, 42, 43, 44, 54, 61, 70), according to the Interagency for Research on Cancer (IARC) [31]. This assay has been compared with the Digene HC2 HPV DNA assay (Qiagen) and found to be analytically more sensitive in detecting the 13 h-HPV types identifiable by both assays, in addition to having higher concordance with comprehensive genotyping based on sequencing analysis [32].

The isolation of nucleic acid was done by QIAamp DNA Mini kit (Qiagen, USA) following manufacturer’s instructions. Extracted DNA samples were aliquoted and stored at -20 °C. The next step was the multiplex real time PCR using the Anyplex™ II HPV28 (Seegene, Korea) following manufacturer’s protocol with the CFX96™ Real-time PCR System (Bio-Rad, USA). In brief, this involved preparation of a master mix which contained RNAase free water (5 μl per sample), 4X Anyplex solution (5 μl per sample) and 4X HPV28 TOM A or TOM B solutions (5 μl per sample for each type). TOM A contained the primers for hr.-HPV genotypes, while TOM B contained primers for non hr.-HPV types. Both reactions were set up to run concurrently. To 15 μl of reagents was added 5 μl of the prepared DNA sample.

A set of 3 external positive controls and a negative control were included in each set to run with the samples. To be valid for any sample, both the internal and external controls must have expected result. When the external controls failed the whole set was invalid and no results were read. When the internal control for a sample failed, whilst all external controls were correct, that particular sample result was considered invalid and it tested again.

Cervical cytology

Following a standardized protocol for Papanicolaou (pap) staining, cervical smears were prepared in the laboratory and read by a consultant cytopathologist at CCTH using the Bethesda 2001 guidelines and the LAST guidelines [33, 34].

Sample size and statistical analysis

Sample size was calculated to allow comparisons of HIV-1 seropositive and HIV-seronegative women. Assumptions on HPV and cytological abnormality prevalence in each group were based on findings from different studies around the Africa region and sample size was powered to detect differences in the lowest frequency outcome (cytological abnormalities). Data analyses were performed using Stata version 13 software (STATA Corp, Texas USA).

A descriptive analysis of socio-demographic, behaviour and other relevant characteristics of the study population was done according to HIV serostatus. P-values were used to compare the parameters between the groups based on student’s t-test for continuous variables or chi- square test for categorical variables. Bivariate analysis was done separately for the two study outcomes (HPV and cytological abnormalities) stratified by HIV serostatus.

Based on outcome frequency, associations are presented as prevalence or risk ratios (PR/RR) with 95% confidence intervals (CI) for the HPV outcomes, and odds ratios (OR) and 95%CI for the cytological abnormality outcomes. Variables with P-values ≤0.20 and a priori factors like age were put in the model for multivariate analysis.

Study participants

Prevalence of HPV, genotype distribution and risk factors for hr.-HPV

Prevalence of, and risk factors associated with abnormal cytology

The distribution of cytological results by HIV serostatus is shown in Table 1. The prevalence of abnormal cytology was higher among HIV-1 seropositive women (any SIL: 14.1% vs. 1.2%, P < 0.0001; low-grade SIL [LSIL]: 4.9% vs. 0.6%, P = 0.02; high-grade SIL [HSIL]: 1.8% vs. 0%, P = 0.07). HPV35 was the most frequent hr.-HPV type associated with LSIL and above (LSIL+) (Fig. 1).

HPV genotype prevalence and associated risk factors

Since protection afforded by HPV vaccines is essentially type specific [40, 41], knowledge of the genotype distribution in a specific population has implications for vaccine choice and predicted impact. The most prevalent hr.-HPV type in both groups in this study was HPV35, which none of the available vaccines on the market includes. The quadrivalent HPV vaccine is used in most countries and covers for only HPV6, 11, 16 and 18 [42, 43]. The recent introduction of the nonavalent vaccine to cover for additional HPV31, 33, 45, 52 and 58 may lead to a change in vaccine choice among countries and countries yet to implement wide scale or routine vaccination programs including Ghana.

HPV16 and 18 are said to contribute to about 70% of cervical cancer cases [44] and this has influenced vaccine development. This study demonstrated the association between HPV35 and abnormal cytology in Ghana. A study in neighbouring Burkina Faso found HPV35 to be the second most prevalent among sex workers with high prevalence of HIV [37]. Another study conducted in South Africa reported HPV35 as the third most prevalent genotype identified among 154 HIV-seronegative women with ICC [45]. Pirek et al. [46] found HPV35 the fifth commonest among Cameroonian women with ICC and HPV45 was the second most prevalent type. Maranga et al. [47] in their study among Kenyan women also found that HPV45 contributed significantly to cervical cancer among HIV-seropositive women. Attoh et al., [25] in their study among Ghanaian women with cervical cancer detected 8 h-HPV genotypes (16, 18, 35, 39, 45, 52, 56 and 66) with HPV18 being the most prevalent. Another study based in sub-Saharan Africa with inclusion of samples from Ghana found that HPV type distribution appeared to differ according to tumor type and HIV status and HPV16, 18, 45 and 35 were the most common HPV types in women with ICC [48]. In this study, hr.-HPV types included in the bi- or quadrivalent vaccines (HPV16/18) and nonavalent vaccine (HPV16/18/31/33/45/52 and 48) were found in higher proportions among HIV-1 seropositive compared with HIV-seronegative women.

Among study participants, there was evidence of an association between having hr.-HPV infection and younger age among HIV positive participants and lack of circumcision of the male partner among both HIV positive and negative groups. These findings are consistent with existing literature [49, 50]. Other studies in Ghana and other parts of the world have reported association between employment status, marital status and educational level with HPV [51, 52].

Epidemiology of cytological abnormalities and associated risk factors

Compared with HIV-seronegative women, HIV-1 seropositive women in this study had a higher prevalence of SIL and higher grade cytological lesions. Indeed, HSIL/ASC-H were only identified among HIV-1 seropositive women. A study conducted among women in selected communities in the Ashanti region of Ghana found any SIL prevalence ranging between 3.5% and 12.6% [53], although this study did not report HIV status. Elsewhere in sub-Saharan Africa, Hood et al. reported that cervical lesions were significantly associated with the detection of plasma HIV RNA (with an adjusted relative risk of 1.16, 95% CI: 1.05–1.28) among women in Senegal [54]. Another study in Kenya among 267 HIV-seropositive women on ART found a much higher prevalence of abnormal cytology of 46%. The median duration of antiretroviral therapy was 13 months (IQR: 8–19) [55]. In another study conducted in South Africa among 109 HIV-seropositive women before initiation of ART, the prevalence of abnormal cytology was 66.3%. The median CD4 count among these women was very low at 125 cells/mm³ [56].

In the sub-group analysis for HIV-1 seropositive women, factors which showed strong evidence of an association with SIL included nadir CD4+ T-cell count. A higher CD4+ T-cell count reflects a stronger immune system, which may be associated with greater ability to clear HPV infection compared with women who acquire HPV whilst more seriously immunocompromised, which may lead to more frequent persistence and a greater chance to develop lesions. The association with nadir CD4+ T-cell count has been demonstrated in other studies [5, 6, 36]. This study had some limitations. It was facility-based, which allowed for more convenience in recruitment, but its results may not be generalizable. Histology confirmation of lesions would have also been preferable as it is more specific diagnosis than cytology, although this has been a usual limitation of several epidemiological studies in sub-Saharan Africa. An important limitation is the unavailability of current HIV viral load and CD4 count for the HIV positive participants. Nonetheless, this study brings useful information for decision makers in the country, demonstrating that the burden of HPV infection and hr.-HPV infection is high in Ghana. The study also showed that the most prevalent HPV genotype was HPV35, and that among HIV-seronegative women, HPV16 and 18 were not among the top five most prevalent genotypes. The established contribution from HPV 16 and 18 has appropriately informed vaccine choices, nevertheless, the genotype distribution pattern found in this and other studies imply the need to look carefully at vaccine choices for primary prevention of HPV and cervical cancer among women in Ghana.