Cancer burden among HIV-positive persons in Nigeria: preliminary findings from the Nigerian AIDS-cancer match study

Background Although Nigeria has a large HIV epidemic, the impact of HIV on cancer in Nigerians is unknown. Methods We conducted a registry linkage study using a probabilistic matching algorithm among a cohort of HIV positive persons registered at health facilities where the Institute of Human Virology Nigeria (IHVN) provides HIV prevention and treatment services. Their data was linked to data from 2009 to 2012 in the Abuja Cancer Registry. Match compatible files with first name, last name, sex, date of birth and unique HIV cohort identification numbers were provided by each registry and used for the linkage analysis. We describe demographic characteristics of the HIV clients and compute Standardized Incidence Ratios (SIRs) to evaluate the association of various cancers with HIV infection. Results Between 2005 and 2012, 17,826 persons living with HIV (PLWA) were registered at IHVN. Their median age (Interquartile range (IQR)) was 33 (27–40) years; 41% (7246/17826) were men and 59% (10580/17826) were women. From 2009 to 2012, 2,029 clients with invasive cancers were registered at the Abuja Cancer Registry. The median age (IQR) of the cancer clients was 45 (35–68) years. Among PLWA, 39 cancer cases were identified, 69% (27/39) were incident cancers and 31% (12/39) were prevalent cancers. The SIR (95% CI) for the AIDS Defining Cancers were 5.7 (4.1, 7.2) and 2.0 (0.4, 3.5), for Kaposi Sarcoma and Cervical Cancer respectively. Conclusion The risk of Kaposi Sarcoma but not Cervical Cancer or Non-Hodgkin’s Lymphoma, was significantly increased among HIV positive persons, compared to the general population in Nigeria.


Background
HIV infection is associated with an increased risk of diverse cancers. Classically, the Centers for Disease Control and Prevention (CDC) designated Kaposi sarcoma (KS) cervical cancer (CC) and non-Hodgkin's lymphoma (NHL) as AIDS defining cancers (ADCs) because of the association of these cancers with increased risk of HIV infection. The advent of highly active antiretroviral therapy (HAART) in 1996 has led to significant reduction in the incidence and prevalence of these ADCs [1]. Nevertheless, the risk of ADCs remains high in many low and middle income countries (LMIC) where there are significant populations with HIV infection who are not on treatment. In contrast, in countries with mature HIV epidemics and combination ART, there is a modest increase in other cancers which have been designated non-AIDS defining cancers (NADCs), including anal, head and neck, lung and liver cancers, Hodgkin's lymphoma, and melanoma [2][3][4][5][6].
Most of the studies conducted to investigate the relationship between HIV infection and malignancies in sub-Saharan Africa to date have largely emanated from East and Southern Africa and this may not accurately reflect the pattern of AIDS associated malignancies in Africa [7][8][9][10][11][12][13]. Furthermore only one of these African studies was population based and it showed that about 70% of cancers occurring in the studied HIV infected population in Uganda were ADCs. The risks of five other NADCs -cancers of the kidney, thyroid, uterus, conjunctiva and Hodgkin's lymphoma -were also increased [12]. Apart from such systematic studies, other reports from Africa were based on hospital or pathology case series with a majority of such studies emanating from Nigeria, West Africa. Table 1 is a summary of the AIDS-associated malignancy studies among adults in Africa from January 2001 to December 2012.
Nigeria being the most populous country in Africa, with an estimated population of 170,123,740 (41% of whom are below age 15 years) in 2012 [23] and the third highest population of persons living with HIV/AIDS (PLWA) in the world, is expected to bear a significant proportion of the projected AIDS associated cancer burden in Africa. The incidence and prevalence of cancer among PLWA in Nigeria is not known. This lack of data hampers the ability to formulate sound policy for clinical and public health intervention. The successful implementation of record-linkage studies in different parts of the world, e.g. Uganda, India, suggests that this may be a feasible approach to the study of AIDS associated cancers in Nigeria.
We conducted this study to evaluate the spectrum of cancers diagnosed among PLWA in Nigeria using the record-linkage technique.

Discussion
This study is the first to use record linkage analysis to estimate cancer incidence among PLWA in West Africa. Among the ADCs, only KS was significantly increased among HIV positive persons and there was no statistically significant increase in incidence of NHL, CC and NADCs. These results for KS and CC are consistent with the findings from studies conducted in other countries where increased risk of KS among HIV positive persons and no association between CC and HIV was found [24][25][26]. Although several studies suggest that the burden of NHL is increased among HIV positive persons in Africa [12,13,27,28], studies from West Africa have not supported this, just like our results. Similar to Bolarinwa et al., we identified several cases of NHL in the general population but none of these were matched in the HIV registry [18].
The relatively low incidence of cancers among HIV positive persons in Nigeria is probably multifactorial. An important possibility may be due to lack of cancer diagnosis. Given the size of the HIV/AIDS epidemic and main focus to diagnose and treat HIV or roll out prevention programs, care providers may miss clinical features suggestive of cancer in persons with HIV infection or fail to investigate those clients thoroughly. Although not well documented, our causal impression is that some physicians who are concerned about occupational exposure to HIV may provide limited quality services to HIV positive persons [29,30]. Finally, the infrastructure to diagnose, refer, and register patients may also have caused low levels of utilization of diagnostic procedures needed to confirm cancer diagnosis in PLWA [29].
There is some evidence that ascertainment of cancer by the ABCR may not be complete. For example, the Kampala Cancer Registry, which covers a population of about 2.1 million, registered 4235 cancer cases during 2007 -2009 [31]. ABCR, which covers a population of about 1.4 million, registered 1815 cancer cases from 2009-2011 registered. This may be due to a longer experience with cancer registration resulting in more complete data or a lifestyle with a higher cancer risk profile in Uganda. We also note that the HIV prevalence is higher in Kyadondo compared to Abuja, 15% and 8.6%, respectively [12,32], which may lead to a higher burden of HIV-associated cancers, such as KS in Uganda.
In this regard, it is worth noting that the prevalence of human herpesvirus 8 is lower in Nigeria compared to Uganda [33], which would also explain the lower prevalence of cancer in ABCR, particularly KS. Our study highlights the opportunity to improve case finding methods of cancer registries in Nigeria. Similar to other countries in Africa, the infrastructure for population-based cancer registries is poorly developed and not fully linked to the HIV screening and registration process. Increased utilization of electronic medical records and databases, improved collaboration between registries and a multidisciplinary approach to data collection and management will improve case ascertainment of HIV and cancer cases, their report to registries and the linkage of registry data. A separate study evaluating the completeness of data obtained by the ABCR is under way. Other possible explanations for the low cancer incidence include the availability of HAART, which would be associated with improved immune function and reduced risk of AIDS-associated cancers, such as KS [34,35]. The coverage of HAART in HIV positive persons in Nigeria is relatively low (24% of individuals who should be on HAART [36]), thus the contribution of this factor is uncertain. Loss to follow-up is another possible explanation [36], as is competing mortality from common diseases, such as tuberculosis and malaria.
Our study is also limited by the small sample size thus we were unable to compare prevalent versus incident cancers and we lacked sufficient power to detect small differences that may exist between the general population and the HIV positive population. Another limitation is lack of unique individual personal identification numbers, such as social security numbers, which would increase the specificity and accuracy of the match. Nevertheless, thorough clerical review of the linked records increased our confidence in the matched records in study. Additional research is needed to understand and improve the specificity of matches in resource-limited settings, where unique individual identification numbers are generally lacking. As the HIV positive population in Nigeria continues to age, coverage with HAART improves and the registry systems are strengthened, more record linkage studies to examine the growing public health impact of cancers occurring in people living with HIV/AIDS will be warranted.

Study population
Abuja where this study was conducted is Nigeria's Federal Capital Territory, it had a population of 1.4 million in 2006 [37]. Because of its role as the capital of Nigeria, individuals from all the country's ethnic groups, tribes and religions live there. Muslims make up approximately 50% of the population, Christians 40%, while the remainder adhere to indigenous beliefs. Abuja has five districts and several surrounding towns and villages. The estimated HIV prevalence in Abuja FCT was 8.6% in 2010, the fifth highest in Nigeria [32].

HIV registry
The study population is a cohort of 18

Cancer registry
Abuja Cancer Registry (ABCR) is a population based cancer registry established in 2009. It covers the population of Abuja where majority of its clients (>97%) permanently reside. National Hospital Abuja (NHA), which is located in one of Abuja's main districts, and University of Abuja Teaching Hospital (UATH), which is located in the main suburban district of Abuja are the main sources of data for the ABCR. Other ABCR data sources include public and private hospitals, and free-standing pathology laboratories in Abuja. Cancer registrars visit the data sources once a month to perform medical chart abstraction and identify new cases of cancer diagnosed during the month. For each new case, hospital number, first and last names, age and date birth, sex, tribe, address, date of diagnosis, tumor site and histology are collected.

Registry linkage
The registry linkage was performed with a probabilistic matching algorithm, using the software for automated linkage in Italy (SALI) software. SALI is a software designed for a linkage analysis of small registries, with few expected linked cases [38]. Files with first name, last name, sex, date of birth and HIV or cancer registration numbers were provided by each registry and used for the linkage analysis. The names were edited by deleting titles, prefixes, suffixes and by correcting common variations in spelling in order to improve match sensitivity. The registry linkage was conducted in collaboration with the developers of SALI. Linkage was done by researchers who were independent of the HIV and cancer registries. To ascertain the validity of the linkage analysis, an extensive review of the medical records was conducted, to confirm the identifiers and diagnosis on all matched cases.

Statistical analysis
Descriptive analyses were performed to characterize the demographic attributes of the clients in the IHVN and ABCR registries. Cancer incidence rates among HIV positive persons, were estimated by dividing the number of cancers by person-years of follow-up. In addition, we categorized cancers in persons with HIV or AIDS occurring 60 months before to 3 months after registration at IHVN's HIV treatment and prevention sites as prevalent; cancers occurring 4 to 27 months after registration as early-incident; and cancers occurring 28 months after registration as late-incident cancers [12]. Standardized incidence ratios (SIRs), a ratio of the number of cancers observed among subjects, to the number of cancers expected, were estimated and used to evaluate the association of various cancers with HIV infection in the incident period. The number of expected cases for each cancer type was calculated as a sum of the product of person-time of HIV positive clients and the corresponding sex, age and period specific cancer rates from the ABCR. The source of the population count was Nigeria's National Population Council (NPC) data, which is stratified by age, sex and geographic location [37]. As in the Ugandan record-linkage study, we considered cancers to be associated with HIV infection if SIRs were significantly increased in the incident period [12]. Statistical analyses were performed with SAS for UNIX statistical software (version 9.2; SAS Institute).

Consent
Written informed consent was obtained from the patient for the publication of this report and any accompanying images.

Ethics
The study was conducted according to the Nigerian National Code for Health Research Ethics and the Declaration of Helsinki. Ethical approval to conduct this study was obtained from the Institute of Human Virology Health Research Ethics Committee.