Prevalence and Molecular Epidemiology of Human T-Lymphotropic Virus Type 1 among Women Attending Antenatal Clinics in Sub-Saharan Africa: A Systematic Review and Meta-Analysis

Abstract.

Human T-cell lymphotropic virus type 1 (HTLV-1) imposes a substantial disease burden in sub-Saharan Africa (SSA), which is arguably the world’s largest endemic area for HTLV-1. Evidence that mother-to-child transmission persists as a major mode of transmission in SSA prompted us to estimate the pooled prevalence of HTLV-1 among pregnant women throughout the region. We systematically reviewed databases including EMBASE, MEDLINE, Web of Science, and the Cochrane Database of Systemic Reviews from their inception to November 2018. We selected studies with data on HTLV-1 prevalence among pregnant women in SSA. A random effect meta-analysis was conducted on all eligible data and heterogeneity was assessed through subgroup analyses. A total of 18 studies, covering 14,079 pregnant women, were selected. The evidence base was high to moderate in quality. The pooled prevalence, per 100 women, of the 18 studies that screened HTLV-1 was 1.67 (95% CI: 1.00–2.50), a figure that masks regional variations. In Western, Central, Southern, and Eastern Africa, the numbers were 2.34 (1.68–3.09), 2.00 (0.75–3.79), 0.30 (0.10–0.57), and 0.00 (0.00–0.21), respectively. The prevalence of HTLV-1 infection among pregnant women in SSA, especially in Western and Central Africa, strengthens the case for action to implement routine screening of pregnant women for HTLV-1. Rigorous studies using confirmatory testing and molecular analysis would characterize more accurately the prevalence of this infection, consolidate the evidence base, and further guide beneficial interventions.

INTRODUCTION

The global burden of human T-lymphotropic virus type 1 (HTLV-1) remains inadequately characterized. Human T-cell lymphotropic virus type 1 now infects 5 to 10 million people worldwide,¹ causing adult T-cell leukemia (ATL) and HTLV-1–associated myelopathy/tropical spastic paraparesis (HAM/TSP) in up to 5% of infected individuals, leading to significant morbidity from HAM/TSP and mortality from ATL.² Other confirmed links include inflammatory disorders such as infective dermatitis and uveitis.² In addition, a recent study found that HTLV-1 was associated with atherosclerosis among the elderly in an area with high seroprevalence.³ Although most of those infected are lifelong asymptomatic carriers, the individual and community impact of HTLV-1–associated diseases can be devastating, yet there is neither a preventive vaccine nor any definitive treatment. The prognosis of ATL and HAM/TSP is dismal, in terms of both survival and quality of life. HTLV-1–associated myelopathy/tropical spastic paraparesis, as a chronic, progressive disease, in which 50% become wheelchair dependant, imposes extraordinary social and financial burdens. Likewise, beyond the lives it might claim, HTLV-1 infection can exact a heavy toll on health systems. Comorbid opportunistic infections often develop in ATL patients as a consequence of immunosuppression caused by dysfunctional HTLV-1–infected T cells. Parasitic infections, especially strongyloidiasis, and fungal infections are frequently associated with all forms of ATL.^4,5 Thus, interventions such as screening, counseling, and educating high-risk individuals and populations are morally compelling and have the potential to decrease the spread of this burdensome, often deadly, blood-borne virus.

Regions of high endemicity include sub-Saharan Africa (SSA), South America, Caribbean Islands, Aboriginal Australia, and Japan.^1,4 HTLV-1 prevalence varies greatly across regions and populations. Sub-Saharan Africa is considered to be highly endemic.¹ As a blood-borne pathogen, HTLV-1 may propagate by transfusion or intravenous drug use, but most new HTLV-1 infections occur horizontally through sexual intercourse, mainly from male to female, and vertically from mother to child, especially after breastfeeding for more than 6 months.^6–8

Mother-to-child transmission (MTCT) occurs in 20% of offspring from an infected mother and is associated with the greatest risk of ATL. Mother-to-child transmission correlates with proviral load in breast milk, concordance of HLA class-I antigens between mother and child, high antibodies titers, and prolonged breastfeeding.^9–11 In fact, among children breastfed for a prolonged period, the rate of transmission ranged between 15% and 25%. Postnatal infection by breastfeeding seems to play the most important role in vertical transmission, and there is evidence that a reduction in the number of mothers breastfeeding and a shortening of the breastfeeding period decreased MTCT by about 80%.^12,13 Thus, because of inadequate treatment options for TSP or ATL and the lack of a vaccine, prevention should focus on interrupting parenteral, sexual, and vertical transmission of HTLV-1.

Human T-cell lymphotropic virus type 1 has at least three major genetic subtypes. The most common subtype is the Cosmopolitan (a), which has the widest distribution worldwide and includes five subgroups. Subgroup A (transcontinental) has spread worldwide, subgroup B (Japanese) is found in Japan and India, subgroup C (West African) may have originated from Africa and came to the Caribbean through slave trade, and subgroups D (North Africa) and E (Peruvian Black) are found in the populations for which they are named. Other HTLV-1 subtypes are mainly found in Central Africa (HTLV-1b),¹⁴ whereas the Melanesian subtype (HTLV-1c) is found in Australia and the islands of the southern Pacific.¹⁵

The number of women of childbearing age (15–49 years) in Africa is increasing and is one of the main factors driving the continent’s surge in births. Estimated to number 280 million in 2015, women of childbearing age are predicted to reach 407 million by 2030, then 607 million by 2050, and approximately 1 billion at the end of the century.¹⁶ This surge will represent enormous challenges for health-care delivery, including antenatal and postnatal care. Antenatal routine screening for HTLV-1 is not now a standard clinical practice in SSA. Based on its association with various severe diseases and clinical manifestations, this virus should receive more attention from public health systems. Also, because ATL is associated with asymptomatic infection over decades that include prime reproductive years, preventing MTCT is critical as an immediate intervention with long-term benefits. Thus, screening pregnant women for anti-HTLV antibodies and counseling seropositive mothers to avoid or limit breastfeeding could prevent many vertical transmission events.

With effective preventive measures available, not screening for HTLV-1 misses an opportunity to curb its spread and reduce future morbidity and mortality. However, widely held assumptions about endemicity have not been backed by a comprehensive investigation of HTLV-1 prevalence among pregnant women in SSA. Sound epidemiological data are needed to inform and motivate policies at local, national, regional, and global levels. Otherwise, HTLV-1 as a neglected tropical disease will continue to propagate in and beyond SSA. Therefore, we performed a systematic review and meta-analysis of available data concerning the prevalence and molecular epidemiology of HTLV-1 among pregnant women in SSA and its subregions.

MATERIALS AND METHODS

Search strategy.

Studies were selected from the following databases: MEDLINE, EMBASE (via Ovid), Web of Science, and the Cochrane Database of Systematic Reviews, covering studies published since the inception of each database until November 30, 2018. No restrictions on language or study design were imposed. Medical Subject Heading (MeSH) terms were used in PubMed, Emtree terms in EMBASE, and keyword search in other databases. This search was supplemented by a hand-search of relevant studies. A full description of the search strategy is presented in Table 1. The search terms were combined using “and” statements and searches were performed on article titles, abstracts, and subjects. Additional studies were identified through hand-searching the references of relevant studies and reviews. The database searches were updated regularly, with the last update on January 15, 2019. The search methodology and reported findings comply with the relevant sections of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. The title and abstracts of identified publications were screened independently by two reviewers (M.O.D. and P.M.B.), with any publication deemed potentially relevant by either reviewer carried forward to full-text evaluation. Disagreements during full-text review were resolved by consensus or, when necessary, by a third independent reviewer (A.M.N.).

Table 1.

Systematic review search strategy

	Search statement	Results
EMBASE (OvidSP) search strategy
#1	exp AFRICA, WESTERN/ or exp AFRICA, NORTHERN/ or exp SOUTH AFRICA/ or exp AFRICA, EASTERN/ or africa.mp. or exp “AFRICA SOUTH OF THE SAHARA”/ or exp AFRICA, CENTRAL/ or exp AFRICA/ or exp AFRICA, SOUTHERN/	321,062
#2	exp HTLV-I Infections/ or exp Human T-lymphotropic virus 2/ or exp Human T-lymphotropic virus 1/ or exp HTLV-II Infections/ or Human T-Cell Lymphotropic Viruses.mp. or exp Leukemia-Lymphoma, Adult T-Cell/ or blood borne viruses.mp. or exp Blood-Borne Pathogens/ or transfusion transmitted infection*.mp. or TTI$.mp.	16,365
#3	pregnancy/ or pregnancy.mp. or exp pregnancy/ or exp pregnant women/or prenatal care.mp. or exp prenatal care/ or antenatal care.mp.	860,457
#4	1 AND 2 AND 3	25

Search	Description	Items found
PubMed search strategy
#1	(pregnancy [Mesh] OR pregnant women [Mesh] OR prenatal care [Mesh] OR antenatal care [tiab])	851,726
#2	((((((HTLV-I Antigens [Mesh]) OR HTLV-II Antigens [Mesh]) OR HTLV-I Infections [Mesh]) OR HTLV-II Infections [Mesh]) OR HTLV-II Antibodies [Mesh]) OR HTLV-I Antibodies [Mesh]) OR transfusion reaction [Mesh] OR transfusion transmitted infection* [tiab] OR TTI* [tiab] OR Human T-Cell Lymphotropic Viruse* [tiab] OR Blood-Borne Pathogens [Mesh]	25,874
#3	(((((((((Africa, Central [Mesh]) OR Africa, Eastern [Mesh]) OR Africa, Northern [Mesh]) OR Africa, Southern [Mesh]) OR Africa, Western [Mesh]) OR Africa, Southern [Mesh]) OR Africa, Western [Mesh]) OR Africa South of the Sahara [Mesh]) OR africa [Mesh]) OR afric* [tiab]	364,292
#4	#1 AND #2 AND #3	88
Web of Science search strategy
#1	TOPIC: (pregnant women) OR TOPIC: (pregnan*) OR TOPIC: (prenatal care) OR TOPIC: (pre-natal care) OR TOPIC: (antenatal care) OR TOPIC: (ante-natal care) OR TOPIC: (antenatal clinic*)	474,335
#2	TOPIC: (human T lymphotropic virus*) OR TOPIC: (human T lymphotropic virus type 1) OR TOPIC: (HTLV-1) OR TOPIC: (HTLV-I) OR TOPIC: (HTLV-1 antibodies) OR TOPIC: (HTLV-1 antigens) OR TOPIC: (sexually transmitted infections)	36,201
#3	TOPIC: (africa) OR TOPIC: (sub-Saharan Africa) OR TOPIC: (Western Africa) OR TOPIC: (Central Africa) OR TOPIC: (Eastern Africa) OR TOPIC: (afric*) [Mesh]) OR afric* [tiab]	522,963
#4	#1 AND #2 AND #3	786

Date of search: November 30, 2018. Search updated: January 15, 2019.

Study selection.

Articles and reports were included if they 1) estimated the prevalence of HTLV-1 among pregnant women in SSA and 2) used confirmatory tests, such as western blots (WB) or molecular assays such as polymerase chain reaction (PCR) in some cases. Case studies, reviews, commentaries, and reports on Africans residing outside Africa were excluded. We also excluded studies presented as abstracts or conference proceedings if their results did not contain sufficient information to adequately assess study quality.

Data extraction.

The selection of articles and the data extraction were independently conducted by two reviewers (P.M.B. and E.S.M.) using a standardized instrument that collected the following information: name of the first author, year of publication, setting, sample size, screening methods, and number of positive cases. Relevant information from each article was entered into a spreadsheet. Following the database search, duplicates were discarded. Titles and abstracts were then screened and irrelevant studies removed. Full-text articles of potentially relevant studies were evaluated by both authors according to eligibility criteria for inclusion. Disagreements between reviewers were resolved by consensus or by a third reviewer (A.M.N.). The study outcome of interest was the proportion of subjects with the presence of HTLV-1 antibody detected by an immunoassay with reported confirmatory testing.

Quality assessment.

Two authors (M.O.D. and P.M.B.) independently appraised the quality of selected articles, with disagreements adjudicated by a third reviewer (A.M.N.). Methodological quality of included studies was evaluated using the Standards for Reporting of Diagnostics accuracy for diagnostic studies and criteria specific to prevalence studies. Details are provided elsewhere.¹⁷ The tool contained seven criteria, three of which focused on external validity and four on internal validity. Quality for each criterion was assessed as either “plus” or “minus.” Studies were then classified as high (score of 7), moderate (score 4–6), or low (score < 4) quality.

Data synthesis and analysis.

For each study, the prevalence of HTLV-1 antibodies was calculated as the reported numbers of subjects screened positive for these markers, divided by the total number of subjects screened. Data were analyzed using the metaprop command of the meta package (4.9–2) in R (version 3.5.1, R Foundation for Statistical Computing, Vienna, Austria).¹⁸ Pooled seroprevalence was calculated with the DerSimonian–Laird random-effects model with Freeman–Tukey double arcsine transformation.¹⁹ We chose a random-effects model a priori because we anticipated heterogeneity arising from variation between studies conducted in various subregions. The transformed proportions were back-transformed and results were presented as percentages. Heterogeneity across the studies was assessed using the I-squared (I²) statistic, with high heterogeneity defined as when I² was equal to or greater than 75%.²⁰ Sources of heterogeneity were explored by comparing prevalence results between subgroups. A subgroup meta-analysis was performed among different regions (Western, Central, and Southern Africa). Small study effects (e.g., publication bias) were not assessed as this is not reliable for seroprevalence studies,²¹ but it is assumed to exist to some degree in all systematic reviews. A leave-one-out sensitivity analysis was performed by iteratively removing one study at a time while recalculating the prevalence to confirm that our findings were not driven by any single study and to assess the robustness of our findings. In a second sensitivity analysis, we compared different methods used in meta-analysis of proportions, including the logit transformation, arcsine, and Freeman–Tukey methods. The latter minimizes influence on the overall estimate arising from studies with extreme prevalence results, by stabilizing the variance of study-specific prevalence. In addition, a cumulative meta-analysis was performed to further evaluate the possible impact of publication year by entering the older studies at the top and adding the newer studies at the bottom.

RESULTS

Search results.

The flow diagram for our electronic search strategy and study selection is shown in Figure 1. The search identified 903 potentially relevant publications. After removing duplicates and screening titles, abstracts, and full texts, we included 18 publications that met our criteria and proceeded with systematic review and meta-analysis.^22–39

Figure 1.

Summary of data extraction history.

Study characteristics.

The study design and characteristics are shown in Table 2. All 18 studies were cross-sectional in design, spanned from 1989 to 2014, and included a total of 14,079 pregnant women. All regions of SSA were represented, including eight studies from Western Africa, six studies from Central Africa, three studies from Southern Africa, and two studies from Eastern Africa. One study by Goubeau et al. reported seroprevalence in two different countries: South Africa and Democratic Republic of the Congo, formerly Zaire. Screening tests were mainly based on enzyme immune assay, including the enzyme-linked immunosorbent assay. Other tests included chemiluminescence enzyme-linked immunoassay and particle agglutination. Confirmatory tests were mainly WB. Results of the quality assessment are depicted in Table 2. In brief, the vast majority of studies (90%) were deemed high or moderate quality.

Table 2.

General characteristics of included studies

First author (year)*	Country	Region	Study design	Study size	HTLV-1 cases	Proportion (%)	Screening test†	Confirmation test‡	Quality assessment score§
Goubau† (1993)	DR Congo	Central	Cross-sectional	414	19	4.6	ELISA	WB	B
Delaporte (1995)	DR Congo	Central	Cross-sectional	1,160	43	3.7	ELISA	WB	B
Etenna (2008)	Gabon	Central	Cross-sectional	907	19	2.1	ELISA	WB	A
Ndumbe (1992)	Cameroon	Central	Cross-sectional	170	1	0.6	ELISA and PA	WB	B
Tuppin (1996)	Congo	Central	Cross-sectional	2,070	14	0.7	EIA	WB	C
Berteau (1993)	Gabon	Central	Cross-sectional	633	35	5.5	ELISA	WB	A
Ramos (2011)	Ethiopia	East	Cross-sectional	165	0	0.0	EIA	NA	B
Ramos (2012)	Ethiopia	East	Cross-sectional	556	0	0.0	CMI	NA	C
Melo (2000)	Mozambique	South	Cross-sectional	132	1	0.8	ELISA	WB	B
Taylor (1996)	South-Africa	South	Cross-sectional	2,357	10	0.4	PA	WB	B
Goubau‡ (1993)	South-Africa	South	Cross-sectional	428	1	0.2	ELISA	WB	B
Andersson (1997)	Guinea-Bissau	West	Cross-sectional	1,231	27	2.2	ELISA	WB	B
Armah (2006)	Ghana	West	Cross-sectional	960	20	2.1	PA	WB	A
Okoye (2014)	Nigeria	West	Cross-sectional	200	1	0.5	ELISA	WB	A
Olaleye (1995)	Nigeria	West	Cross-sectional	364	20	5.5	EIA	WB	A
Verdier (1989)	Ivory Coast	West	Cross-sectional	513	10	1.9	ELISA	WB	A
Zehender (2008)	Guinea-Bissau	West	Cross-sectional	427	11	2.6	ELISA	PCR	B
Naucler (1992)	Guinea-Bissau	West	Cross-sectional	272	9	3.3	ELISA	WB	B
Biggar (1993)	Ghana	West	Cross-sectional	1,120	22	1.9	ELISA	WB	A

* The number of studies exceeds the 18 studies included in the flowchart because one study by Goubeau et al. has reported seroprevalence in two different countries: Democratic Republic of the Congo (formerly Zaire: Goubeau†) and South Africa (Goubeau‡).

† CMI = chemiluminescent microparticle immunoassay; EIA = enzyme immunoassays; ELISA = enzyme-linked immunosorbent assay; PA = particle agglutination test.

‡ NA = not applicable; PCR = polymerase chain reaction; WB = western blot.

§ A: 7, “high”; B: 4–6, “moderate”; C: < 4, “low.”

Prevalence of HTLV-1 among pregnant women.

In a cumulative sample size of 14,079 pregnant women, pooling of the 18 studies that screened only HTLV-1 yielded an overall prevalence of 1.67 (95% CI: 1.00–2.50) (Figure 2). The heterogeneity between studies was high (I² = 90.5% [95% CI: 86.6–93.2]), which we explored in a subgroup analysis.

Figure 2.

Forest plot of proportion estimates of HTLV-1 in pregnant women.

Subgroup analysis.

After stratifying by subregions, the pooled estimate was higher in Western (n = 8) and Central (n = 6) Africa than that in Southern (n = 3) and Eastern Africa (n = 2) (2.34 [95% CI: 1.68–3.09], 2.00 [95% CI: 0.75–3.79], 0.30 [95% CI: 0.10–0.57], and 0.00 [95% CI: 0.00–0.21], respectively) (Table 3).

Table 3.

Overall and stratified HTLV-1 prevalence

	N	Prevalence (95% CI)	I2 (%)
Overall	14,079	1.67 (1.00–2.50)	90
Subregion
Western Africa	5,087	2.34 (1.68–3.09)	59
Central Africa	5,354	2.00 (0.75–3.79)	93
Southern Africa	2,917	0.30 (0.10–0.57)	0
Eastern Africa	721	0.00 (0.00–0.21)	0

Sensitivity analysis and cumulative meta-analysis.

We conducted a series of sensitivity analyses to assess the robustness of the findings. First, we performed a leave-one-out sensitivity analysis by iteratively removing one study at a time and recalculating the pooled prevalence. The pooled prevalence remained stable, indicating that the overall results were robust and not dominated by any single study (Table 4). Second, we calculated the pooled prevalence using three different methods of transformation used in meta-analysis of proportions. We found that the overall pooled estimates using the logit transformation, arcsine, and Freeman–Tukey double arcsine were overlapping (1.85 [95% CI: 1.28–2.67], 1.61 [95% CI: 0.96–2.44], and 1.67 [95% CI: 1.00–2.50], respectively). The cumulative meta-analysis did not display any major change of the prevalence over time. Thus, time period did not appear to impact the prevalence (Figure 3).

Table 4.

Sensitivity analysis using the leave-one-out method

Study	Prevalence (95% CI)
All	1.67 (1.00–2.50)
Goubau† (1993)	1.55 (0.90–2.36)
Delaporte (1995)	1.57 (0.92–2.39)
Etenna (2018)	1.54 (0.85–2.40)
Ndumbe (1992)	1.63 (0.94–2.48)
Tuppin (1996)	1.65 (0.93–2.55)
Berteau (1993)	1.52 (0.90–2.28)
Ramos (2011)	1.69 (1.00–2.55)
Ramos (2012)	1.75 (1.07–2.59)
Melo (2000)	1.61 (0.93–2.46)
Taylor (1996)	1.68 (0.99–2.53)
Goubau‡ (1993)	1.68 (0.99–2.55)
Andersson (1997)	1.64 (0.93–2.52)
Armah (2006)	1.65 (0.94–2.53)
Okoye (2014)	1.64 (0.95–2.49)
Olaleye (1995)	1.42 (0.80–2.18)
Verdier (1989)	1.55 (0.87–2.41)
Zehender (2008)	1.52 (0.85–2.36)
Naucler (1992)	1.60 (0.93–2.44)
Biggar (1993)	1.65 (0.94–2.54)

Goubeau et al. has reported seroprevalence in two different countries: Democratic Republic of the Congo (formerly Zaire: Goubeau†) and South Africa (Goubeau‡).

Figure 3.

Cumulative meta-analysis by publication year.

Genotypes of HTLV-1 among pregnant women.

Two included studies investigated molecular characteristics of the HTLV-1 strains by analyzing the env and long terminal repeat regions of genotype isolates from confirmed positive samples. One study conducted in Gabon (Central Africa) found that all but one of the HTLV-1 strains belonged to molecular subtype b, the most prevalent subtype in Central Africa (HTLV-1b).²⁶ In the second study conducted in Guinea-Bissau (Western Africa), analysis showed that the cosmopolitan subtype (HTLV-1a) was prevalent and only one isolate belonged to the b subtype (HTLV-1b).³⁶

DISCUSSION

Building on the pioneering efforts of other investigators, we report what appears to be the largest and most comprehensive analysis of HTLV-1 prevalence in pregnant women in SSA. In this systematic review and meta-analysis, using a random-effects model, we found a pooled prevalence of 1.67 (95% CI: 1.00–2.50) per 100 pregnant women, with differences across subregions. The pooled prevalence in Western and Central Africa was higher than that in Southern and Eastern Africa, for reasons yet to be elucidated. The zero prevalence reported from Eastern Africa (Ethiopia) may represent an absence of proof rather than proof of absence, and serves as a reminder that more data, especially from the latter two subregions, would be helpful.

Compared with blood donors in SSA, the prevalence in pregnant women was nearly 3-fold higher.⁴⁰ Why? The prevalence of HTLV-1 infection increases with age, particularly among women. Early evidence suggested that women were more vulnerable to HTLV-1 infection because the infection may be transmitted more efficiently from males to females than vice versa.^41,42 Moreover, blood donors are predominantly younger and male, and regardless of age or gender are screened to identify and exclude high-risk populations from blood donation. However, compared with the general population in SSA, the prevalence of HTLV-1 among pregnant women lies in-between. Indeed, prevalence in the general population has been reported from 1.1% in Guinea to 6.6% to 8.5% in Gabon.⁴³ Thus, pregnant women may to some extent reflect the general population. However, comparisons with the general population are complicated because HTLV-1 infection increases with age, predominantly among women.^1,4,44 As anticipated, our estimates for SSA were higher than those of other regions. In Western Europe and some Asian countries, HTLV-1 prevalence is very low, for example, from 0.007/100 in Germany to 0.034/100 in England, and 0.1/100 in Japan.^45,46 This is in-line with the fact that SSA is known as a region of high HTLV-1 endemicity. The pooled prevalence in this review is comparable to that of Martinique (1.9% [95% CI: 0.68–3.18]).⁴⁷ This is consistent with the known endemicity of HTLV-1 in in both regions, SSA and the Caribbean.

The heterogeneity of the overall pooled estimates from our study was high and we observed a significant variation among subregions. Sample size also varies substantially across studies. The observed heterogeneity may also be attributed to the residual effect of unmeasured variables. Another possibility is that the observed heterogeneity reflects the true variability of HTLV-1 epidemiology within countries and subregions. Nonetheless, we adopted a random-effects model a priori to account for anticipated heterogeneity between studies. Reassuringly, a series of sensitivity analyses showed that our estimate was robust.

Phylogenetic analysis has revealed two main genotypes of HTLV-1: HTLV-1 subtype b (HTLV-1b), most prevalent in Central Africa, and the cosmopolitan subtype (HTLV-1a). Analysis of HTLV-1 genotypes is important in understanding the natural history, transmission chains, and evolution of HTLV-1 infection. Although we found few studies that assessed HTLV-1 genotypes, their findings are consistent with reports from Bissau, Senegal, and Mozambique.^48–50

In aggregate, data from the articles on which our meta-analysis is based support the idea of screening pregnant women for HTLV-1 in SSA. Evidence implicates breastfeeding as the most important mode of vertical transmission; elsewhere, seropositive mothers have been counseled to refrain from breastfeeding.^12,51 In well-resourced Japan, which has led the world with interventions against HTLV-1, prenatal screening for anti-HTLV-1 antibodies and counseling of seropositive mothers to avoid breastfeeding have dramatically reduced vertical transmission.¹² Therefore, preventing MTCT would probably have the most significant impact on the occurrence of HTLV-1–associated diseases. In addition, antenatal screening and the promotion of bottle-feeding for children of seropositive mothers have been cost-effective in Japan.⁵² In light of such measures to minimize MTCT of HTLV-1, questions persist about the most appropriate and cost-effective strategies for resource-limited settings such as SSA. Would the benefits of avoiding HTLV-1 infection outweigh the risks of not breastfeeding? Breast milk is not only nourishing but also transfers maternal immunity to protect infants from infectious agents. In fact, breastfeeding for 6–11 months is thought to be the most effective strategy to prevent child mortality in resource-limited settings, and the WHO recommends at least 6 months of exclusive breastfeeding.⁵³ Therefore, potential consequences of interrupted breastfeeding include malnutrition and increased infant mortality.⁵⁴ Where safe alternatives to breastfeeding are not readily available, limiting breastfeeding to the first 6 months may be a viable compromise to convey nutrition and maternal antibodies when most needed, while reducing the likelihood of MTCT. Clearly, an integrated approach to the prevention, detection, and control of HTLV-1 is needed in SSA. These measures should include the systematic screening and counseling of pregnant women, seeking safer alternatives to prolonged breastfeeding for seropositive mothers, and the prevention of sexual transmission by educational programs emphasizing the importance of safe sex. However, low incomes, difficulties around implementation of screening programs, and inadequate access to affordable health care in SSA are impediments to these goals. Also, the costs of systematic screening with current technology can be prohibitive where scarce financial resources may already be earmarked for the prevention of other infectious diseases such as HIV and HBV. Cooperation among governments, stakeholders, and the technology sector has the potential to increase access to testing, decrease its cost, and ultimately decrease the spread and transmission of HTLV in SSA.

Strengths of this review include a robust methodology, starting with a wide and deep literature search using clearly defined terms, followed by stringent inclusion and exclusion criteria, and the use of robust statistical methods, including the Freeman–Tukey double arcsine transformation and sensitivity analysis. The literature review was not arbitrarily time-limited or restricted to any particular language(s). Particular care was taken to include only studies that used confirmatory tests for positive screening results, as prevalence estimates can otherwise be skewed by false positives. Thus, our review summarizes the best available epidemiological information for the prevalence of HTLV-1 in pregnant women in SSA. Our data should be interpreted in light of a number of limitations. First, there were a limited number of acceptable studies. Subregions of SSA were variably represented and the most studies came from West Africa, which point to important research needs particularly for Eastern and Southern Africa and limit the scope for generalizations and comparisons. Thus, differences between subregions need to be interpreted with caution. Second, prevalence data were based on sampling from participants not necessarily representing national prevalence rates. Third, despite being an important variable associated with HTLV-1, age was not reported consistently across studies. Hence, we could not adjust for age in the meta-regression and this may account for some of the residual heterogeneity observed. However, because the mean age of pregnant women is generally comparable (about 22 to 26 years) in most countries, it is unlikely that age could have a major impact. Fourth, the paucity of data on HTLV-1 genotypes precludes any firm conclusions about it. Finally, this study is sensitive to all limitations associated with systematic reviews, such as publication bias. Nevertheless, we believe that the broad search strategy used in this review and the fact that we did not impose any restrictions on time or language yielded a wide range of articles, and it is unlikely that we missed large and major studies.

In conclusion, this review has sought to assemble and analyze conscientiously acquired data throughout SSA to guide thinking and action for subsequent generations. Having established that HTLV-1 is prevalent among pregnant women in SSA, our findings suggest the need for a wider approach to its prevention, detection, and control. The evidence provided might serve as a basis to decide, implement, and sustain public health policies aimed at reducing the vertical transmission of HTLV-1. Global strategies ultimately require active involvement of local governments, other stakeholders, and the technology sector. In addition, more rigorous studies with confirmatory testing and molecular analysis will expand the evidence base, from which further analysis and action can improve health in SSA.