Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Epidemiology of tuberculous lymphadenitis in Africa: A systematic review and meta-analysis

  • Daniel Mekonnen ,

    Contributed equally to this work with: Daniel Mekonnen, Abraham Aseffa

    Roles Formal analysis, Methodology, Writing – original draft, Writing – review & editing

    nigusdaniel@gmail.com

    Affiliations Department of Medical Microbiology, Immunology and Parasitology, College of Medicine and Health Sciences, Bahir Dar University, Bahir Dar, Ethiopia, Biotechnology Research Institute, Bahir Dar University, Bahir Dar, Ethiopia

  • Awoke Derbie,

    Roles Data curation, Methodology, Writing – review & editing

    Affiliations Department of Medical Microbiology, Immunology and Parasitology, College of Medicine and Health Sciences, Bahir Dar University, Bahir Dar, Ethiopia, The Centre for Innovative Drug Development and Therapeutic Trials for Africa (CDT-Africa), Addis Ababa University, Addis Ababa, Ethiopia

  • Andargachew Abeje,

    Roles Formal analysis, Software

    Affiliation Geospatial Data and Technology Center, Bahir Dar University, Bahir Dar, Ethiopia

  • Abebe Shumet,

    Roles Data curation, Methodology, Writing – review & editing

    Affiliation Amhara Regional State Health Bureau, Felege Hiwot Referral Hospital, Bahir Dar, Ethiopia

  • Endalkachew Nibret,

    Roles Data curation, Methodology, Writing – review & editing

    Affiliations Biotechnology Research Institute, Bahir Dar University, Bahir Dar, Ethiopia, Department of Biology, Bahir Dar University, Bahir Dar, Ethiopia

  • Fantahun Biadglegne,

    Roles Data curation, Formal analysis, Methodology, Validation, Writing – review & editing

    Affiliation Department of Medical Microbiology, Immunology and Parasitology, College of Medicine and Health Sciences, Bahir Dar University, Bahir Dar, Ethiopia

  • Abaineh Munshae,

    Roles Data curation, Formal analysis, Methodology, Writing – review & editing

    Affiliations Biotechnology Research Institute, Bahir Dar University, Bahir Dar, Ethiopia, Department of Biology, Bahir Dar University, Bahir Dar, Ethiopia

  • Kidist Bobosha,

    Roles Writing – original draft, Writing – review & editing

    Affiliation Armauer Hansen Research Institute, Addis Ababa, Ethiopia

  • Liya Wassie,

    Roles Methodology, Writing – original draft, Writing – review & editing

    Affiliation Armauer Hansen Research Institute, Addis Ababa, Ethiopia

  • Stefan Berg,

    Roles Validation, Writing – review & editing

    Affiliation Animal and Plant Health Agency, Weybridge, the United Kingdom

  • Abraham Aseffa

    Contributed equally to this work with: Daniel Mekonnen, Abraham Aseffa

    Roles Conceptualization, Investigation, Supervision, Validation, Writing – review & editing

    Affiliation Armauer Hansen Research Institute, Addis Ababa, Ethiopia

Abstract

Introduction

Tuberculous lymphadenitis is the most frequent form of extra-pulmonary TB (EPTB) and accounts for a considerable proportion of all EPTB cases. We conducted a systematic review of articles that described the epidemiological features of TBLN in Africa.

Methods

Any article that characterized TBLN cases with respect to demographic, exposure and clinical features were included. Article search was restricted to African countries and those published in English language irrespective of publication year. The articles were retrieved from the electronic database of PubMed, Scopus, Cochrane library and Lens.org. Random effect pooled prevalence with 95% CI was computed based on Dersimonian and Laird method. To stabilize the variance, Freeman-Tukey double arcsine root transformation was done. The data were analyzed using Stata 14.

Results

Of the total 833 articles retrieved, twenty-eight articles from 12 African countries fulfilled the eligibility criteria. A total of 6746 TBLN cases were identified. The majority of the cases, 4762 (70.6%) were from Ethiopia. Over 77% and 88% of identified TBLN were cervical in type and naïve to TB drugs. Among the total number of TBLN cases, 53% were female, 68% were in the age range of 15–44 years, 52% had a history of livestock exposure, 46% had a history of consuming raw milk/meat and 24% had history of BCG vaccination. The proportion of TBLN/HIV co-infection was much lower in Ethiopia (21%) than in other African countries (73%) and the overall African estimate (52%). Fever was recorded in 45%, night sweating in 55%, weight loss in 62% and cough for longer than two weeks in 32% of the TBLN cases.

Conclusions

TBLN was more common in females than in males. The high prevalence of TBLN in Ethiopia did not show directional correlation with HIV. Population based prospective studies are warranted to better define the risk factors of TBLN in Africa.

Introduction

Tuberculosis (TB) is one of the oldest chronic and complex infectious diseases and is caused by a group of bacteria belonging to the Mycobacterium tuberculosis complex (MTBC). The complex includes the human adapted species of M. tuberculosis and M. africanum, and zoonotic pathogens; M. bovis, M. caprae, M. microti and M. pinnipedii which affect cattle, goats/sheep, voles and seals/lions, respectively [1, 2]. The current body of evidence suggests that these mycobacteria might have co-evolved along with early hominids in East Africa since as far back as 3 million years ago [3, 4].

The 2017 WHO global TB report on Africa showed high TB mortality (41/100,000), incidence (254/100,000) and TB/HIV co-infection (34%) rates. Moreover, the treatment success rate was below the WHO target of ≥85% [5]. In 2016, as many as 82% and 85% of TB deaths were reported among HIV-negative and total TB patients, respectively, in Africa and WHO South East Asia Region [57]. Altogether, Africa is the worst affected region with TB.

On average in the world, pulmonary TB (PTB) accounted for 85% of the clinical forms of TB whereas extra-pulmonary TB (EPTB) accounted for the remaining 15% [5, 8]. The most common types of EPTB include TB of the lymphatics (TBLN), pleural, bone, meningeal, genitourinary and peritoneal TB [911]. However, the prevalence of EPTB and its predominant forms varies from country to country [10, 1215]. For instance, Ethiopia reports an EPTB proportion of 32%; ranking third in case number globally next to India and Pakistan despite their much larger total population size [5] and this level has remained high over the years [10, 16, 17].

Global TB control efforts have largely ignored EPTB. This is because EPTB is generally considered non-infectious and as such inconsequential to the global epidemic [18]. However, recent data from northwest England have shown that the prevalence of active TB disease among household contacts of EPTB was high (440 per 100 000 contacts screened), indicating that EPTB cases might have substantial impact on TB transmission [19]. Moreover, it is conceivable that the slower annual decline rate of EPTB compared to PTB [11] could retard the progress towards the END-TB targets set by WHO [20, 21].

Among the risk factors studied for EPTB, immunological drivers are well known and have been comprehensively reviewed by O’Garra and colleagues [22]. Reports from USA and South Africa showed that race, sex and HIV are important risk factors for the development of EPTB [15, 23]. Another study from Brazil also described an association of EPTB with HIV and ethnicity [24]. On the other hand, Mehta et al. (1991) compared TB data between pre and post HIV era in the USA and concluded that EPTB was not associated with HIV there [25]. Berg et al. (2015) did not find any association of ethnicity, TB strain type or HIV co-infection with TBLN prevalence in Ethiopia [26]. Additionally, the role of other potential factors such as over diagnosis [27], bovine origin [28] or lineage tropism [29] were minimal and unable to explain the high incidence rate of TBLN in Ethiopia.

Several attempts have been made to investigate differences in infectivity and virulence of strains isolated from PTB and EPTB patients. Viedma et al. (2005) used an ex vivo competitive macrophage co-infection assay and a murine aerosol-infection model and reported that strains isolated from EPTB were more efficient and showed higher infectivity than strains derived from PTB sites. This report suggests a possible role for bacterial factors in determining the clinical phenotypes [30]. On the other hand, a study by Gomes et al. (2013) failed to detect any association between clinical phenotypes of TB and MTBC genotypes. Rather, they found a link between clinical phenotypes of TB and host factors [24].

Tuberculosis may be considered as a disease with a continuous and dynamic spectrum [31]; TBLN as one pole of the spectrum occurring with a relatively strong host resistance, and disseminated TB as the other pole with relatively weaker host resistance. TB lymphadenitis is distinct from disseminated TB where lymph nodes could be involved in addition to pulmonary illness. TBLN is a form of TB with no evidence of pulmonary involvement or TB illness in any other organ of the body. The same strain types have been isolated from TBLN and PTB cases in Ethiopia, suggesting that host and/or environmental factors might play a role in the pathogenesis of TBLN rather than strain tropism [29].

In general, pooled data on the epidemiology of TBLN in Africa were not available. Moreover, the factors behind the development of TBLN are not well understood and this might require a combined analysis of data from host (genomics, immunity, co-morbidity), environment and pathogen genomics and be triangulated using powerful statistical and mathematical tools [1].

Objectives

The central thesis of this review was to determine the geo-spatial distribution of TB lymphadenitis in Africa and to characterize TBLN cases by different demographic (gender, age groups), exposure (previous TB treatment history, raw meat/ milk exposure, BCG vaccination) and clinical variables (HIV co-infection, fever, weight loss, night sweat, cough).

Methods

Protocol registration

This review protocol is registered at the National Institute for Health Research; PROSPERO international prospective register of systematic reviews with registration number CRD42018104170 at (https://www.crd.york.ac.uk/PROSPERO/#recordDetails).

Eligibility criteria

Any article that characterized TBLN in African countries with respect to: gender, age, TBLN/HIV co-infection, lymph node features, exposure status (livestock, raw milk/meat, BCG vaccination and TB treatment) and cardinal TB symptoms was included. Peer review articles published in the English language irrespective of publication year were included. Tuberculous lymphadenitis cases diagnosed on clinical criteria plus cytology and/or bacteriology were included. Those, TBLN cases diagnosed on clinical criteria alone were excluded. Sample size was not used as inclusion or exclusion criterion.

Information sources and search strategy

Articles were retrieved from the electronic data bases of PubMed, Scopus, Cochrane library, and Lens.org. The search was done using key words and MeSH term. The key words included tuberculosis lymphadenitis, tuberculous lymphadenitis, lymph node tuberculosis, and Africa. The full search was done by combining key words and related MeSH terms using Boolean operators. S1 Table shows the full search strategy.

Study selection

All of the identified articles were imported to an Endnote library. Initial screenings were done by title followed by abstract and then full text reading. Articles were assessed independently for the fulfillment of the inclusion criteria by two authors (AD, AS). Disagreements regarding the inclusion or exclusion of articles were resolved by discussion.

Data collection process and data items

Data from the selected articles were extracted by two authors (DM, AM) independently using excel data extraction sheet. Key indicators such as first author, year of publication, study period, country, number and types of TBLN cases, lymph node features, sex, age and TBLN/HIV co-infection status were extracted. Moreover, history of exposure to raw milk/meat, BCG vaccination, contact with chronic cougher, previous TB treatment history were also extracted. Furthermore, cardinal TB symptoms (fever, night sweat, weight loss, and cough for longer than two weeks) were extracted. Distribution of patients’ place of origin and number of TBLN cases were mapped using ArcGIS 10.3 (ArcGIS Desktop, ESRI 2011. Redlands, Canada).

Risk of bias in individual studies

To assess risk of bias, two authors of this paper (DM, EN) independently used the seven item-based ROBINS-I risk of bias assessment tool [32]. Each item scored one point and discrepancies were resolved by a third independent author (FB). Moreover, to determine the certainty of evidence generated and strength of recommendations; Grading of Recommendations Assessment, Development and Evaluation (GRADE) tool was applied [33].

Summary measures and synthesis of results

The collected data were analyzed using quantitative measures. For random effect meta-analysis, approximate likelihood approach was followed. Moreover, to make the normal distribution assumptions more applicable to significance testing and stabilize the variances; Freeman-Tukey double arcsine rooted transformation was done [34]. Furthermore, to estimate the transformed pooled prevalence, Dersimonian and Laird method was used [35]. Taken together, using the metan command in Stata, study estimate (ES) as prevalence was computed using Freeman-Tukey double arcsine root transformation with 95% confidence interval. In the forest plot, the box indicated weight of articles from random effect analysis. The crossed line is the 95% confidence interval (CI), the solid vertical line is zero to x-axis. The analysis was done using Stata 14 (Stata Corp. College Station, TX, US). Country/sub-region/ wise sub-group analysis was done with regard to TBLN types, TBLN features, gender, age and TBLN/HIV co-infection status. The summary measures were presented as forest plots and table.

Risk of bias across studies

Statistical heterogeneity estimate among the articles estimate was assessed using Cochrane Q, I2 statistic and P-value. The I2 value of <25%, 25–50% and ≥ 50% was taken as low, moderate and high degree of heterogeneity, respectively [36]. To deal with heterogeneity, sub-group and sensitivity analyses were performed; possible publication bias was assessed using funnel plot asymmetry.

Results

Study selection

A total of 831 articles were retrieved from the four electronic databases and imported to an Endnote library. Two additional articles were identified through hand searching in the Ethiopian Journal of Health Development and African Journals Online. After removing duplicates (97 articles), 736 articles were screened. Of these, 632 articles did not fulfill the inclusion criteria and they were removed. A further 55 articles were excluded for the same reason after reading the abstract. Twenty-eight articles were included in the quantitative analysis. Over all, full screening was done based on the preferred reporting items for systematic reviews and meta- analysis (PRISMA) flow diagram (Fig 1).

thumbnail
Fig 1. PRISMA flow diagram of literature selection, Africa, 2018.

https://doi.org/10.1371/journal.pone.0215647.g001

Study characteristics

A total of 6746 TBLN cases from 12 African countries were reviewed. Majority of the cases, 4762 (70.6%) were from Ethiopia; Djibouti reported only eight confirmed TBLN cases. The geographic distribution of TBLN cases is summarized in Fig 2A and 2B. The spatial data used for the maps were taken from Map library which is a public domain that can be accessed at www.maplibrary.org.

thumbnail
Fig 2. Geographic distribution and number of TBLN cases in Africa and Ethiopia, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g002

Of the total of 28 articles reviewed, 14 articles were from Ethiopia [16, 26, 3748] two articles each from Zambia [49, 50], South Africa [51, 52] and Nigeria [53, 54]. One article each were identified from Burkina Faso [55], Uganda [56], Djibouti [57], Mozambique [58], Sudan [59], Tunisia [60], Tanzania [61], and Malawi [62]. Data collection period of articles was between 1970 and 2015 (46 years) while the publication years range was between 1975 and 2018 (44 years) (Table 1).

thumbnail
Table 1. Reviewed articles and characteristics of TBLN cases, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.t001

Most articles contained complete and clear data about sex, age, TBLN/HIV co-infection status and types of TBLN (Table 1). However, some articles lacked complete information about livestock exposure, history of consumption of raw milk/meat/ and history of BCG vaccination. Different articles categorized age differently. Thus, best educated guess was applied to assign data to the respective age ranges. Groups with unknown HIV status were removed in the meta- analysis. Meta-analysis was done when at least two articles had the variables of interest.

Risk of bias within studies

The risk of bias for each individual article was measured as no risk of bias, probably yes, yes and no information. Probably yes, yes and no information scored zero and no risk of bias got a score of one. The total score therefore ranged from zero to seven, with higher scores indicating higher quality of outcome. Of the total 28 articles reviewed; 17, 10 and one article showed an overall low, moderate and critical risk of bias, respectively (Table 2). Further, S2 Table shows that patient classification, measurement of outcome and reporting bias were the identified source of bias in the included articles. Overall, the included articles were judged as good quality.

thumbnail
Table 2. Robbins-I risk of bias summary for the included articles, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.t002

Results of individual studies

In this review, TBLN was disaggregated against 15 variables. The result of individual studies and its summary measure is presented using forest plot (Figs 37).

thumbnail
Fig 3. Pooled prevalence of TBLN types and features, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g003

thumbnail
Fig 4. Proportion of female and age group of 15–44 years among TBLN cases, Africa 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g004

thumbnail
Fig 5. Proportion of TBLN/HIV co-infection among TBLN cases, Africa 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g005

thumbnail
Fig 6. Prevalence of key exposure variables among TBLN cases, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g006

thumbnail
Fig 7. Prevalence of cardinal TB symptoms among TBLN cases, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.g007

Cervical TBLN is the most prevalent form and ranged from 47% [45] to 98% [50]. The sub-group analysis showed that cervical LN type was lower in Ethiopia (69%) compared with articles from northern and eastern African countries (89%) and southern and western sub-regions (85%) (Fig 3A). On the other hand, Bem (1997) reported low prevalence of matted type TBLN,17% [50], while other studies reported higher than 50% prevalence of the matted types [16, 38, 41, 51, 61] (Fig 3B).

Three sub-group analyses were done on proportion of females among TBLN subjects; Ethiopia in one group, southern and western African articles in the second group and that of northern and other eastern African countries in a third group. Pooled prevalence of female gender among north and eastern African studies, excluding Ethiopia, reported the lowest prevalence (42%) followed by southern and western African countries (47%). In Ethiopia it was 54%. Individual article estimates of female proportion ranged from 38% to 75%. The pooled African female proportion was 53% (95%CI: 51–55%) showing more females with TBLN than males (Fig 4A). When sensitivity analysis was performed, Muluye et al [45] showed an influence on Ethiopian and African overall estimates. After removing the Muluye et al article, the female proportion turned out to be 52% in Ethiopia and 50% for overall Africa. We noted that Muluye et al [45] analyzed a large number of TBLN cases (Table 1) and the study quality was rated as good (Table 2). Part A of S1 Fig depicts the sensitivity analyses of articles included in gender wise meta- analysis.

Most of TBLN patients were in the age range of 15–44 years. Articles reported as low as 12% [26], 39% [42] and as high as 98% [39] prevalence of age range of 15–44 years among TBLN cases (Fig 4B). The sensitivity analysis in Part B of S1 Fig shows the influence of Berg et al. [26] and Muluye et al. [45] studies. Before removing these two studies, the pooled prevalence of age range of 15–44 years was 67%, 70% and 68% in Ethiopia, in other Africa countries and also in the overall African pooled estimate, respectively. However, when these two influential articles [29,45] were removed, the pooled prevalence turned out to be 72%, 70% and 72% in Ethiopia, in other African countries and in overall Africa, respectively.

The prevalence of HIV among TBLN cases showed a clear difference between Ethiopia and other African countries. The majority of cases in Africa other than Ethiopia showed higher prevalence of HIV among TBLN cases (Fig 5). The sensitivity analysis for TBLN/HIV co-infection showed minimal influence by a single study. Part C of S1 Fig depicts the sensitivity analysis.

The majority of TBLN cases were new which ranged from 67% [58] to 94% [16]). The individual and pooled prevalence of various exposure status are summarized in Fig 6. Moreover, Fig 7 shows that the prevalence of the cardinal TB symptoms lies between 32% and 62%) among TBLN cases.

Synthesis of results

Seventy seven percent of TBLN cases included in this review were cervical lymph node (Fig 3A). Closer inspection of Fig 3B and Table 3 shows that matted and mobile type TBLN were more frequent than discrete and firm types, respectively. Moreover, TBLN was more frequent in female (53%) than in male patients (Fig 4A) and also more frequent in the age range of 15–44 years (68%) than in other age ranges (Fig 4B). The most surprising difference was the HIV prevalence among TBLN patients in Ethiopia, which differed significantly from the average in other African countries; 21% versus 73% (Fig 5).

thumbnail
Table 3. A summary of pooled prevalence to key variables among tuberculous lymphadenitis patients, Africa, 1970–2015.

https://doi.org/10.1371/journal.pone.0215647.t003

The prevalence of exposure variables such as history of anti-TB drugs, history of contact with TB patients, BCG vaccination history, history of drinking raw milk and eating raw meat was generally lower than 50% and did not show any trend (Table 3 and Fig 6). Likewise, prevalence of cardinal TB symptoms among TBLN cases also varied. For example, the prevalence of weight loss and night sweating was 62% and 55%, respectively (Table 3 and Fig 7).

Risk of bias across studies.

Except for female gender, the meta-analysis results were very heterogeneous and therefore, a random effect meta-analysis was done. The random effect analysis was also heterogeneous. To sort out the cause, publication bias was assessed. The funnel plot figures in S2 Fig shows the presence of possible publication bias. This bias might be due to missing of grey literatures across the continent and exclusion of non-English language articles. The other causes of heterogeneity might be due to differences in the recruitment criteria of TBLN cases and measurement of outcome variables. Sensitivity analysis was done for gender, age and TBLN/HIV co-infection for which over ten articles had been included. While omission of a single study at a time had minimal influence on the pooled prevalence of TBLN/HIV co-infection, it showed an influence on pooled prevalence of age groups [29, 45] and gender [45]. We noted that, the two influential articles have good methodological and outcome quality. Thus, it is less likely to influence the result away from the true pooled estimate. The GRADE pro system of grading the quality of evidence showed low quality of evidence. This is due to the methodological quality of the articles included in the review.

Discussion

Tuberculous lymphadenitis (scrofula) has been recognized for thousands of years and remains one of the most common forms of EPTB [63]. Cervical TBLN is the most frequent form followed by axillary and inguinal TBLN. In the middle ages in Europe, it was believed that a touch from royalty could heal this disease [64]. Unlike PTB which is more common in males [5], our review identified a relatively higher percentage of females (53%) than males among TBLN cases (low quality of evidence) (Table 3 and Fig 4A). The link between being female and TBLN is not well known. However, reports have shown that differences in tumor necrosis factor, interleukin-10, CD4+ lymphocyte counts, endocrine, socioeconomic and cultural factors [64] might influence the development of TBLN. A review of 31 articles from Afghanistan, Pakistan, India and Bangladesh agrees with our report [18]. Katsnelson (2017) discussed pregnancy, diabetes, vitamin D deficiency and low protein consumption as potential factors associated with TBLN.

The pooled prevalence for the age group of 15–44 years was higher than for other age groups among TBLN cases (low quality of evidence). TBLN was previously considered a disease of childhood [65]. Recent reports also showed that high TBLN cases occur in the age range of 20 to 40 years [9, 64]. A critical review by Biadglegn et al. (2013) showed that EPTB (with TBLN being the most common presentation) was more common among young adults [17].

Sub-group analysis of TBLN/HIV co-infection by country/sub-region/ showed that the pooled prevalence of HIV among TBLN in African countries other than Ethiopia was 73% whereas it was 21% in Ethiopia (low quality of evidence). This indicates that Ethiopia’s high TBLN rate is probably unique in its epidemiology and seems to lack directional correlation with HIV. Multiple studies showed that HIV infection was significantly associated more with allopatric than sympatric host-pathogen relationships [6668]. Absence of directional correlation between TBLN and HIV in Ethiopia might be a consequence of co-evolution. Moreover, evidence shows that EPTB is associated with HIV when it is the disseminated form rather than when it is exclusively localized [9]. The TBLN cases reviewed here are exclusively TBLN cases having no apparent pulmonary involvement. When considering Africa, the epidemiology of HIV among TBLN cases appears to be in line with other parts of the world in which HIV is the main driver of EPTB, including TBLN [69].

The majority (88%) of TBLN cases were newly identified cases showing no association between TBLN and TB treatment history. The history of eating raw meat/drinking raw milk among TBLN cases in Africa was 46% (low quality of evidence). In the past, it has been reported that 10–20% of all TBLN in Europe was caused by M. bovis, which was acquired from drinking unpasteurized milk [70]. However, recent studies from countries with similar settings (endemic bovine TB in cattle and no pasteurization) [71] have not shown such high prevalence of zoonotic TB. For instance, molecular analysis of 173 isolates from pastoral communities who had contact with livestock revealed as many as 160 M.tuberculosis and three M. bovis. Similarly, molecular analysis of 39 isolates from their camels, cattle and goats showed 24 M. bovis and 1 M. tuberculosis [72]. These data confirmed the low incidence of M. bovis as a cause for human TB. On the contrary, a systematic review of global epidemiology of TB due to M. bovis showed a higher rate (2.8%) among humans in Africa [73]. Such high prevalence of TB due to M. bovis is possibly because Müller et al. (2013) included articles reporting M. bovis using biochemical methods as diagnostic tool. It is known that biochemical method lacks specificity for identification of M. bovis. Taken together M. bovis was rarely detected in human TB [29, 74] but should not be ruled out as a zoonotic disease.

The prevalence of BCG vaccination history among TBLN cases in this review was 24%. However, the number of articles was a few (only 4 articles). Thus, the quality of the evidence is low; pending further investigation between BCG vaccination and TBLN. There are few reports about the effects of BCG on the incidence of TBLN except a report on its adverse effects among infants [75].

The prevalence of the cardinal TB symptoms among TBLN cases in Africa ranged from 32% with history of cough for longer than two weeks to 62% with record of weight loss. Overall, the prevalence of one or more systemic symptoms was 49%. Based on this report history of cough was less prevalent than weight loss. Unlike this study, a study from India showed that fever was the most prevalent symptom in TBLN [76]. Moreover, the prevalence of one or more systemic symptoms was 56.6% which is slightly higher than 49% in this report [76]. Another report from Turkey showed the prevalence of cough to be 26–33% which is in line with this report. However, night sweating of 29–36% which was reported by a study from Turkey [77] was lower than the present report (55%).

Limitations

Due to methodological exclusion of articles published in languages other than English and missing of grey literatures; publication bias is likely high. Although we included a large number of articles in the review, each article however contained only few variables. Thus, prevalence estimate was based on a small sample size which might make our pooled estimate imprecise. In addition, most of the included articles were chart reviews and retrospective in nature likely introducing clinical and methodological heterogeneity. These collectively reduce the quality of the generated evidence.

Conclusions

This review is the first comprehensive meta-analysis that estimated pooled prevalence for key demographic, exposure, and clinical variables that could characterize TBLN. Of the total 28 articles included in the review, 19 were from the Horn of Africa with most of these from Ethiopia (14 studies) suggesting more clustering of TBLN in Eastern Africa than in other sub regions. Within Ethiopia, TBLN was also relatively more clustered in agrarian than in pastoral regions.

Most TBLN (77%) were cervical in type, matted (67%) and mobile (64%) in their feature. The majority (68%) were in the age-range of 15–44 years. Unlike PTB which is more prevalent among males; TBLN is slightly higher among females (53%) and this requires further investigation. The TBLN/HIV co-infection rate was 52% in overall Africa, 21% in Ethiopia and 73% in the rest of Africa excluding Ethiopia which indicates the unique feature of TBLN epidemiology in Ethiopia. Eighty-eight percent of TBLN cases had no prior TB treatment history, 52% had livestock exposure and 24% had BCG vaccination scar. The prevalence of cardinal systemic symptoms among TBLN cases were 45%, 55%, 62% and 32% for fever, night sweating, weight loss, and history of cough for longer than two weeks, respectively.

To identify the most informative risk factors for TBLN, a meta-analysis and/or a prospective double population-based study is highly desirable. In addition, the host and pathogen genomic dimension and their evolutionary relationship should be investigated.

Supporting information

S2 Table. Full Robbins-I risk of bias assessment result.

https://doi.org/10.1371/journal.pone.0215647.s002

(XLSX)

S1 Fig. Sensitivity analysis of gender, age and TBLN/HIV co-infection.

https://doi.org/10.1371/journal.pone.0215647.s003

(DOCX)

S2 Fig. Funnel plots showing publication bias.

https://doi.org/10.1371/journal.pone.0215647.s004

(DOCX)

S1 File. Completed PRISMA check list of the review.

https://doi.org/10.1371/journal.pone.0215647.s005

(DOCX)

References

  1. 1. Comas I, Gagneux S. A role for systems epidemiology in tuberculosis research. Trends Microbiol. 2011;19(10):492–500 pmid:21831640
  2. 2. Smith NH, Gordon SV, de la Rua-Domenech R, Clifton-Hadley RS, Hewinson RG. Bottlenecks and broomsticks: the molecular evolution of Mycobacterium bovis. Nat Rev Microbiol. 2006;4(9):670 pmid:16912712
  3. 3. Kritski ACA, Reyes A, Martin A,Gicquel B, Martin C,Sola Ch,Espitia CI,. Tuberculosis.From basic science to patient care. Belgium, Brazil, Argentina: Institute of Tropical Medicine 2007.
  4. 4. Gutierrez MC, Brisse S, Brosch R, Fabre M, Omaïs B, Marmiesse M, et al. Ancient origin and gene mosaicism of the progenitor of Mycobacterium tuberculosis. PLoS Pathog. 2005;1(1):e5 pmid:16201017
  5. 5. WHO. Global tuberculosis report 2017. Geneva: World Health Organization 2017 978-92-4-156551-6 Contract No.: WHO/HTM/TB/2017.23.
  6. 6. Pai M, Behr MA, Dowdy D, Dheda K, Divangahi M, Boehme CC, et al. Tuberculosis. Nat Rev Dis Primers. 2016;2:16076 pmid:27784885
  7. 7. Nunn P, Williams B, Floyd K, Dye C, Elzinga G, Raviglione M. Tuberculosis control in the era of HIV. Nat Rev Immunol. 2005;5:819 pmid:16200083
  8. 8. Group BP. BMJ Best practices: Extrapulmonary Tuberculosis 2017: BMJ Publishing Group; 2017 [cited 2017 Nov 11]. https://bestpractice.bmj.com/topics/en-gb/166.
  9. 9. Qian X, Nguyen DT, Lyu J, Albers AE, Bi X, Graviss EA. Risk factors for extrapulmonary dissemination of tuberculosis and associated mortality during treatment for extrapulmonary tuberculosis. Emerg Microbes Infect.2018;7(1):102 pmid:29872046
  10. 10. Alemie GA, Gebreselassie F. Common types of tuberculosis and co-infection with HIV at private health institutions in Ethiopia: A cross sectional study. BMC Public Health. 2014;14(1) pmid:24708793
  11. 11. Peto HM, Pratt RH, Harrington TA, LoBue PA, Armstrong LR. Epidemiology of extrapulmonary tuberculosis in the United States, 1993–2006. Clin Infect Dis. 2009;49(9):1350–1357 pmid:19793000
  12. 12. Golden MP, Vikram HR. Extrapulmonary tuberculosis: an overview. Am Fam Physician. 2005;72(9)
  13. 13. Ramirez-Lapausa M, Menendez-Saldana A, Noguerado-Asensio A. Extrapulmonary tuberculosis: an overview. Rev Esp Sanid Penit. 2015;17:3–11
  14. 14. Lee JY. Diagnosis and treatment of extrapulmonary tuberculosis. Tuberc Respir Dis. 2015;78(2):47–55
  15. 15. Yang Z, Kong Y, Wilson F, Foxman B, Fowler AH, Marrs CF, et al. Identification of risk factors for extrapulmonary tuberculosis. Clin Infect Dis. 2004;38(2):199–205 pmid:14699451
  16. 16. Biadglegne F, Tesfaye W, Sack U, Rodloff AC. Tuberculous lymphadenitis in Northern Ethiopia: in a public health and microbiological perspectives. PLoS One. 2013;8(12):e81918 pmid:24349151
  17. 17. Biadglegne F, Tesfaye W, Anagaw B, Tessema B, Debebe T, Anagaw B, et al. Tuberculosis lymphadenitis in Ethiopia. Jpn J Infect Dis. 2013;66(4):263–268 pmid:23883834
  18. 18. Katsnelson A. Beyond the breath: Exploring sex differences in tuberculosis outside the lungs. Nat Med. 2017;23:398 pmid:28388608
  19. 19. Wingfield T, MacPherson P, Cleary P, Ormerod LP. High prevalence of TB disease in contacts of adults with extrapulmonary TB. Thorax. 2017:thoraxjnl-2017-210202
  20. 20. Lönnroth K, Raviglione M. The WHO’s new End TB Strategy in the post-2015 era of the Sustainable Development Goals. Trans R Soc Trop Med Hyg. 2016;110(3):148–150 pmid:26884490
  21. 21. Raviglione M, Director GT. Global strategy and targets for tuberculosis prevention, care and control after 2015. World Health Organization, Geneva. 2013
  22. 22. O’Garra A, Redford PS, McNab FW, Bloom CI, Wilkinson RJ, Berry MP. The immune response in tuberculosis. Annu Rev Immunol. 2013;31:475–527 pmid:23516984
  23. 23. Gounden S, Perumal R, Magula N. Extrapulmonary tuberculosis in the setting of HIV hyperendemicity at a tertiary hospital in Durban, South Africa. South Afr J Infect Dis. 2017:1–8
  24. 24. Gomes T, Vinhas SA, Reis-Santos B, Palaci M, Peres RL, Aguiar PP, et al. Extrapulmonary tuberculosis: Mycobacterium tuberculosis strains and host risk factors in a large urban setting in Brazil. PLoS One. 2013;8(10):e74517 pmid:24098337
  25. 25. Mehta JB, Dutt A, Harvill L, Mathews KM. Epidemiology of extrapulmonary tuberculosis: a comparative analysis with pre-AIDS era. Chest. 1991;99(5):1134–1138 pmid:2019168
  26. 26. Berg S, Schelling E, Hailu E, Firdessa R, Gumi B, Erenso G, et al. Investigation of the high rates of extrapulmonary tuberculosis in Ethiopia reveals no single driving factor and minimal evidence for zoonotic transmission of Mycobacterium bovis infection. BMC Infect Dis. 2015;15(1) pmid:25886866
  27. 27. Iwnetu R, van den Hombergh J, Woldeamanuel Y, Asfaw M, Gebrekirstos C, Negussie Y, et al. Is tuberculous lymphadenitis over-diagnosed in Ethiopia? Comparative performance of diagnostic tests for mycobacterial lymphadenitis in a high-burden country. Scand J Infect Dis. 2009;41(6–7):462–468 pmid:19382003
  28. 28. Ameni G, Vordermeier M, Firdessa R, Aseffa A, Hewinson G, Gordon SV, et al. Mycobacterium tuberculosis infection in grazing cattle in central Ethiopia. Vet J 2011;188(3–4):359–361 pmid:20965132
  29. 29. Firdessa R, Berg S, Hailu E, Schelling E, Gumi B, Erenso G, et al. Mycobacterial lineages causing pulmonary and extrapulmonary tuberculosis, Ethiopia. Emerg Infect Dis. 2013;19(3):460–463 pmid:23622814
  30. 30. García de Viedma D, Lorenzo G, Cardona P-J, Alonso Rodriguez N, Gordillo S, Ruiz Serrano MJ, et al. Association between the infectivity of Mycobacterium tuberculosis strains and their efficiency for extrarespiratory infection. J Infect Dis. 2005;192(12):2059–2065 pmid:16288368
  31. 31. Delogu G, Goletti D. The spectrum of tuberculosis infection: new perspectives in the era of biologics. J Rheumatol Suppl. 2014;91:11–16 pmid:24788995
  32. 32. Sterne JA, Hernán MA, Reeves BC, Savović J, Berkman ND, Viswanathan M, et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016;355:i4919 pmid:27733354
  33. 33. GRADEpro Guideline Development Tool [Software] [Internet]. McMaster University, (developed by Evidence Prime, Inc.). 2015.
  34. 34. Freeman MF, Tukey JW. Transformations related to the angular and the square root. Ann Math Stat. 1950:607–611
  35. 35. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–188 pmid:3802833
  36. 36. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557 pmid:12958120
  37. 37. Abdissa K, Tadesse M, Bezabih M, Bekele A, Apers L, Rigouts L, et al. Bacteriological methods as add on tests to fine-needle aspiration cytology in diagnosis of tuberculous lymphadenitis: can they reduce the diagnostic dilemma? BMC Infect Dis. 2014;14:720 pmid:25551280
  38. 38. Abebe G, Deribew A, Apers L, Abdissa A, Deribie F, Woldemichael K, et al. Tuberculosis lymphadenitis in Southwest Ethiopia: a community based cross-sectional study. BMC Public Health. 2012;12:504 pmid:22770435
  39. 39. Atomsa D, Abebe G, Sewunet T. Immunological Markers and Hematological Parameters among Newly Diagnosed Tuberculosis Patients at Jimma University Specialized Hospital. Ethiop J Health Sci. 2014;24(4):311 pmid:25489195
  40. 40. Beyene D, Ashenafi S, Yamuah L, Aseffa A, Wiker H, Engers H, et al. Diagnosis of tuberculous lymphadenitis in Ethiopia: correlation with culture, histology and HIV status. Int J Tuberc Lung Dis. 2008;12(9):1030–1036 pmid:18713500
  41. 41. Bezabih M, Mariam DW, Selassie SG. Fine needle aspiration cytology of suspected tuberculous lymphadenitis. Cytopathology. 2002;13(5):284–290 pmid:12421444
  42. 42. Bezabih M, Mariam DW. Determination of aetiology of superficial enlarged lymph nodes using fine needle aspiration cytology. East Afr Med J. 2003;80(11):559–563 pmid:15248672
  43. 43. Fanosie A, Gelaw B, Tessema B, Tesfay W, Admasu A, Yitayew G. Mycobacterium Tuberculosis complex and HIV co-infection among Extrapulmonary Tuberculosis suspected cases at the University of Gondar hospital, northwestern Ethiopia. PLoS One. 2016;11(3) pmid:26950547
  44. 44. Kidane D, Olobo JO, Habte A, Negesse Y, Aseffa A, Abate G, et al. Identification of the causative organism of tuberculous lymphadenitis in Ethiopia by PCR. J Clin Microbiol. 2002;40(11):4230–4234 pmid:12409403
  45. 45. Muluye D, Biadgo B, Gerima EW, Ambachew A. Prevalence of tuberculous lymphadenitis in Gondar University Hospital, Northwest Ethiopia. BMC Public Health. 2013;13(1) pmid:24499165
  46. 46. Tadesse M, Abebe G, Bekele A, Bezabih M, de Rijk P, Meehan CJ, et al. The predominance of Ethiopian specific Mycobacterium tuberculosis families and minimal contribution of Mycobacterium bovis in tuberculous lymphadenitis patients in Southwest Ethiopia. Infect Genet Evol. 2017;55:251–259 pmid:28919549
  47. 47. Yassin MA, Olobo JO, Kidane D, Negesse Y, Shimeles E, Tadesse A, et al. Diagnosis of tuberculous lymphadenitis in Butajira, rural Ethiopia. Scand J Infect Dis. 2003;35(4):240–243 pmid:12839151
  48. 48. Zewdie O, Mihret A, Abebe T, Kebede A, Desta K, Worku A, et al. Genotyping and molecular detection of multidrug-resistant Mycobacterium tuberculosis among tuberculosis lymphadenitis cases in Addis Ababa, Ethiopia. New Microbes New Infect. 2018;21:36–41 pmid:29675262
  49. 49. Bem C, Patil PS, Luo N. The Increased Burden of Tuberculous Lymphadenitis in Central Africa: Lymph Node Biopsies in Lusaka, Zambia, 1981 and 1990. Trop Doct. 1996;26(2):58–61 pmid:8685966
  50. 50. Bem C. Human immunodeficiency virus-positive tuberculous lymphadenitis in Central Africa: Clinical presentation of 157 cases. Int J Tuberc Lung Dis. 1997;1(3):215–219 pmid:9432366
  51. 51. Marais BJ, Wright CA, Schaaf HS, Gie RP, Hesseling AC, Enarson DA, et al. Tuberculous lymphadenitis as a cause of persistent cervical lymphadenopathy in children from a tuberculosis-endemic area. Pediatr Infect Dis J. 2006;25(2):142–146 pmid:16462291
  52. 52. Khuzwayo ZB, Naidu TK. Head and neck tuberculosis in KwaZulu-Natal, South Africa. J Laryngol Otol. 2014;128(1):86–90 pmid:24423085
  53. 53. Onuigbo WIB. Tuberculous peripheral lymphadenitis in the Igbos of Nigeria. Br J Surg. 1975;62(4):323–325 pmid:1131513
  54. 54. Ukekwe FI, Olusina D, Banjo A, Akinde O, Nzegwu M, Okafor O, et al. Tuberculous Lymphadenitis in South-Eastern Nigeria; a 15 Years Histopathologic Review (2000–2014). Ann Med Health Sci Res. 2016;6(1):44–49 pmid:27144076
  55. 55. Béogo R, Birba NE, Coulibaly TA, Traoré I, Ouoba K. Presentations of tuberculous adenitis of the head and neck at the University Hospital of Bobo-Dioulasso, Burkina Faso. Pan Afr Med J. 2013;15 pmid:24319521
  56. 56. Wamala D, Asiimwe B, Kigozi E, Mboowa G, Joloba M, Kallenius G. Clinico-pathological features of tuberculosis due to Mycobacterium tuberculosis Uganda genotype in patients with tuberculous lymphadenitis: A cross sectional study. BMC Clin Pathol. 2014;14(1) pmid:24690344
  57. 57. Blouin Y, Cazajous G, Dehan C, Soler C, Vong R, Hassan MO, et al. Progenitor "Mycobacterium canettii" clone responsible for lymph node tuberculosis epidemic, Djibouti. Emerg Infect Dis. 2014;20(1):21–28 pmid:24520560
  58. 58. Viegas SO, Ghebremichael S, Massawo L, Alberto M, Fernandes FC, Monteiro E, et al. Mycobacterium tuberculosis causing tuberculous lymphadenitis in Maputo, Mozambique Microbial genetics, genomics and proteomics. BMC Microbiol. 2015;15(1)
  59. 59. Ageep AK. Diagnosis of tuberculous lymphadenitis in Red Sea state, Sudan. Int J Trop Med.2012;7(1):53–56
  60. 60. Smaoui S, Mezghanni MA, Hammami B, Zalila N, Marouane C, Kammoun S, et al. Tuberculosis lymphadenitis in a southeastern region in Tunisia: Epidemiology, clinical features, diagnosis and treatment. Int J Mycobacteriol. 2015;4(3):196–201 pmid:27649866
  61. 61. Richter C, Kitinya JN, Kimara JS, Hirji KF. Clinical Characteristics of Tuberculous Lymphadenitis in Tanzania. Trop Doct. 1992;22(3):129–130 pmid:1641895
  62. 62. Boeree M, Kamenya A, Liomba G, Ngwira B, Subramanyam V, Harries A. Tuberculosis lymphadenitis, a diagnostic problem in areas of high prevalence of HIV and tuberculosis. Malawi Med J. 1998;11(2):56–59
  63. 63. Carrol E, Clark J, Cant A. Non-pulmonary tuberculosis. Paediatr Respir Rev. 2001;2(2):113–119 pmid:12531057
  64. 64. Cataño J, Robledo J. Tuberculous Lymphadenitis and Parotitis. Microbiol Spectr. 2016;4(6)
  65. 65. Fanning A. Tuberculosis: 6. Extrapulmonary disease. CMAJ. 1999;160(11):1597 pmid:10374005
  66. 66. Pasipanodya JG, Moonan PK, Vecino E, Miller TL, Fernandez M, Slocum P, et al. Allopatric tuberculosis host–pathogen relationships are associated with greater pulmonary impairment. Infect Genet Evol. 2013;16:433–440 pmid:23501297
  67. 67. Fenner L, Egger M, Bodmer T, Furrer H, Ballif M, Battegay M, et al. HIV infection disrupts the sympatric host–pathogen relationship in human tuberculosis. PLoS genet. 2013;9(3):e1003318 pmid:23505379
  68. 68. Gagneux S, DeRiemer K, Van T, Kato-Maeda M, De Jong BC, Narayanan S, et al. Variable host–pathogen compatibility in Mycobacterium tuberculosis. Proc Natl Acad Sci 2006;103(8):2869–2873 pmid:16477032
  69. 69. Powell DA, Hunt WG. Tuberculosis in children: an update. Adv Pediatr. 2006;53(1):279–322
  70. 70. Grange JM, Yates MD. Zoonotic aspects of Mycobacterium bovis infection. Vet Microbiol. 1994;40(1–2):137–151 pmid:8073621
  71. 71. Sibhat B, Asmare K, Demissie K, Ayelet G, Mamo G, Ameni G. Bovine tuberculosis in Ethiopia: A systematic review and meta-analysis. Prev Vet Med. 2017
  72. 72. Gumi B, Schelling E, Berg S, Firdessa R, Erenso G, Mekonnen W, et al. Zoonotic transmission of tuberculosis between pastoralists and their livestock in South-East Ethiopia. EcoHealth. 2012;9(2):139–149 pmid:22526748
  73. 73. Müller B, Dürr S, Alonso S, Hattendorf J, Laisse CJ, Parsons SD, et al. Zoonotic Mycobacterium bovis–induced tuberculosis in humans. Emerg Infect Dis. 2013;19(6):899 pmid:23735540
  74. 74. Carrol ED, Clark JE, Cant AJ. Non-pulmonary tuberculosis. Paediatr Respir Rev. 2001;2(2):113–119 pmid:12531057
  75. 75. Szczuka I. Adverse events following immunization with BCG vaccine in Poland 1994–2000. 2002.
  76. 76. Purohit MR, Mustafa T, Mørkve O, Sviland L. Gender differences in the clinical diagnosis of tuberculous lymphadenitis—a hospital-based study from Central India. Int J Infect Dis. 2009;13(5):600–605 pmid:19111495
  77. 77. Tatar D, Senol G, Alptekin S, Gunes E, Aydin M, Gunes O. Assessment of Extrapulmonary Tuberculosis in Two Provinces of Turkey. Iran J Public Health. 2016;45(3):305 pmid:27141492