The results of numerous studies leave no doubt that infection with HIV-1 is currently a powerful catalyst that favours the progression to clinical disease of tuberculous infection, whether latent or newly acquired. Indeed, the risk of developing active tuberculosis among individuals co-infected with HIV-1 and Mycobacterium tuberculosis is 5–15% per year , in comparison with the risk in non-HIV-infected individuals of only 10% per lifetime . Much of the information concerning the interaction between HIV-1 and M. tuberculosis comes from sub-Saharan Africa, the region of the world most heavily afflicted with HIV-1, and where nearly 70% of all dually infected individuals live .
Considerably less is known about a kindred retrovirus, HIV-2, which occurs chiefly in western Africa , and appears to affect older individuals and to suppress the immune system more slowly than HIV-1 [5,6]. A study from Bissau showed that tuberculosis was more common in HIV-2-positive hospitalized patients than in HIV-negative individuals . To provide additional insight into the relationship between HIV-2 and tuberculosis, we analysed data from a community-based surveillance study in Bissau, the capital of Guinea-Bissau, the country with the highest prevalence of HIV-2 in the world. In addition, the onset and spread of HIV-1 in the study area beginning in the early 1990s [5,8] afforded the unique opportunity of comparing the relative impacts of HIV-1 and HIV-2 on the incidence of and mortality from tuberculosis.
The community tuberculosis surveillance programmeBeginning in May 1996, a tuberculosis surveillance programme based on passive and active case finding was implemented in four suburban areas (Bandim 1, Bandim 2, Belem and Mindara) in Bissau, the capital city of Guinea-Bissau. Bissau has an estimated total population of approximately 300 000 people. Bandim 1 has been under demographic surveillance since 1978, Bandim 2 and Belem since 1984, and Mindara since 1996. The total study population was 42 700 inhabitants in May 1997. Two surveys conducted in 1987  and 1996  reported the highest HIV-2 seroprevalence ever observed in a community. Of individuals 15 years of age or older, 6.8% were HIV-2 seropositive. HIV-1 infection recently emerged in this community, its seroprevalence being 0% in the late 1980s  and 2.2% in 1996 .
The only specialized tuberculosis service in Guinea-Bissau is found in the capital city. Many patients from rural areas who have chronic respiratory diseases move temporarily to live with relatives in the capital, where the health services are known to be better. These patients, without a permanent address in the study area, were classified in the present study as ‘visitors'. Following International Union Against Tuberculosis and Lung Disease/World Health Organization guidelines, all residents and visitors aged 15 years or older who presented at either of the two local health centres with signs or symptoms suggestive of tuberculosis were investigated for tuberculosis disease. Because the signs and symptoms of tuberculosis are well known to the population, many patients came directly to Hospital Raoul Follereau, the national referral hospital. To identify patients in the project's demographic surveillance system, the senior nurse, who first interviewed all patients and ascertained their addresses, referred both residents and visitors from the study area to one of the two physicians responsible for the enrolment of tuberculosis patients.
Active case finding among contacts at home was carried out by the field assistant or project nurse as soon as a new case was detected. During home visits, the nurse asked about people suffering from signs and symptoms suggestive of tuberculosis. Suspected cases were referred to Hospital Raoul Follereau for further investigation.
To assess the coverage of the tuberculosis surveillance system, we identified all resident adults who had died in the area during the study period. Verbal autopsies were performed for 511 adults by two Guinean physicians; autopsy forms were reviewed by all participating doctors to decide whether the individual was likely to have died from tuberculosis. Only two of the 39 individuals who were believed to have died from tuberculosis were not included in the study.
Diagnoses and treatment regimen
Morning sputum samples were obtained on three consecutive days, kept in the dark at +4°C, and transported to the National Public Health Laboratory the same day for examination. Direct microscopy was performed using Ziehl–Neelsen staining. For culture all specimens were incubated at 37°C for 7 weeks in conventional Löwenstein–Jensen egg medium and in a modified Löwenstein–Jensen medium with 0.6% pyruvate. Tubes were examined weekly. Growth was confirmed by microscopy.
One or more sputum samples positive by direct microscopy or by culture, together with clinical and radiological findings supporting a diagnosis of tuberculosis, was regarded as bacteriological confirmation. Patients with clinical signs and symptoms of tuberculosis, and X-ray changes compatible with active intrathoracic tuberculosis, but without bacteriological confirmation, were treated with antibiotics (co-trimoxazole or amoxicillin) for 10 days and then re-evaluated clinically and with a chest X-ray. If there was no improvement and the suspicion of tuberculosis remained, the patient was classified as having presumed tuberculosis and was treated.
Tuberculosis patients received chemotherapy for 8 months: 4 months of isoniazid, rifampicin, pyrazinamide and ethambutol, followed by 4 months of isoniazid and ethambutol. In addition, multivitamins including vitamin B were provided daily. During the first 4 months of treatment, drugs were administered using directly observed therapy. During the last 4 months, drugs were given to patients twice a month for 15 days of treatment. To assess compliance, pills were counted at each clinical appointment and urine was tested for isoniazid at 2, 5 and 8 months. Cases presenting with severe disease or patients having problems with housing in Bissau were offered hospitalization usually for approximately 2 months.
After obtaining informed consent to participate, blood was drawn and sera tested for HIV antibodies. Five individuals (one HIV-1-positive, four HIV-negative) were tested at a clinical follow-up examination and not at the initial recruitment. HIV testing was performed at the National Public Health Laboratory. Sera were screened using Enzygnost Anti-HIV 1+2 Plus (Behring Diagnostics Gmbh, Marburg, Germany) and confirmed with Capillus HIV-1/HIV-2 (Cambridge Diagnostics, Galway, Ireland) or Multispot HIV-1/HIV-2 (Sanofi Diagnostics Pasteur, Marnes-la Coquette, France). Dual reactive samples were sent frozen to the Swedish Institute for Infectious Disease Control in Stockholm and confirmed using Immunocomb II HIV-1&2 Bispot (Orgenics, Yavne, Israel).
The study protocol was approved by both the Central Ethical Committee of Denmark and the Ministry of Health of Guinea-Bissau.
Follow-up of tuberculosis patients and mortality
Because directly observed treatment was routinely implemented, tuberculosis patients were seen 6 days a week during the first 4 months of treatment. Defaulting patients were visited by the nurses and encouraged to continue treatment. If defaulting was caused by the physical inability to come to the health centre, hospitalization was offered. During the last 4 months of treatment, patients were invited to attend the health centre twice a month. Clinical examinations were performed at 2, 5 and 8 months by project physicians. As a result of the intensive monitoring of treatment, deaths are reported with a precise date of the event.
HIV sero-survey in the community
To estimate the HIV seroprevalence in the community, an HIV-1 and HIV-2 survey was performed in 1996 . This survey included individuals aged 15 years or older living in a random sample of 212 houses in Bandim 1, Bandim 2, and Belem. Among the 1748 individuals living in these homes, 1505 (86.1%) individuals were tested for HIV. Reasons for not testing were refusals (5.5%), insufficient blood for HIV test (0.4%), and absence because of school, work or travel (7.9%).
Statistical analyses were made using EPI INFO v.6.04 and SAS for Windows v.6.12. The incidence of tuberculosis was determined in relation to the prevalence of HIV infection in the community. To estimate the number of HIV-infected and non-HIV-infected individuals in the community, we assumed that age and sex-specific HIV prevalences obtained in the 1996 survey of three districts applied to the population of all four areas. Census data from May 1997 indicated that the adult population was 24 878 individuals. Individuals afflicted twice with tuberculosis were excluded on the second occurrence. Visitors were not included in these analyses.
The proportion of tuberculosis disease attributable to different HIV types (HIV-1, HIV-2 and dual infections) was estimated using the population attributable fraction (PAF):EQUATION where P is the prevalence of HIV infection in the population, and RR is the relative risk of tuberculosis in HIV-infected versus non-infected individuals.
Patients were included in the survival analysis from the date of the initiation of tuberculosis treatment until death, migration, end of treatment, or 7 June 1998, when a civil war broke out in Guinea-Bissau, whichever came first. Survival curves were estimated using the Kaplan–Meier product limit method. Multivariate Cox regression analyses were used to compare different groups and to adjust for background factors. Effects were expressed as mortality rate ratios (MR) with 95% confidence intervals (CI).
Demographic, clinical and serological findings at inclusion
From 6 May 1996 to 7 June 1998 a total of 520 patients were included in the study. A total of 154 were diagnosed as having non-tuberculosis disease, usually pneumonia or bronchitis, and 366 were diagnosed as having active intrathoracic tuberculosis. The diagnosis of tuberculosis was bacteriologically confirmed in 195 cases (53%). The remaining 171 tuberculosis cases had either initiated treatment elsewhere (n = 35) before inclusion and the sputum results were thus not reliable, or were diagnosed on clinical and radiological grounds. Of the 366 tuberculosis cases 247 were residents and 119 visitors (Table 1). The mean age was 40.1 years (± 15.8) and the female : male ratio was 0.54 (128 : 238). Visitors were older than residents and had a higher female : male ratio (Table 1). Furthermore, visitors were less often HIV-1 positive and more often hospitalized (Table 1).
The proportion of smear-negative tuberculosis was higher in HIV-positive cases than in HIV-negative cases (RR 1.19; CI 1.01–1.40). HIV-2-positive cases were more likely to have negative sputum smears than HIV-1 positive cases (RR 0.73; CI 0.56–0.95). A large number of tuberculosis patients (310/366) came directly to the national hospital without referral from health centres. Therefore, most tuberculosis patients were found through passive case finding. Active case finding added 28 more cases.
Incidence of tuberculosis and tuberculosis disease attributable to HIV
Incidence rates for tuberculosis according to different HIV types (HIV-1, HIV-2, and dual infections) are presented in Table 2. From the serological survey, we estimated that 2048 HIV-positive adults were living in the area. Ninety-seven of these developed active tuberculosis. Adjusted for age, both HIV-1 and HIV-2-positive individuals had higher incidences of tuberculosis compared with HIV-negative individuals (RR 18.3; CI 12.9–26.0 and RR 3.0; CI 2.1–4.3), respectively). The incidence of tuberculosis in HIV-1-positive individuals was 5.2 times higher than in HIV-2-positive individuals (RR 5.2; CI 3.2–8.4). There was no difference between HIV-1-positive and dually reactive individuals (RR 1.13; CI 0.67–1.91). The statistical differences between the HIV types remained essentially unchanged when we restricted the analysis to bacteriologically confirmed cases.
HIV-1 and HIV-2 infections were associated with nearly equal numbers of patients with tuberculosis in the community. HIV-1 and dual infection combined, however, was more important than HIV-2, even though the seroprevalence of HIV-1 and dual infection in the community was much lower than the seroprevalence of HIV-2 infection (Table 2).
Mortality was estimated for the tuberculosis patients during their 8 months of treatment. Follow-up stopped on the 7 June 1998, when a civil war started. During the follow-up, 45 patients abandoned treatment or were transferred out and then lost to follow-up. These patients were censored accordingly. None of the patients was positive in direct microscopy of sputum at the end of the treatment. Adjusted for age and sex, HIV-2-positive patients had no worse survival than HIV-negative patients (Table 3). Compared with HIV-negative patients, HIV-1 (MR 2.7; CI 1.1–6.5) and dually infected (MR 2.9; CI 1.1–7.4) tuberculosis cases had higher mortality rates. These tendencies remained unchanged when analyses were restricted to bacteriologically confirmed tuberculosis cases (Table 3). Survival curves are presented in Fig. 1.
The results of our HIV serological survey and tuberculosis surveillance system in closely monitored areas of Bissau provided the opportunity for defining the relative impacts of HIV-1 and HIV-2 on both the incidence of and mortality from tuberculosis. Infection with each of the two HIV subgroups increased the incidence of tuberculosis in comparison with that in non-HIV-infected individuals, but the risk associated with infection with HIV-1 was substantially greater than that from HIV-2, 18.3 versus 3.0. Moreover, mortality rates in HIV-2 infected patients with tuberculosis did not differ from those in non-HIV-infected patients, whereas mortality rates accompanying HIV-1 infection were 2.7 times greater.
Previous studies have documented that HIV-2 is less immunosuppressive than HIV-1 [6,9–11], but the sixfold decrease in the incidence of tuberculosis in individuals with HIV-2 infection compared with those with HIV-1 infection is, nevertheless, surprising. A slower progression of immunosuppression might lead to a difference in the age of onset of tuberculosis in the two HIV subgroups, older in HIV-2 than in HIV-1, but not necessarily in overall incidence. A lower level of viral replication provides a satisfactory explanation for the longer clinical latency and diminished transmission of HIV-2 compared with HIV-1 [12,13], but may not fully account for their remarkably different effects on favouring the development of tuberculosis. Subtle differences between the two virus types on specific immunological defences may also be involved.
Our finding of an equal susceptibility to tuberculosis in dually infected and HIV-1-infected individuals is consistent with previous studies of immunodepression in these two types of infection [14,15]. A study from Abidjan  estimated the relative risk of developing tuberculosis for HIV-1 and HIV-2-infected individuals by comparing the HIV prevalences among tuberculosis patients and blood donors. The estimates for HIV-2 [odds ratio (OR) 4.2] and dual infections (OR 8.9) corresponded well with what we observed in Bissau. However, their estimate for HIV-1 (OR 4.7) was considerably lower than in Bissau, and may have been too low if the blood donors were not representative of the general population in Abidjan. It is striking that HIV-1 infection, which emerged only a few years ago in this community, has a greater impact on the incidence of tuberculosis than HIV-2 infection. With the current tendency towards a stable or declining prevalence of HIV-2 and increasing prevalence of HIV-1 [17,18], the tuberculosis case load attributed to HIV-1 will increasingly predominate in West African countries.
Accurate estimates of the HIV-associated incidence of tuberculosis are mostly unobtainable in developing countries. Many cases are likely to be missed as a result of a poorly developed notification system and the seroprevalence of HIV in the general population is generally estimated from special subgroups such as blood donors, pregnant women or military recruits. We believe the tuberculosis surveillance system that we instituted in the study area and the concomitant HIV survey in the same communities should yield a reliable estimate of the incidence rate of HIV-associated tuberculosis. Indeed, as a result of active case finding, the referral of suspected cases, and verbal autopsy for all deaths in the designated communities, virtually all tuberculosis cases occurring within the study period are likely to have been identified. Under ordinary circumstances, cases of tuberculosis among visitors would have been a concern for a community-based evaluation but, in our study area, the quality of census taking allowed us to exclude all these individuals.
The high proportion of acid-fast bacilli smear negativity among the tuberculosis patients may raise doubts about the accuracy of diagnostic procedures. However, it is well established that smear-negative tuberculosis is more frequent among HIV-positive than among HIV-negative patients [19,20]. It should also be noted that we have included all forms of intrathoracic tuberculosis, such as pleural effusion and intrathoracic lymphadenopathy, in the analyses. These forms are rarely smear-positive. When analyses were restricted to patients with bacteriologically confirmed tuberculosis, the observed differences remained essentially unchanged.
Several studies have shown that short-term mortality among tuberculosis patients is higher in HIV-1-positive than in HIV-negative patients. Our results are similar to those from two studies from other West African countries, Burkina Faso and Ivory Coast, where both HIV types are prevalent [21,22], as well as with a hospital study from Bissau where no difference was found in the short-term survival between HIV-2-positive and HIV-negative patients . In another study from Abidjan using a 6 month treatment period, the mortality rate ratios for HIV-1 and HIV-2-infected tuberculosis cases compared with HIV-negative patients were 6.5 (CI 2.5–19.5) and 3.0 (CI 0.8–10.6), respectively . In that study, only smear-positive patients were included and the patients were considerably younger. Previous studies have demonstrated that the severity of immune deficiency was the major independent factor of mortality in HIV-infected patients with tuberculosis [22,25]. The relatively low degree of immunosuppression associated with HIV-2 may explain the limited impact on mortality. Data on the extent of immunosuppression from our material have yet to be analysed.
To our knowledge this paper presents the first data from a rigorous community-based study on the relative impacts of HIV-1 and HIV-2 infection on the development of active tuberculosis, and mortality during treatment. As a result of its higher prevalence, HIV-2-infection will continue to make a significant contribution to the incidence of tuberculosis and the associated mortality in Guinea-Bissau. To make matters worse, the recent arrival and ongoing spread of HIV-1 heralds a dramatic future increase in the incidence of tuberculosis and tuberculosis-related mortality in Guinea-Bissau and neighbouring countries. Health authorities should be ready to reinforce national tuberculosis control programmes in order to respond to this inevitable and growing public health challenge.
The following persons contributed invaluably to the study: Peter Aaby, Francisco Dias, Mali Jalo, Peter Johansson, John F. Murray, Anders Naucler, Anita Sandström, Margarida G.D.S. Teixeira. Particular thanks to the assistants and nurses working daily in the study area.
1. Raviglione MC, Harries AD, Msiska R, Wilkinson D, Nunn P. Tuberculosis
and HIV: current status in Africa
. AIDS 1997, 11 (Suppl. B) : S115–S123.
2. Styblo K. Epidemiology of tuberculosis.
The Hague: Royal Netherlands Tuberculosis
Association; 1991. pp. 55–66.
3. Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Consensus statement.Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. WHO Global Surveillance and Monitoring Project.
JAMA 1999, 282: 677–686.
4. Romieu I, Marlink R, Kanki P, M'boup S, Essex M. HIV-2
link to AIDS in West Africa
. J Acquired Immune Defic Syndr 1990, 3: 220–230.
5. Poulsen AG, Kvinesdal B, Aaby P. et al
. Prevalence of and mortality
from human immunodeficiency virus type 2 in Bissau, West Africa
[see Comments]. Lancet 1989, 1: 827–831.
6. Lisse IM, Poulsen AG, Aaby P, Knudsen K, Dias F. Serial CD4 and CD8 T-lymphocyte counts and associated mortality
in an HIV-2
-infected population in Guinea-Bissau. J Acquired Immune Defic Syndr Hum Retrovirol 1996, 13: 355–362.
7. Naucler A, Albino P, Da Silva AP, Andreasson PA, Andersson S, Biberfeld G. HIV-2
infection in hospitalized patients in Bissau, Guinea-Bissau. AIDS 1991, 5: 301–304.
8. Larsen O, da Silva Z, Sandstrom A. et al
. Declining HIV-2
prevalence and incidence
among men in a community
study from Guinea-Bissau. AIDS 1998, 12: 1707–1714.
9. Lisse IM, Poulsen AG, Aaby P. et al
. Immunodeficiency in HIV-2
infection: a community
study from Guinea-Bissau. AIDS 1990, 4: 1263–1266.
10. Jaffar S, Wilkins A, Ngom PT. et al
. Rate of decline of percentage CD4+ cells is faster in HIV-1
than in HIV-2
infection. J Acquired Immune Defic Syndr Hum Retrovirol 1997, 16: 327–332.
11. Marlink R, Kanki P, Thior I. et al
. Reduced rate of disease development after HIV-2
infection as compared to HIV-1
. Science 1994, 265: 1587–1590.
12. De Cock KM, Adjorlolo G, Ekpini E. et al
. Epidemiology and transmission of HIV-2.Why there is no HIV-2 pandemic [published erratum appears inJAMA1994 19 Jan; 271:196] [see Comments].
JAMA 1993, 270: 2083–2086.
13. Shanmugan V, Switzer WM, Nkengasong JN. et al
. Lower HIV-2
plasma viral loads may explain differences between the natural histories of HIV-1
infections. J Acquired Immune Defic Syndr 2000, 24: 257–263.
14. Kestens L, Brattegaard K, Adjorlolo G. et al
. Immunological comparison of HIV-1
- and dually-reactive women delivering in Abidjan, Cote d'Ivoire. AIDS 1992, 6: 803–807.
15. Ghys PD, Diallo MO, Ettiegne-Traore V. et al
. Dual seroreactivity to HIV-1
in female sex workers in Abidjan, Cote d'Ivoire. AIDS 1995, 9: 955–958.
16. De Cock KM, Gnaore E, Adjorlolo G. et al
. Risk of tuberculosis
in patients with HIV-I and HIV-II infections in Abidjan, Ivory Coast. BMJ 1991, 302: 496–499.
17. Djomand G, Greenberg AE, Sassan-Morokro M. et al
. The epidemic of HIV/AIDS in Abidjan, Cote d'Ivoire: a review of data collected by Projet RETRO-CI from 1987 to 1993. J Acquired Immune Defic Syndr Hum Retrovirol 1995, 10: 358–365.
18. Norrgren H, Andersson S, Dias F, Naucler A, Biberfeld G. Trends of incidence and prevalence of HIV-1 and HIV-2 in Guinea-Bissau, West Africa.
In:Xth International Conference on AIDS and STD in Africa
. Abidjan, 1997 [Abstract B.045].
19. Samb B, Sow PS, Kony S. et al
. Risk factors for negative sputum acid-fast bacilli smears in pulmonary tuberculosis
: results from Dakar, Senegal, a city with low HIV seroprevalence. Int J Tubercl Lung Dis 1999, 3: 330–336.
20. Karstaedt AS, Jones N, Khoosal M, Crewe-Brown HH. The bacteriology of pulmonary tuberculosis
in a population with high human immunodeficiency virus seroprevalence. Int J Tubercl Lung Dis 1998, 2: 312–316.
21. Malkin JE, Prazuck T, Simonnet F. et al
and human immunodeficiency virus infection in west Burkina Faso: clinical presentation and clinical evolution. Int J Tubercl Lung Dis 1997, 1: 68–74.
22. Ackah AN, Coulibaly D, Digbeu H. et al
. Response to treatment, mortality
, and CD4 lymphocyte counts in HIV-infected persons with tuberculosis
in Abidjan, Cote d'Ivoire. Lancet 1995, 345: 607–610.
23. Naucler A, Winqvist N, Dias F. et al
. Pulmonary tuberculosis
in Guinea-Bissau: clinical and bacteriological findings, human immunodeficiency virus status and short term survival of hospitalized patients. Tubercl Lung Dis 1996, 77: 226–232.
24. Kassim S, Sassan-Morokro M, Ackah A. et al
. Two-year follow-up of persons with HIV-1
- and HIV-2
-associated pulmonary tuberculosis
treated with short-course chemotherapy in West Africa
. AIDS 1995, 9: 1185–1191.
25. Shafer RW, Bloch AB, Larkin C. et al
. Predictors of survival in HIV-infected tuberculosis
patients. AIDS 1996, 10: 269–272.