|Home | About | Journals | Submit | Contact Us | Français|
A few reports have suggested that HTLV-1 may influence immunological response and therefore, clinical course of tuberculosis in co-infected individuals. We wished to determine the prevalence of HTLV-1 infection among hospitalized patients in Salvador, Brazil, a region endemic for both HTLV-1 infection and latent tuberculosis infection.
A cross-sectional study was conducted at a pulmonary disease hospital between September 1st of 2006 to August 31st of 2007. Study participants were interviewed and tested for HTLV-1 infection and current or past episode of tuberculosis.
Of 607 participants recruited into the study, 360 (59.3%) had current or past history of tuberculosis and 50 (8.2%) had HTLV-1 infection; 39 (6.4%) had both. After controlling for confounding variables, we found that the odds of patients with a positive HTLV-1 test having tuberculosis were 2.57 times the odds (95%, CI: 1,23, 5:35) in those who tested negative for HTLV-1 infection.
In a region endemic for both tuberculosis and HTLV-1 infection, HTLV-1 infection increases the risk of Mycobacterium tuberculosis infection. Such a risk may influence tuberculosis transmission and therefore epidemiology of the disease in this community.
The human T cell lymphotropic virus type 1 infects about 20 million individuals worldwide with highest prevalence in Africa, Central and South America and Southeastern Japan 1. HTLV-1 predominantly infects T cells leading to a disregulation in the immune response, which may not only increase susceptibility but also influence the clinical course of other concurrent infectious diseases. Salvador, Bahia, is a city in Brazil that has the highest prevalence of HTLV-1 infection with a seroprevalence of 1.76% documented in one population-based survey 2 and 1.35% among blood donors 3. The prevalence of tuberculosis in Salvador is 3.7% and the prevalence of infection among household contact in early 2000s was about 64% 4. The association between HTLV-1 and tuberculosis has also been observed in this population. Among patients with tuberculosis identified at health posts, the risk of HTLV-1 infection (4.27%) was higher than in controls (1.32%) 5. However other studies have not found an association between tuberculosis and HTLV-1 6, 7. The present study was conducted to determine the influence of HTLV-1 on the epidemiology of tuberculosis, and it examined hospital admitted patients with and without tuberculosis in a setting endemic for both infections.
This study was conducted at Hospital Especializado Octávio Mangabeira, a 217-bed pulmonary disease referral hospital in the city of Salvador-Bahia, Brazil. This hospital is the main center for treatment of tuberculosis patients in the state of Bahia. The study was approved by the Institution Review Board of the Hospital and the study subjects gave written informed consent before entering the study.
The study subjects were consecutively recruited from September 1st 2006 to August 31st 2007 in four wards of the hospital - two wards reserved for men and two for women. A case was defined as a patient diagnosed with active tuberculosis at the time of recruitment, or having a history of previous treatment or hospitalization for tuberculosis. Controls were defined as those with no history of tuberculosis admitted to the hospital for diseases other than tuberculosis. A standardized questionnaire soliciting for demographic and clinical information was administered to these study subjects.
Five milliliters of blood were drawn from each study subject. Sera were screened for antibodies to HTLV-1 and HTLV-2 by enzyme linked immunosorbent assay (Murex HTLV-I + II, Abbot, Dastford, UK), and for HIV-1 and 2 by HIV-1/HIV-2 Ab capture ELISA test system, Orhoclinical Diagnostic). All samples with positive results for HTLV or HIV were submitted for further serological confirmation. The discrimination between HTLV-1 and HTLV-2 was performed by Western blot analysis (HTLV Blot2.4; Genelabs, Singapore). The discrimination between HIV I and HIV II was performed by Western blot assay (Cambridge Biotech Corp. Worchester, USA).
Based on univariate analysis and unadjusted odds ratios, five variables in addition to the main exposure of interest, HTLV-1, were considered for inclusion in a multivariate logistic regression model. All variables were independently assessed for colinearity and effect modification. The final model was performed by incorporating variables based on the “likehood ratio” with α = 0.05, and odds ratio (OR) with 95% confidence intervals (CIs). All analyses were done by statistical software STATA 10.1 (Stata Corp, College Station, TX) and SPSS version 11.0 (SPSS, Inc., Chicago, IL).
Questionnaires were administered to 607 participants; 360 (59.3%) met the case definition and 247 (41%) met the definition for controls. Of the cases, 39 (10.8%) were co-infected with HTLV-1, while among the controls, 11 (4.5%) were infected with HTLV-1 (p=0.005). The demographic and clinical data of these cases and controls are shown in Table 1. By univariate analyses, cases were found to be younger (mean=42.9 years, p-value <0.0005), more likely to be males (69.4%, p-value = 0.0009) and were more frequently involved in a sexual relationship with a member of the same sex (6.4%, p-value = 0.0126). Presence of tattoo (15.3%, p-value<0.005) and HIV infection (11.1%, p-value= 0.0004) were also associated with tuberculosis.
A manual forward stepwise procedure was carried out to arrive at the final statistical model (Table 2) describing the association between HTLV-1 and tuberculosis in this study population. Each of the five variables - HIV, gender, sexual relationship with an IV drug user, presence of tattoo and age over 50 – were added singly and assessed for potential confounding. Addition of a variable leading to a greater than 10% change in the effect estimate of our primary exposure, HTLV-1 infection, was considered to be confounding the relationship between HTLV-1 and tuberculosis and therefore retained in the model. Considerable attention was given to the exposure of HIV infection with respect to the association between HTLV-1 and tuberculosis.
Independent analysis of these three variables revealed that the association between HTLV-1 and tuberculosis vary by HIV status. Among HIV positive participants there was no association between HTLV-1 and tuberculosis (OR: 1.0, 95% CI: 0.15, 11.7) while among HIV negative participants, the association was significant (OR: 2.5, 95% CI: 1.14, 6.23). However, this study was not properly powered to detect such an interaction, resulting in a non-significant HIV-HTLV interaction term (p>0.1). Regardless of possible interactions, HIV infection was clearly confounding the main association and was therefore retained in the model as a single term. The elevated p-value for the HIV and tattoo variables in the final model of 0.069 and 0.063, respectively, is largely due to the inclusion of the dichotomous age variable (above or below 50 years of age). Only four of the 69 participants with tattoos and four of the 48 HIV positive participants were over the age of 50. The sparseness of individuals with these characteristics over the age of 50 resulted in the elevated p-values. Both variables were retained for the final model.
We also compared the same variables among the group of patients who had tuberculosis and HTLV-1 with those with tuberculosis without HTLV-1 (Table 3). Patients with HTLV-1 infection and tuberculosis were more likely to have a history of intravenous drug use, previous treatment for tuberculosis, previous hospitalization for tuberculosis and to be infected with HIV.
In this study, we had an opportunity to determine the association of HTLV-1 infection in a setting endemic for both latent tuberculosis infection and HTLV-1 infection among patients admitted to a referral pulmonary disease hospital in Salvador, Brazil. We found that indeed, individuals with HTLV-1 infection were significantly more likely to have current or past tuberculosis. In the same city, Marinho et al previously showed that the prevalence of HTLV-1 among outpatients with tuberculosis was higher than in patients without tuberculosis 5. More recently several studies have shown similar results 8, 9, but studies did not show that this association was significant 6, 7.
By univariate analysis, many other variables were found to be associated with tuberculosis, including age over 50 years, male gender, non-white race, history of sexual relationship with a member of the same sex, presence of tattoo, and HIV infection. When a preliminary logistic regression model was used, only HTLV-1, male gender, and age over 50, HIV and presence of a tattoo, remained significantly associated. As other studies have shown an association between HIV infection and tuberculosis, we kept HIV infection in the final regression model. Again only HTLV-1 infection, male gender, and age over 50 remained statistically significant. In previous studies that showed no association of HTLV-1 infection with tuberculosis, the sample sizes were generally small. Here using hospitalized patients we confirm a previous observation performed in a population-based study 5 that HTLV-1 infection is associated with tuberculosis.
The association of tuberculosis with HTLV-1 infection could be due to factors other than HTLV-1, since many of these co-infected subjects share similar demographic and epidemiologic characteristics in this endemic setting. However, we found no such characteristics common to both. Thus, this association is likely to be biologic.
HTLV-1 has been associated with other infectious diseases such as strongyloidiasis and scabies. In these cases, HTLV-1 infection was associated with a severe clinical manifestation of strongyloidiasis and scabies 10, 11. There is also an increased prevalence of helminthes infection in HTLV-1 infected subjects as compared to seronegative controls 12. Type 2 T-helper cell immune response participates as part of the defense mechanisms against helminthes infections and it has been well documented that HTLV-1 decreases IL-4, IL-5 and IgE production in patients with strongyloidiasis 13. These abnormalities facilitate auto-infection and decrease worm killing allowing dissemination of Strongyloides stercoralis. The host defense against Mycobacterium tuberculosis infection is believed to involve predominantly activated macrophages and CD8 T cells 14, 15. HTLV-1 is associated with T cell activation and elevated production of IFN-γ and TNF-α in unstimulated cultures 16, 17. However antigen specific immune response may be compromised in HTLV-1. Compared to seronegative controls, HTLV-1 infected carriers have more negative intradermal PPD test 18, 19, and lymphocyte proliferative response to PPD is decreased in HTLV-1 infected subjects 20. In such case the association of tuberculosis with HTLV-1 could be related to a decreased specific immunological response to mycobacterium antigens.
In a retrospective study performed by Pedral-Sampaio et al, a high rate of mortality was observed in patients co-infected with HTLV-1 and tuberculosis 21. In the present study a retrospective analysis of the clinical manifestations and laboratory data did not allow us to determine the influence of HTLV-1 on the clinical course of tuberculosis, due to lack of relevant clinical and microbiological information in the charts. However, patients with HTLV-1 plus tuberculosis had some indirect findings that could be related to severity of diseases. For instance, patients with HTLV-1 infection and tuberculosis had a greater number of previous hospital admissions for tuberculosis and treatment for tuberculosis, despite no evidence of their defaulting on treatment for tuberculosis.
HTLV-1 and HIV are retroviruses that infect T cells. While opportunistic infection is highly associated with HIV, in HTLV-1 the only well established associated infectious diseases are helminthic infections. In such case the exaggerated type 1 immune response induced by HTLV-1 decrease production of Th2 cytokines and IgE molecules that participate of the defense against helminthes. The increasing observation of an association between tuberculosis and HTLV-1 is a strong indicator that HTLV-1 may influence also immunologic response against infectious agents that are controlled by the type 1 immune response.
HTLV-1 infection increases the risk of Mycobacterium tuberculosis infection and patients with HTLV-1 and tuberculosis had a greater number of hospital admissions for tuberculosis and treatment for tuberculosis. HTLV-1 infection may influence the epidemiology of tuberculosis in regions endemic for both diseases.
This study was supported by the National Institute of Tropical Diseases (INCT/CNPq), NIH/FIC grant D43 TW007127 and NIH grant # R01 AI079238A. We thank the secretarial assistance of Elbe Silva and Cristiano Sampaio in the preparation of the manuscript. EMC is a senior investigator of CNPq.
MLB and EMC coordinated the study, reviewed the data and wrote the final version of the manuscript. OB performed the statistical analysis, DLM, CAC, MJA and LS applied the questionnaires, DNP applied the questionnaires and entry the data in the database and LR participated on the study design and statistical analysis.