Surveillance for sickle cell disease, United Republic of Tanzania

Abstract Objective To determine the regional- and district-level newborn prevalence of sickle cell trait and disease, and the prevalence of haemoglobin variants and genetic modifiers of sickle cell disease, in the nine regions of north-western United Republic of Tanzania. Methods We repurposed dried blood spot samples from children (aged 0–24 months) born to mothers living with human immunodeficiency virus (HIV), collected as part of the HIV Early Infant Diagnosis programme, for sickle cell diagnosis. We performed isoelectric focusing to determine whether samples had normal haemoglobin, sickle cell trait, sickle cell disease or a rare haemoglobin variant. We shipped samples diagnosed as disease or variant to Cincinnati Children’s Hospital in the United States of America for deoxyribonucleic-acid-based analyses to determine the prevalence of α-thalassaemia, glucose-6-phosphate dehydrogenase (G6PD) deficiency or fetal haemoglobin genetic modifiers. Findings We analysed a total of 17 200 specimens during February 2017–May 2018. We observed a prevalence of sickle cell trait and disease of 20.3% (3492/17 200) and 1.2% (210/17 200), respectively. District-level trait varied from 8.6% (5/58) to 28.1% (77/274). Among confirmed sickle cell disease specimens, we noted 42.7% (61/143) had 1-gene deletion and 14.7% (21/143) had 2-gene deletion α-thalassaemia trait. We documented G6PD A– deficiency in 19.2% (14/73) of males. Conclusion Our calculated prevalence is twice as high as previously reported and reinforces the need for enhanced sickle cell diagnostic services. Our district-level data will inform public health policy, allowing screening and disease-modifying hydroxyurea therapy to be focused on high-prevalence areas, until universal newborn screening is available.


Introduction
Sickle cell disease is an inherited disorder of haemoglobin, caused by a mutation in the β-globin subunit of adult haemoglobin. In classic autosomal recessive fashion, inheritance of one abnormal and one normal allele confers sickle cell trait, a carrier state without clinical symptoms. Inheritance of two mutated alleles causes sickle cell disease, characterized by varying amounts of chronic haemolytic anaemia, recurrent debilitating pain and an array of clinical sequelae, including increased risk of infection, stroke, lung disease, splenic dysfunction and bone infarction. 1 Sickle cell disease imposes a significant global burden of disease that remains underrecognized, 2 especially in Africa. Approximately 400 000 infants are born each year with sickle cell disease; 3-5 75% of these infants are born in the tropical regions of sub-Saharan Africa, 6 home to most of the > 25 million people who live with sickle cell disease globally. 7 Sickle cell disease causes substantial morbidity 8 and is responsible for 5-16% of mortality in children younger than 5 years. 9,10 Cumulative data from prior studies suggest that more than half of the children with sickle cell disease in sub-Saharan Africa die in early childhood, 11 with substantial differences in mortality between historic rural communities 12 and modern urban centres. 13 Early enrolment in a comprehensive care programme that includes preventive care (immunizations and prophylactic antimicrobials) and disease-modifying therapy (hydroxyurea or prophylactic blood transfusions) can reduce symptoms and improve survival.
The World Health Organization acknowledged the global importance of addressing sickle cell disease almost 15 years ago, 10,14,15 and in 2010 African leaders formally proposed sickle cell disease prevention and control strategies for the African Region. 5 In response, the United Republic of Tanzania has embedded sickle cell disease targets within the national noncommunicable disease policy, 16 increased advocacy, 17 created a centre of excellence, 18 educated health-care workers and increased research output. 19 The majority of these efforts have been focused in Muhimbili National Hospital in the coastal city of Dar es Salaam, but prevalence estimates suggest that the greatest burden of sickle cell disease is in the north-western regions of the country around Lake Victoria. 6,20,21 The United Republic of Tanzania does not yet have a national newborn screening programme. In the absence of recent reports, Tanzanian estimates of sickle cell disease are based on sparse data from studies performed over the past 65 years in only seven of the 30 regions of the country. [22][23][24][25][26][27][28][29][30][31][32] Recognizing that data extracted from isolated reports can poorly reflect variation within a country, 33,34 we initiated and conducted the United Republic of Tanzania Sickle Surveillance Study in the north-western regions of the country. Our primary objective was to provide contemporary regional-and district-level data on the newborn prevalence of sickle cell trait and disease to inform the national noncommunicable disease policy goals.
Objective To determine the regional-and district-level newborn prevalence of sickle cell trait and disease, and the prevalence of haemoglobin variants and genetic modifiers of sickle cell disease, in the nine regions of north-western United Republic of Tanzania. Methods We repurposed dried blood spot samples from children (aged 0-24 months) born to mothers living with human immunodeficiency virus (HIV), collected as part of the HIV Early Infant Diagnosis programme, for sickle cell diagnosis. We performed isoelectric focusing to determine whether samples had normal haemoglobin, sickle cell trait, sickle cell disease or a rare haemoglobin variant. We shipped samples diagnosed as disease or variant to Cincinnati Children's Hospital in the United States of America for deoxyribonucleic-acid-based analyses to determine the prevalence of α-thalassaemia, glucose-6-phosphate dehydrogenase (G6PD) deficiency or fetal haemoglobin genetic modifiers.  36 An infant born to a mother living with HIV is brought to a Reproductive and Child Health clinic at 4-6 weeks of age for the first preventive health visit. A dried blood spot is collected from the infant, and transported to a reference laboratory for detection of HIV using a polymerase chain reaction (PCR). The dried blood spots are labelled with key demographic information including date of birth, sex, referring health facility, date of collection, date of dispatch and date of receipt at the laboratory. HIV results are communicated to the referring health facility for prompt initiation of antiretroviral medication. Dried blood spots are then stored at room temperature and made available for additional testing for sickle cell trait and disease.

Sickle cell diagnosis
We analysed all repurposed dried blood spots collected as part of the HIV Early Infant Diagnosis programme during February 2017-May 2018 by using the isoelectric focusing technique. 34 Our laboratory equipment included Resolve Hemoglobin kits and JB-2 Staining System reagents (both PerkinElmer, Inc., Waltham, United States of America, USA), as well as control specimens and other consumables, all donated to the haematology section of the Bugando laboratory. Staff were trained on-site by a board-certified haematologist, and attended a 2-day seminar in Dar es Salaam, organized by the equipment manufacturer and funded jointly by the manufacturer and the United States Association of Public Health Laboratories. At the seminar, staff acquired theoretical knowledge of the isoelectric focusing technique and had the opportunity to process samples while supervised by an experienced manufacturer representative. The senior Tanzanian paediatrician leading the haematology clinic also completed a further two months of clinical and laboratory training funded and hosted by Cincinnati Children's Hospital, USA.
All dried blood spots were analysed with a standard control specimen containing adult, fetal and sickling haemoglobin (haemoglobin A, F and S) and haemoglobin C. Isoelectric focusing results were scored independently by two Bugando Medical Centre staff for the presence and abundance of each type of haemoglobin. The results were interpreted as: normal if haemoglobin A (± haemoglobin F) was present; sickle cell disease if haemoglobin S (± haemoglobin F) was present; sickle cell trait if both haemoglobin A and S (± haemoglobin F) were present; variant if a band was present at any location other than haemoglobin A, S or C (± haemoglobin F); and uninterpretable if poor quality precluded interpretation. Dried blood spot analyses interpreted as sickle cell disease, variant or uninterpretable were repeated for confirmation and frozen for later deoxyribonucleic acid (DNA) studies. Regular teleconferences were convened with collaborators based in the Cincinnati Children's Hospital, USA, to provide ongoing feedback on the quality of laboratory techniques and interpretation of gels; however, the Tanzanian team was responsible for all final interpretations.

DNA-based testing
Any dried blood spot diagnosed as sickle cell disease or variant was stored at −20 °C then shipped to the USA for future DNA testing. Upon arrival, we stored dried blood spots at −80 °C until testing could be performed. We extracted DNA from dried blood spots using an adapted protocol from Instagene (Bio-Rad Laboratories, Hercules, USA). 37 We performed amplification of β-globin gene exons 1 and 2 using a PCR, and confirmed the presence of a haemoglobin S mutation at rs334 (c.334T > A;p.Glu6Val) using a custom TaqMan PCR probe (Applied Biosystems, Foster City, USA). 37 We analysed specimens interpreted as uncommon and atypical haemoglobin variants by isoelectric focusing via the previously published algorithm 37 used to investigate uncommon variants in East Africa. Uncommon variants include the α-chain variants haemoglobin G-Pest (HBA1: p .Asp75Asn) and haemoglobin Stanleyville II (HBA1: p .Asn79Lys), and the fusion variants haemoglobin P-Nilotic (β-globin gene (HBB)-δ-globin gene (HBD) fusion: β31-δ50) and haemoglobin Kenya (γ-globin gene (HBG1)-HBB fusion: A γ81-β86). 37 We detected α-thalassaemia trait resulting from the 3.7-kilobase α-globin gene deletion and glucose-6-phosphate dehydrogenase (G6PD) Avariant using DNA-based techniques, [37][38][39] and determined the modifiers of baseline haemoglobin F production. 40 We genotyped the BCL11A polymorphisms (rs1427407, rs7557939 and rs11886868) and HBS1L-MYB intergenic polymorphisms (HMIP) (rs28384513 and rs9399137) using commercially available real-time PCR assays (Applied Biosystems, Foster City, USA). To identify the XmnI single-nucleotide polymorphism at −158 basepairs to G-γ globin (rs7482144), we amplified the G-γ gene (HBG2) using PCR with G-γ-specific forward and reverse primers to ensure that the G-γ rather than an A-γ product was isolated. We then further amplified this product by performing PCR with Classic 1 forward and reverse primers. We genotyped the final product using a custom-made TaqMan PCR probe set (Applied Biosystems, Foster City, USA).

Data analysis
We entered data into an Excel database (Microsoft, Redmond, USA), treating age as a continuous variable (summarized using median and interquartile range) and treating haemoglobin type, HIV status, sex, region of origin, district of origin, G6PD status and α-thalassaemia status as categorical variables (summarized using frequencies). We calculated the prevalence of sickle cell trait and disease by dividing the number of specimens with sickle cell trait and disease by the total number of non-missing specimens with interpretable results. We determined allelic frequency by dividing the number of times that an allele was observed (once in heterozygotes, twice in homozygotes) by the total number of all alleles (twice the total number of specimens). We compared the proportions of participants living with HIV with and without sickle cell disease using a χ 2 test.

Ethical considerations
Our study protocol was approved with a waiver for informed consent by the joint Catholic University of Health and Allied Sciences-Bugando Medical Centre Research and Ethics Committee, as well as the Tanzanian National Institute for Medical Research, to perform disease surveillance on de-identified archived dried blood spots previously collected by the HIV Early Infant Diagnosis programme. The study was also approved by the Cincinnati Children's Hospital Medical Center Institutional Review Board. A formal material transfer agreement was obtained so that specimens could be shipped to the USA for genetic analysis.

Sample collection and testing
Our local Bugando Medical Centre staff completed a total of 232 isoelectric focusing gels during February 2017-May 2018. After samples from children older than 24 months were excluded to obtain a more accurate prevalence for infants, the median age of infants included in the sampling population was 52 days (interquartile range, 41-93 days). Staff scored a total of 17 274 unique dried blood spot specimens from the catchment area. The quality of laboratory testing was extremely high with only 20 specimens scored as uninterpretable and 54 with missing results, meaning that we performed our primary analysis on 17 200 specimens.

Prevalence
We observed an overall prevalence of sickle cell trait and disease in our cohort of 20.3% (3492/17 200) and 1.2% (210/17 200), respectively, with a 0.1% (17/17 200) prevalence of atypical or uncommon haemoglobin variants. Our data yielded an allelic frequency of 0.114 ([3492+(2 × 210)]/(2 × 17 200)) for the sickle gene (haemoglobin S) mutation, and demonstrate perfect Hardy-Weinberg equilibrium, as expected in a stable population. 41 We did not identify any haemoglobin C or other common β-globin variants. Our geospatial mapping revealed mild variation between regions, with the prevalence of sickle cell trait and disease ranging from 16.6% (146/880) to 22.5% (253/1126) and 0.5% (4/880) to 1.5% (17/1126), respectively ( Table 1). Analysis of individual districts that provided more than 48 specimens (i.e. excluding Buhigwe, from which only 12 samples were provided) revealed a wider geographic variability, with sickle cell trait and disease ranging from 8.6% (5/58) to 28.1% (77/274) and from zero to 4.3% (9/208), respectively ( Fig. 1 and Table 1). Using our regional prevalence estimates of sickle cell trait and disease, and regional census data from the 2012 Population and Housing Census for the United Republic of Tanzania, 36 we calculated the estimated number of annual births with sickle cell trait or disease in each region within the study area ( Table 2). We estimated that the number of births with sickle cell disease per year was 10 056 within the nine regions comprising the study area. Contributions varied by region according to their population size and sickle cell disease prevalence. We projected the lowest number of births with sickle cell disease per year for the region of Singida (526 births), and the highest for the region of Mwanza (1730 births; Table 2).

Relation to HIV status
For the 16 479 samples for which HIV results were available, we analysed the co-morbidity of HIV and sickle cell disease to compare the potential effect of HIV status on mortality, as previously performed in Uganda. 34 The prevalence of sickle cell disease was 1.2% among both HIV-infected (9/732) and HIV-negative (192/15 747) infants, indicating that HIV status has no effect on early mortality (Table 3).
We provide the minor allelic frequencies for genetic modifiers that affect haemoglobin F production for the 143 samples available for genetic testing in Table 5. Three single-nucleotide polymorphisms in BCL11A had minor allelic frequencies ranging from 0.266 (rs1427407) to 0.325 (rs4671393). Our frequencies were slightly higher than those observed within the African subgroup of the 1000 Genomes Project, 42 which reported values of 0.238 (rs1427407) and 0.269 (rs4671393). The frequency of two single-nucleotide polymorphisms within HMIP was more variable; the value at rs28384513 (in the HMIP-1 region) was 0.238, which is slightly higher than the 1000 Genomes Project African subgroup (0.184), while the value at rs9399137 (in the HMIP-2 region) was low (0.045) and almost identical to the 1000 Genomes Project African subgroup (0.042). Finally, the A variant of rs7482144 in HBG2 was not detected, as expected in an East African cohort.

Discussion
We estimate that the annual number of live births with sickle cell disease is at least twice that previously thought to occur in the United Republic of Tanzania; we project that over 10 000 births among just 40% of the Tanzanian population are affected annually, compared with previous estimates of 8655 6 20 Although we observed geographical differences between districts, our variations were not as high as those reported from neighbouring Uganda 34 where specimens were collected from a roughly equivalent land mass and population. Our high prevalence and relative lack of variation between districts in the north-western part of the country may reflect the high selection pressure from malaria in a holoendemic area, and possibly lower migration rates. Our data reinforce the urgent need to enhance sickle cell diagnostic services, the obligatory first step in the cascade of care for this neglected patient population. The sickle cell clinic at Bugando Medical Centre has approximately 600 patients currently enrolled in care, some of whom travel from neighbouring regions. However, we estimate that 1730 infants are born with sickle cell disease in Mwanza region alone each year, which indicates that many affected children in sub-Saharan Africa have not even been diagnosed with the condition. The nine regions included in this study have the highest childhood mortality in the country, ranging from 30 to 38 deaths per 1000 live births, 43 and sickle cell disease contributes significantly to mortality in this age group. [11][12][13] We also investigated the extent to which additional erythrocyte disorders are commonly co-inherited with sickle cell disease; cross-sectional analyses in the United Republic of Tanzania have reported on the impact of such disorders on sickle cell patients. 44 Ongoing prospective cohort studies will help to clarify the unique interactions between specific sickle cell disease morbidities and co-inherited erythrocyte disorders, endemic arboviral and plasmodial infections, and environmental factors. Expression of haemoglobin F is another well-recognized factor affecting disease severity. Some modifiers of haemoglobin F are similar between Tanzanian and British patients, 45 but genetic studies among Tanzanians have also identified unique candidate variants and pathways. 46 Understanding these genetic factors and their influence on disease outcomes will become more important, especially as hydroxyurea therapy is introduced to the region. 47 An important strength of our study is that we developed a sickle cell disease diagnostic service using existing infrastructure used for HIV infant diagnosis, demonstrating a potential of using this approach across sub-Saharan Africa. For example, in Uganda decision-makers have successfully expanded the use of their HIV infant diagnosis screening platform for sickle cell disease diagnosis, and have launched focused screening in 18 high-prevalence areas. 48 Until the emergence of government-sponsored universal screening, local providers can implement their own strategies for optimal screening. Regardless of the location (e.g. hospitals, schools, or reproductive and child health clinics), population (e.g. all patients, mothers, newborns or children) or testing modality (isoelectric focusing, electrophoresis, point-of-care testing or high-performance liquid chromatography) used for initial screen-  ing, the HIV infant diagnosis infrastructure can be used to transport dried blood spots to a central laboratory equipped for confirmatory testing and diagnosis of sickle cell disease. Expanding testing in this way would represent a crucial step towards achieving some of the targets of the sustainable development goal 3, which are to end preventable deaths among children and reduce premature mortality from noncommunicable diseases. 5,49 The testing of infants was another strength of our study. By excluding samples from children older than 24 months from our analyses, we avoided the errors of previous reports 22-32 that used adults and children to develop inaccurately low estimates, or that extrapolated sparse results over large geographic areas.

Region or district
Our study had two limitations. First, we analysed dried blood spots only after they had been processed by the HIV Early Infant Diagnosis laboratory. In some cases, several months had elapsed before the dried blood spots became available for sickle cell disease testing. During that time, dried blood spots were stored at suboptimal conditions for haemoglobin preservation (usually room temperature); however, our isoelectric focusing testing was robust and able to score 99.6% (17 200/17 274) of all dried blood spot specimens. Second, our study population was restricted to children born to mothers living with HIV; however, it has been reported that HIV status is unlikely to affect the genetic inheritance of the sickle allele. 34 Since universal newborn screening is not immediately feasible in most countries, we recommend initially focusing screening efforts on high-prevalence districts to invest in those communities most affected by sickle cell disease, using existing public health infrastructure with minimal start-up cost and training. The cost-effectiveness of implementing a screening programme in conjunction with treatment for children who are subsequently diagnosed with sickle cell disease varies between regions and countries, especially regarding salary for personnel, but a detailed analysis in Angola provides some helpful comparisons. 50 Inaccurate national and subnational estimates obscure the true burden of sickle cell disease, which is a common and important cause of death in young children in low-income countries. It is   α-thalassaemia (n = 143) 5 copies, αα/ααα 1-gene duplication, unaffected 0 (0.0) 4 copies, αα/αα 0-gene deletion, unaffected 61 (42.7) 3 copies, αα/-α 3. 7 1-gene deletion, α-thalassaemia minima 61 (42.7) 2 copies, -α 3.7 /-α 3   Вывод Рассчитанная авторами распространенность заболевания вдвое превышает опубликованные ранее значения и еще раз указывает на необходимость укрепления и расширения услуг по диагностике СКА. Полученные данные по районам станут основой для выработки политики в области общественного здравоохранения, которая бы позволяла сосредоточить скрининговое обследование и лечение гидроксимочевиной (изменяющее течение заболевания) в областях с высоким распространением заболевания до тех пор, пока не станет возможным всеобщее скрининговое обследование новорожденных.

Vigilancia de la drepanocitosis en la República Unida de Tanzania
Objetivo Determinar la prevalencia de la enfermedad y del rasgo de las células falciformes en los recién nacidos a nivel regional y de distrito, y la prevalencia de las variantes de hemoglobina y de los modificadores genéticos de la drepanocitosis en las nueve regiones del noroeste de la República Unida de Tanzania.

Research Tanzania Sickle Surveillance Study
Emmanuela E Ambrose et al.