In human immunodeficiency virus (HIV)-infected patients, bacterial lower respiratory tract infections are the most frequent respiratory diseases. They are frequently the first clinical manifestation of HIV infection.
The incidence and severity of bacterial lower respiratory tract infections increase with the degree of immunosuppression. At the acquired immune deficiency syndrome (AIDS) stage, the responsible bacteria and clinical presentation may be atypical. Bacterial pneumonia may be fatal, particularly in AIDS patients, and its occurrence is predictive of a reduced survival time.
Pneumococcal vaccine is recommended in patients with a CD4 T‐lymphocyte count of >200 cells·mm−3 and cotrimoxazole (trimethoprim/sulfamethoxazole) in patients with a CD4 T‐lymphocyte count of <200 cells·mm−3. Unfortunately, such prophylaxis remains insufficiently prescribed and its protective effect is limited.
Highly active antiretroviral treatment has dramatically reduced the incidence of lower respiratory tract infection due to Pseudomonas aeruginosa and opportunistic bacteria. In contrast, successful highly active antiretroviral therapy slightly decreased the risk of bacterial pneumonia due to usual bacteria, even in patients on successful highly active antiretroviral therapy.
- acquired immune deficiency syndrome
- bacterial bronchitis
- bacterial pneumonia
- human immunodeficiency virus infection
At the beginning of the acquired immune deficiency syndrome (AIDS) epidemic, it was clear that the lung of human immunodeficiency virus (HIV)-infected patients was the major target for many infections and tumours, and, in the first series reported, Pneumocystis carinii pneumonia (PCP) and bronchopulmonary localisation of Kaposi's sarcoma were mainly incriminated 1, 2. It is only gradually that pyogenic bacteria have been recognised to be a major cause of lower respiratory tract infection (LRTI) in HIV-infected patients, whatever their level of immunosuppression 3–5.
During the first decade of the AIDS epidemic, it was shown that the occurrence of pyogenic bacterial infection might partly be prevented by the use of prophylaxis, and, more recently, the use of highly active antiretroviral therapy (HAART) has been shown to have an indirect (immune restoration) but long-lasting preventive effect. Thus, in the current world, three very different situations have to be considered 6.
Bacterial lower respiratory tract infections in human immunodeficiency virus-infected patients who do not receive prophylaxis or antiretroviral treatment
This situation concerns the majority of HIV-infected patients in developing countries, but, in developed countries, some patients without knowledge of their seropositivity or without appropriate follow-up are also concerned. In such patients, the natural history of HIV-associated LRTI is obviously the same as it was at the beginning of the AIDS epidemic.
Incidence of pyogenic bacterial lower respiratory tract infection
Incidence in developed countries
In developed countries, Wallace et al. 7 examined trends in the incidence of specific respiratory disorders in a multicentric cohort with progressive HIV disease during a 5‐yr period. Individuals with a wide range of HIV disease severity and belonging to the three main transmission categories were evaluated at regular intervals and for episodic respiratory symptoms using standard diagnostic algorithms. Acute bronchitis and bacterial pneumonia were the most common respiratory diseases of cohort members. The respective incidences per 100 person-yrs were 13.7 for bronchitis and 5.5 for bacterial pneumonia. When HIV-infected individuals were compared with controls of similar age, race, sex and transmission group, the incidence rates were two-fold greater for bronchitis and six-fold greater for bacterial pneumonia 7.
Similar results were found in a US case controlled study performed in an intravenous drug user cohort, in which the incidence of bacterial pneumonia per 100 person-years was 1.93 in HIV-seropositive and 0.45 in HIV-seronegative subjects 8.
Finally, in an Italian intravenous drug user cohort, Boschini et al. 9 reported 149 episodes of community-acquired pneumonia among HIV-seropositive patients and 61 among HIV-seronegative subjects with incidence rates per 1,000 person-yrs of 90.5 and 14.2, respectively.
Incidence in developing countries
Gilks et al. 10 followed up a cohort of HIV-positive and -negative female sex workers in Nairobi, Kenya. The incidence rates of pneumonia per 1,000 person-yrs were 38.6 in HIV-seropositive females and 3.7 in HIV-seronegative females. More precisely, the incidence rates of invasive pneumococcal disease and pneumococcal bacteraemia per 1,000 person-yrs were 42.5 and 23.8 in HIV-seropositive females and 3.7 and 0 in HIV-seronegative females, respectively. The relative risk of development of invasive pneumococcal disease with underlying HIV infection was 17.8 (95% confidence interval 2.5–126.5) 10. These high incidence rates explain why, in prospective studies, pyogenic bacteria as well as Mycobacterium tuberculosis were recognised as the major cause of respiratory diseases in HIV-infected Africans 11–13.
Risk factors for bacterial lower respiratory tract infection
The risk of bacterial pneumonia is not the same in all HIV-infected persons. The most important risk factor for bacterial pneumonia is the degree of immunosuppression, as reflected by the CD4 T‐lymphocyte count. In the previously mentioned US cohort study 7, 14, acute bronchitis was equally prevalent during all stages of HIV disease, whereas the risk of bacterial pneumonia was clearly related to the study entry CD4 count (2.3, 6.8 and 10.8 episodes per 100 person-yrs in the subgroups of cohort members with >500, 200–500 and <200 CD4 T‐lymphocytes·mm−3) 14, 15. A similar relationship between a decreased CD4 T‐lymphocyte count and an increased risk of bacterial pneumonia has been found in European studies 9, 16. Other risks factors for bacterial pneumonia have also been identified. Hirschtick et al. 15 showed that the risk was greater in intravenous drug users than in homosexual male or female partners. In the same study, tobacco was equally shown to be an independent risk factor in the subgroup of HIV-seropositive subjects with <200 CD4 lymphocytes·mm−3. Another risk factor is neutropenia, which may result from direct retroviral infection, the use of antiretroviral and other drug therapy, systemic opportunistic infections and autoimmune mechanisms 17. Splenectomy 18, 19, previous pneumonia, whatever the cause 8, smoking illicit drugs 8 and low Karnofsky score 14 have also been associated with bacterial pneumonia.
Regarding specifically nosocomial pneumonia, advanced HIV infection, high Acute Physiology and Chronic Health Evaluation III score and central nervous system diseases have been shown to be risk factors in the case controlled study of Tumbarello et al. 20.
Bacteria responsible for lower respiratory tract infection
The first series investigating bacterial pneumonia in patients with AIDS or AIDS-related complex have shown the predominant role of Streptococcus pneumoniae and, to a lesser degree, Haemophilus influenzae, in adults as well as children 21, in developed countries 22–25 and developing countries 10, 13. Subsequently, other series with greater numbers of patients followed up over a long period have emphasised that three kinds of bacteria might be the cause in HIV-infected patients.
Typical pyogenic bacteria
In all series, typical pyogenic bacteria and, more particularly, S. pneumoniae and H. influenzae, were the major responsible bacteria. For example, in a French clinical epidemiology database, S. pneumoniae and H. influenzae were the cause of 52 and 16% of bacteriologically confirmed pneumonia, respectively 26. Similarly, in the US cohort study of Hirschtick et al. 15, S. pneumoniae and H. influenzae were found in 52 and 15% of patients with confirmed pneumonia, respectively.
In the African cohort study by Gilks et al. 10, 91% of pneumococcal serotypes incriminated in HIV-seropositive adults were included in the currently available pneumococcal vaccine. The principal serotypes were 1, 3, 5, 7, 19 and 23. In South Africa, 83% of serotypes incriminated in HIV-seropositive children were included in the nine-valent conjuguated pneumococcal vaccine 27. Finally, in the USA, Frankel et al. 28 also found that 90% of the serotypes responsible for invasive pneumococcal disease, particularly serotypes 9, 14, 6b and 22 f, were equally included in the polyvalent pneumococcal vaccine.
Concerning susceptibility of S. pneumoniae to penicillin, Bedos et al. 29 performed a retrospective study in France, to investigate the susceptibility of 10,350 S. pneumoniae strains to penicillin G and identify risk factors for infection with nonsusceptible strains. In a logistic regression model, HIV infection was associated with nonsusceptible strains (odds ratio (OR) 2.01). In US children, Mao et al. 30 also found that HIV-infected children were more likely to yield isolates exhibiting penicillin resistance, but such a relation was not found by Frankel et al. 28 in US adults. In South Africa, Jones et al. 27 also found that resistance to penicillin was increased in S. pneumoniae isolates from HIV-infected patients.
If bacteria other than those usually responsible for community-acquired pneumonia in HIV-negative adults are considered, Legionella pneumophila serogroup 1 has been infrequently incriminated in HIV-infected subjects 9. In a cohort of HIV-infected patients from the US air force, Blatt et al. 33 identified eight cases of legionella pneumonia. These cases represented 1.7% of patients with late-stage HIV infection and five of these cases were nosocomial. At the present time, L. pneumophila remains an uncommon pathogen in HIV-infected people.
Besides encapsulated pathogens and L. pneumophila, Boschini et al. 9 found a high incidence of community-acquired pneumonia due to Chlamydia pneumoniae (incidences per 1,000 patient-yrs of 18.64 for S. pneumoniae, 12.43 for H. influenzae and 11.30 for C. pneumoniae). This diagnosis was based on serological criteria, and, during the period of the study, there were two outbreaks of C. pneumoniae infection. In Germany, Dalhoff et al. 34 detected C. pneumoniae by polymerase chain reaction in bronchoalveolar lavage (BAL) fluid from 13% of hospitalised HIV-infected patients. However, acute respiratory symptoms were not always present, and, in 50% of cases, copathogens were found in the BAL fluid. Today, the precise role of C. pneumoniae as a cause of community-acquired pneumonia in HIV‐infected patients remains debatable.
Atypical pyogenic bacteria
Atypical pyogenic bacteria may also be the causative agent, particularly in patients with advanced HIV disease. For example, Klebsiella pneumoniae, other members of the Enterobacteriaceae family and Pseudomonas aeruginosa were present in 13, 10 and 8% of cases, respectively, of confirmed pneumonia in the US cohort study of Hirschtick et al. 15. Similarly, P. aeruginosa, the Enterobacteriaceae family and Staphylococcus aureus were the cause of 25, 9 and 10% of community-acquired pneumonia, respectively, in the series of Afessa et al. 31. These kinds of pathogen and their relative frequencies were consistent with most published reports 16, 35–37 and the microbiological data in the French epidemiological database 26.
Concerning S. aureus, Levine et al. 35 recovered this pathogen in 23% of respiratory tract cultures performed in 129 consecutive HIV-infected patients with an episode of respiratory disease. According to the authors, this presence of S. aureus was found to be community-acquired pneumonia in 28% of cases, of indeterminate significance in 62% and colonisation in 10%. None of the patients with pneumonia were neutropenic or on corticosteroids.
Some series on P. aeruginosa have been published. In HIV-infected patients, this pathogen may cause not only nosocomial pneumonia but also community-acquired pneumonia or chronic bronchitis. Traditional risk factors for the development of P. aeruginosa infections such as neutropenia, steroids, previous antimicrobial treatment and recent hospitalisation were absent in most patients. Nevertheless, this infection occurred almost exclusively at the AIDS stage in patients with a very low level of CD4 T‐lymphocytes 36, 38–40.
Finally, opportunistic bacteria have also been incriminated, although to a lesser extent.
Pasteurella multocida 50, Bordetella bronchiseptica 51, 52, Neisseria spp. 53, Rochalimaea spp. 54 Corynebacterium pseudodiphteriticum 55 and nonserotype 1 legionella 56 were equally the cause of pneumonia in case reports of HIV-infected patients with a high level of immunosuppression.
Nontyphoid strains of salmonella should be considered separately. In the series of Casado et al. 57, lung involvement was found in 18 of 51 (35%) of AIDS individuals with salmonella bacteraemia, but definite salmonella pulmonary infection was diagnosed in only six of 51 (12%). It is clear that not all respiratory diseases observed in patients with salmonella bacteraemia are due to this pathogen 11. Nevertheless, in a series of 38 HIV-infected patients with documented salmonellosis, two had a lung abcess and/or empyema, with Salmonella enteritidis being isolated from blood, sputum and pleural effusion 58.
Clinical presentation and diagnostic management of lower respiratory tract infection
Typical clinical presentation
In the vast majority of cases, the clinical presentation of HIV-infected subjects with bacterial pneumonia is similar to that of HIV-seronegative patients with community-acquired pneumonia. Onset is acute 59 or subacute 32 with fever, cough, purulent sputum, dyspnoea and chest pain. Physical examination reveals abnormalities 59, mainly crackles 37. Unilateral alveolar opacities are the most frequent radiological abnormalities 16, 37, 59–62. Laboratory evaluation usually shows a leucocytosis but leucopenia can occur. Serum lactate dehydrogenase levels are typically normal or mildly elevated 63. Hypoxaemia may be present 37. In the stepwise logistical analysis of Tumbarello et al. 16, the independent indicators of mortality were: CD4 T‐cell levels <100 cells·mm−3; arterial oxygen tension ≤9.3 kPa (≤70 mmHg); Karnofsky score <50; and neutrophil count <1,000 cells·mm−3 16. In severe cases, sepsis syndrome associated with hypotension, coagulopathy and multiorgan dysfunction can develop.
In a retrospective chart review, Selwyn et al. 59 tried to differentiate PCP, bacterial pneumonia and tuberculosis on the basis of clinical characteristics at presentation. On regression analysis, independent predictors for bacterial pneumonia included rales on examination (OR 12.4), a chart mention of “toxic” appearance (OR 9.1), fever for ≤7 days (OR 6.6) and lobar infiltrate (OR 5.8).
Nevertheless, there are four important differences between bacterial pneumonia occurring in HIV-infected patients and that in HIV-seronegative patients.
High frequency of bacteraemia
The first difference is the high frequency of bacteraemia in HIV-infected patients. In the prospective study of Plouffe et al. 64, the relative risk of pneumococcal bacteraemia among HIV-infected persons was 41.8 times that of controls aged 18–64 yrs. A similar increase in the incidence of pneumococcal bacteraemia was observed by Gilks et al. 65 in Africa. This high frequency is found in the great majority of series 22, 23, 66, 67, reaching 50% of bacterial pneumonia in the series of Gil Suay et al. 68. Bacteraemia occurred with this unusual frequency, whatever the pathogen, e.g. S. pneumoniae 25, 27, 31, 69, H. influenzae 21 S. aureus 21, Escherichia coli 21, P. aeruginosa 36–39 and even R. equi 47.
Unusual radiographic abnormalities
The second difference is the high frequency of unusual radiographic abnormalities. Magnenat et al. 70 reviewed the radiological features of 60 consecutive cases of bacterial pneumonia in HIV-infected patients. Forty-five per cent showed classic segmental or lobar alveolar consolidation patterns but 33 (55%) showed a predominantly interstitial infiltrate, this interstitial infiltrate being diffuse in 21 cases. There was no significant difference in the radiographic patterns between AIDS- and HIV-infected patients. In other series, a high frequency of multilobar consolidation or of bilateral involvement with either patchy bronchopneumonia or alveolointerstitial infiltrates has also been reported 16, 22, 35, 37, particularly with H. influenzae 32. In such cases, it might be difficult to distinguish bacterial pneumonia from opportunistic infections like PCP 70.
High rate of pleural effusions
The third difference is the high rate of pleural effusions associated with pneumonia. In the case controlled study of Gil Suay et al. 68, the rate of pleural effusions was two-fold higher in HIV-infected patients. These pleural effusions were mainly due to S. aureus. Their clinical course was more indolent with a long duration of symptoms before hospitalisation and more severe with a high percentage of bacteraemia and an unusual need for drainage. In the same way, bacterial pneumonia was the condition most commonly associated with pleural effusions in the series of Afessa 71. Such pleural effusions have been found with typical bacteria 37, as well as opportunistic bacteria such as N. asteroides 46 or R. equi 44.
Bacterial pneumonia due to opportunistic bacteria
The last difference concerns the unusual presentation of cases of bacterial pneumonia due to opportunistic bacteria, particularly N. asteroides and R. equi. The most frequent clinical and radiological features at presentation are subacute onset (median duration of symptoms prior to presentation frequently >1 month), respiratory symptoms, fever, weight loss and fatigue, lung consolidation or cavitary lesions. Extrapulmonary localisation, such as in pericardial effusion or a brain abcess, may be present at the time of presentation 44, 46.
Diagnostic management of lower respiratory tract infection
In practice, the clinical hypotheses and diagnostic approach clearly depend on clinical presentation and degree of immunosuppression. It is possible to distinguish between four schematic situations 72.
In the first situation, the clinical features are those of classic bacterial pneumonia with acute onset in an HIV-infected patient with a CD4 T‐cell count of >200 cells·mm−3. The pathogens responsible are typical bacteria, mainly S. pneumoniae. In all cases, emergency treatment, including a β‐lactam antibiotic effective against S. pneumoniae, should be started after blood culture. In cases with acute respiratory failure, fibreoptic bronchoscopy with protected bronchial brushing might be useful.
The second situation, defined by subacute progression of the condition, fever and diffuse opacities in a patient with a CD4 T‐cell count of <200 cells·mm−3, usually indicates an opportunistic pneumonia, such as PCP, but, in AIDS patients, typical bacteria and particularly H. influenzae might also be the cause. In all cases, fibreoptic bronchoscopy with BAL 73 and protected bronchial brushing is indicated 70. This should be performed before starting empirical treatment 74.
The third situation with subacute or moderate progression of the condition, fever, weight loss, nodules 75 or alveolar infiltrates, with or without cavitation 76, in a patient with a CD4 T‐cell count of <100 cells·mm−3, usually indicates an opportunistic infection due to mycobacteria, fungus, Nocardia spp. or R. equi, but S. aureus, members of the Enterobacteriaceae family or P. aeruginosa might also be responsible. In this situation, fibreoptic bronchoscopy with BAL and protected bronchial brushing or transthoracic needle aspiration 77 is indicated in all cases. Ideally, this should be performed after computed tomography (CT) to provide guidance regarding the optimal type and risk of sampling. After a negative first-line procedure including BAL, a second BAL combined with transbronchial biopsy allowed a definitive diagnosis in 90% of cases with nodules or focal infiltrates 78.
The last situation concerns HIV-infected patients with chronic sinusitis and chronic bronchitis. Such symptoms or signs usually occur in patients with lymphoid interstitial pneumonia or advanced HIV disease. Sinus and thoracic CT scans are of interest. Microbiological documentation is useful in guiding the choice of antibiotics.
Course of bacterial lower respiratory tract infection
In the majority of series of bacterial pneumonia in HIV-infected patients, the short-term mortality was the same as in HIV-seronegative subjects. In the series of 49 episodes of community-acquired lobar pneumonia of Miller et al. 67, 13 episodes were complicated by pleural effusion or empyema and 11 by lung cavitation. In this series, in spite of an unusual rate of complications, the mortality rate was only 8%. Many other series with S. pneumoniae 22, 28, H. influenzae 32, L. pneumophila 33 and other typical pathogens 37 have also shown a percentage of deaths similar to that usually observed in HIV-seronegative people. However, the bacteraemia level and mortality increase with the progression of HIV disease. This is clear for pneumonia due to P. aeruginosa 16, 36 or other opportunistic bacteria 44 observed exclusively at the AIDS stage. This is also true for pneumococcal pneumonia complicated by bacteraemia, the mortality clearly being higher in the subgroup of AIDS patients 79.
In patients cured of their acute episode of bacterial pneumonia, two late unfavourable events might occur: recurrence of LRTI and decrease in survival time.
High rate of recurrence
In the absence of immune restoration, obvious recurrence was observed in cases of opportunistic bacterial pneumonia, e.g. those due to R. equi 47, 48 or N. asteroides 44 needing a very prolonged period of curative treatment 44, 47. Nevertheless, recurrence has also been observed with typical bacterial pneumonia. In the cohort study of Gilks et al. 10, a first episode of bacterial pneumonia was observed in 54 patients during a 36-month follow-up. During this relatively short follow-up period, a first recurrence of bacterial pneumonia occurred in 24% of cases, which is the very high recurrence rate usually reported for HIV-infected patients 22, 28, 69, 76. The recurrence might be due to the same or a new pathogen 80.
Such recurrence has also been observed for bacterial bronchitis and mainly described for H. influenzae and P. aeruginosa. With this latter organism, the patients were at a late stage of their HIV disease and suffered from repeated exacerbations of chronic bronchitis needing repeated antibiotic treatment. In spite of antibiotic therapy, P. aeruginosa persisted. When performed, serial CT scans showed a course towards bronchiectasis, as seen in patients with cystis fibrosis 40. It should be noted that such bronchiectasis has been particularly well described in HIV-infected children 81.
Decrease in survival time
In a recent case controlled study, Osmond et al. 82 compared the survival time of patients that had recovered from their bacterial pneumonia or PCP and two control subjects matched by CD4 T‐lymphocyte count. On multivariate analysis, the occurrence of the PCP or bacterial pneumonia was a highly significant predictor of shorter survival time 82. The differences between cases and controls were not due to early mortality after initial diagnosis.
There are two main explanations for this shorter survival time 83. 1) It is known that bacterial pneumonia increases local 84, 85 and systemic 86 HIV replication. Consequently, bacterial pneumonia might accelerate the progression of HIV disease. 2) Bacterial pneumonia could be a single marker of more severe immunodepression than is reflected in the CD4 T‐cell count.
In addition, in a recent study, Morris et al. 87 showed that bacterial pneumonia, as well as PCP, resulted in expiratory airflow reductions that persisted after the acute infection resolved. These changes might contribute to prolonged respiratory complaints in HIV-infected patients who have had pneumonia.
Host defence deficiencies incriminated
Abnormalities in host defences caused by HIV infection are probably present early on in the course of the illness and increase in severity with the duration of infection. If the defects in cell-mediated immunity play a central role in opportunistic infections 88, the resulting local and systemic defects in humoral immunity play the major role in bacterial pneumonia 89–93.
Other defects 94, such as nonspecific host defence-like dysfunction of phagocytes 95, 96, abnormal activation of the complement system, alterations in hepatic and splenic clearance, alteration of surfactant and a low mannose-binding lectin serum concentration 97 have also been incriminated.
Human immunodeficiency virus-infected patients receiving prophylaxis for pulmonary infections but not antiretroviral treatment
In the past, this was the situation of the majority of patients in developed countries. Currently, this is the situation of the majority of patients in developing countries.
Changes in the incidence of bacterial pneumonia in the era of prophylaxis.
A first approach to studying these changes is to consider the period around 1990, as, at this time, HIV-infected patients had access to prophylaxis but not to HAART. For example, Lyon et al. 98 compared autopsy findings from before 1984 with those after 1989. During the earlier time period, PCP was more prevalent and was the major cause of death. During the latter time period, as prophylaxis was mainly effective against PCP, bacterial infection, predominantly bronchopneumonia due to S. aureus or P. aeruginosa, became most frequent and the major cause of death. Stein et al. 99 also found that, between January 1988 and July 1990, bacterial infections were the most common cause of death (30%), whereas PCP was responsible for only 16% of deaths. Finally, Afessa et al. 100 showed that, during the period 1985–1996, of all pulmonary complications found at autopsy, only the incidence of PCP decreased.
Another approach to investigating these changes may be through cohort studies. Hoover et al. 101 clearly showed that M. avium-intracellulare disease, oesophageal candidiasis, wasting syndrome and cytomegalovirus disease were more common in HIV-infected patients who had received prophylaxis against P. carinii than in those who had not. Unfortunately, bacterial pneumonia and recurrent bacterial pneumonia were not mentioned.
A last approach is to observe the changes in the causes of respiratory diseases requiring hospitalisation of HIV-infected patients. In the UK, Pitkin et al. 102 noted a decrease in the relative frequency of PCP (68% of admissions in 1986–1987 and 48% in 1990–1991) and an increase in that of bacterial infections (14% of admissions in 1986–1987 and 23% in 1990–1991). During the same period, similar changes were also found in the present author's department, where the relative frequency of PCP was reduced two-fold (40% in 1980–1985 and 24% in 1990–1993) and that of bacterial pneumonia increased three-fold (8% in 1980–1985 and 25% in 1990–1993) (data not shown).
It is very difficult, from these studies, to determine the respective role played by each prophylactic treatment recommended and the first antiretroviral drugs, such as zidovudine, in these changes 103. Consequently, the preventive effectiveness of each prophylactic treatment should be considered separately. In the current review, tobacco cessation, vaccination against H. influenzae b, recommended for HIV-infected children, prophylactic antibiotics in cases of repeated recurrence of severe bacterial LRTI and intravenous immunoglobulin, which may be useful in preventing serious bacterial infections in HIV-infected children, are only mentioned in passing 104, 105.
Effectiveness of pneumococcal vaccine
In the US guidelines 106, pneumococcal vaccine is indicated in all patients with a moderate level of immunosuppression. Indeed, previous studies have clearly shown that some asymptomatic HIV-infected patients and, more particularly, those with a CD4 T‐lymphocyte count of >500 cells·mm−3, developed appropriate antibody responses to the 23-valent pneumococcal vaccine 107–110.
Effectiveness of pneumococcal vaccine in developed countries
Two retrospective case controlled studies and one observational cohort study are available to illustrate the effectiveness of pneumococcal vaccine in developed countries.
In the first retrospective case controlled study, Guerrero et al. 111 found that pneumococcal immunisation was associated with a reduction in the risk of pneumonia of almost 70%. This result was obtained even when immunisation was given with a CD4 T‐cell count of <100 cells·mm−3.
In the second retrospective case controlled study, the pneumococcal vaccine's effectiveness in preventing invasive pneumococcal disease was 50% (95% confidence interval 12–70%) after adjustment for CD4 T‐cell count 112. However, it should be noted that, in the group of HIV-infected patients of African origin, the pneumococcal vaccine was not significantly effective. This failure to demonstrate effectiveness among Blacks might be due to limited power because of limited use of the vaccine in this population, immunisation at more advanced stages of immunosuppression or unmeasured factors.
Finally, in a US observational cohort study among 40,000 HIV-infected persons, pneumococcal vaccine was also effective in preventing pneumococcal disease 113. This effectiveness was strongly related to the level of immunosuppression at the time of vaccination. A significant decrease in incidence was observed only for invasive pneumococcal disease and only in patients in whom pneumococcal vaccination had been performed at a CD4 lymphocyte count of >500 cells·mm−3. In the same cohort, pneumococcal vaccination was associated with slightly improved survival compared with persons who had not received pneumococcal vaccination 103.
Effectiveness of pneumococcal vaccine in developing countries
French et al. 114 evaluated the effectiveness of pneumococcal vaccine in a prospective randomised trial in a cohort of HIV-infected Ugandans. Surprisingly, the vaccine was ineffective in the prevention of invasive as well as noninvasive pneumococcal disease, including those due to the serotypes included in the vaccine. Moreover, the vaccine appeared to increase the risk of developing pneumonia, whatever its cause. This lack of vaccine efficacy might be related to impaired production of capsule-specific immunoglobulin G 115. However, this poor response would not explain the detrimental effect of immunisation. One possible hypothesis is transient upregulation of HIV1 transcription, but there was no association between immunisation and serious infections other than pneumonia and death. Another possible hypothesis is a direct harmful effect of pneumococcal polysaccharides by destruction of polysaccharide-responsive B‐cell clones. This effect might explain in part the high recurrence rate of pneumococcal disease seen in HIV-infected patients 114.
Today, the effectiveness, and even safety, of pneumococcal vaccination in HIV-infected persons in Africa remains debatable.
Effectiveness of anti-infectious drugs recommended for prophylactic treatment of opportunistic infections
Effectiveness of anti-infectious drugs in developed countries
In developed countries, United States Public Health Service/Infectious Diseases Society of America guidelines recommend cotrimoxazole (trimethoprim/sulfamethoxazole) in HIV-infected patients with a CD4 T‐cell count of <200 cells·mm−3 to prevent PCP, and clarithromycin or azithromycin in HIV-infected patients with a CD4 T‐cell count of <50 cells·mm−3 to prevent M. avium complex disease 106. Thus, in some prospective 116 or case controlled 117 studies of PCP or M. avium complex prophylaxis, the use of cotrimoxazole or macrolides was associated with a low incidence of bacterial infection, although with a lesser efficacy if given as secondary prophylaxis 118. Similarly, in the retrospective cohort study of Buskin et al. 119, cotrimoxazole reduced the risk of major infections. Moreover, in this last study, cotrimoxazole also reduced the risk of death not attributable to PCP. Finally, in a recent prospective study, Currier et al. 120 evaluated the indirect effectiveness of anti-infectious drugs given to prevent opportunistic infections in patients with a CD4 lymphocyte count of <200 cells·mm−3 using multivariate analysis. Four regimens were effective at preventing bacterial infections: cotrimoxazole, clarithromycin, a combination of clarithromycin and rifabutin, and, finally, a combination of cotrimoxazole and clarithromycin. Nevertheless, it should be pointed out that, as for pneumococcal vaccine, the effectiveness remained limited.
Effectiveness of anti-infectious drugs in developing countries
Since the effectiveness and safety of pneumococcal vaccine were questionable in Central Africa, it was logical that alternative strategies for the prevention of bacterial infections be considered. Thus, the World Health Organization (WHO) and the Joint United Nations Programme on HIV/AIDS have recently recommended the use of cotrimoxazole prophylaxis for HIV-infected adults in Africa with symptomatic HIV disease (stage 2, 3 or 4 of the WHO classification of HIV infection and disease) and for asymptomatic individuals with a CD4 T‐cell count of <500 cells·mm−3 121. Indeed, the beneficial effect of cotrimoxazole has been evaluated in Abidjan, Ivory Coast, in two recent prospective randomised trials. In the first study, by Anglaret et al. 122, cotrimoxazole was administered to HIV-infected patients when in clinical stage 2 or 3 of the WHO classification. At 12 months, there was a significant decrease in hospital admissions, mainly those due to bacterial pneumonia, malaria and unexplained fever, in patients receiving cotrimoxazole. In the second study, by Wiktor et al. 123, cotrimoxazole was administered to HIV-infected patients treated for tuberculosis with positive smears. There was a significant decrease in hospital admissions, mainly those due to septicaemia or enteritis. Also, as in the study of Buskin et al. 119, there was a significant decrease in mortality, particularly among patients with a CD4 T‐cell count of <350 cells·mm−3.
Nevertheless, two questions remain unresolved. 1) Which patients are likely to benefit from cotrimoxazole? In a recent observational cohort study, Badri et al. 121 showed that cotrimoxazole reduced mortality and the incidence of severe HIV-related illnesses in patients with evidence of advanced immune suppression on clinical (WHO stage 3 and 4) or laboratory (CD4 count <200 cells·mm−3) assessment. In this study, no significant evidence of effectiveness was found in patients with WHO stage 2 disease or a CD4 count of 200–500 cells·mm−3. 2) What is the long-term risk of raising the cotrimoxazole resistance of typical bacteria? A recent study in the San Francisco (CA, USA) area partly answers this last question. In the HIV units in this area, the frequency of cotrimoxazole-resistant bacteria has recently increased. This increase coincided temporally with an increase in the prophylactic use of cotrimoxazole in HIV-infected patients. Finally, the frequency of cotrimoxazole-resistant bacteria has also increased in non-HIV units and cotrimoxazole resistance was associated with multidrug resistance 124.
Effect of prophylaxis on clinical presentation of bacterial pneumonia
In the recent review of Sepkowitz 125 on the effect of prophylactic treatment on recent clinical manifestations of AIDS-related opportunistic infections, no change concerning bacterial pneumonia was reported. The only reported changes were in the relative frequencies of responsible bacteria. Thus, pneumococcal vaccine specifically reduced the incidence of pneumococcal pneumonia. In the same way, in the recent cohort study of Dworkin et al. 126, multivariate analysis showed that prescription of cotrimoxazole was associated with significant protection against Haemophilus spp., staphylococcal infection (invasive or any), but not against pneumococcal or nonpneumococcal Streptococcus, Klebsiella or Pseudomonas spp. infections.
Human immunodeficiency virus-infected patients receiving highly active antiretroviral therapy, with or without prophylaxis of opportunistic infections
This is the situation of the great majority of HIV-infected patients in developed countries who receive HAART and prophylactic treatment of opportunistic infections until a CD4 cell count of >200 cells·mm−3 is reached 127–129.
Effect of highly active antiretroviral therapy on incidence of bacterial pneumonia
In a German cohort, during 1992–1996, Brodt et al. 130 observed a significant decrease in the number of cases of bacterial pneumonia, clearly related to the number of antiretroviral drugs administered. Similarly, in the important US cohort study of Dworkin et al. 113, antiretroviral treatment was an independent factor that contributed to a two-fold decrease in the incidence of pneumococcal disease during 1990–1998. As in the study of Brodt et al. 130, the protective effect increased with the number of antiretroviral drugs and the resulting antiretroviral effectiveness.
This significant reduction was not observed in the outstanding French clinical epidemiology database in which the decreases in both incidence (<20%) and recurrence (<40%) of bacterial pneumonia were relatively low during 1992–1999 26. In this database, this slight variation contrasted with the dramatic decrease observed for opportunistic infections such as PCP, cytomegalovirus infection or toxoplasmosis. Finally, Tumbarello et al. 131 also observed a decrease in the incidence of bacterial pneumonia in the era of HAART, but this decrease was principally observed for nosocomial pneumonia and remained nonsignificant for community-acquired pneumonia.
Indirect approaches might also be considered for the evaluation of these changes.
One approach is to consider the impact of HAART on the characteristics of HIV-infected patients hospitalised during 1995–1997. During this period, Paul et al. 132 observed a two-fold decrease in the incidence of hospital admissions for HIV-infected patients but, in 1997, bacterial pneumonia remained the most common diagnosis at admission.
A second approach is to consider the impact of HAART on the causes of death in the HIV-infected population. Masliah et al. 133 recently analysed the changes in pathological findings at autopsy in AIDS cases over the last 15 yrs. Overall, the frequency of opportunistic infections decreased, whereas that of bacterial infections increased, with the lung remaining the organ most frequently involved. These results were in accordance with those of Afessa et al. 100, who found that, in 1996, bacterial pneumonia remained the most frequent pulmonary complication at autopsy of persons with AIDS.
A last approach is to compare the causes of respiratory disease in HIV-infected patients hospitalised in a chest department before and after 1997, which was the first year in which there was widespread routine HAART use. In the present author's chest department, the absolute number of bacterial LRTIs decreased between the periods 1993–1996 and 1997–1999, but its relative frequency was similar (data not shown).
Effect of highly active antiretroviral therapy on clinical presentation of bacterial pneumonia
Currently, it is clear that rapid restoration of cellular immunity due to HAART may induce a transitory increase in the symptoms and signs of opportunistic pneumonia 134, such as tuberculosis 135 or PCP 136.
To the best of the present author's knowledge, this transitory exacerbation has not been reported for bacterial pneumonia. The main change observed in the French epidemiological database is that, in 1999, bacterial pneumonia occurred at a lesser degree of immunosuppression than in 1992. In 1992, >50% of bacterial pneumonia occurred in patients with <50 CD4 T‐cells·mm−3. This percentage fell to <20% in 1999 26.
One expected consequence of this occurrence at a lesser degree of immunosuppression was a change in the responsible pathogens. In the present author's chest department, the relative frequencies of both S. pneumoniae and H. influenzae increased in the period 1996–1999 compared with the period 1993–1996. In contrast, no pneumonia or bronchiectasis due to P. aeruginosa has been observed since 1996 (data not shown). Thus, Domingo et al. 40 reported the definite cure of P. aeruginosa LRTI in patients receiving effective HAART.
The way forward
Even though the incidence of opportunistic LRTI is decreasing during the current era of HAART, bacterial pneumonia remains, in 2002, the most frequent respiratory disease in all subgroups of HIV-infected patients: 1) those without knowledge of their HIV-seropositivity, in whom typical bacterial pneumonia may reveal the HIV infection 69; 2) those without follow-up or noncompliant with treatment 137, in whom typical as well as opportunistic bacterial pneumonia may be observed; 3) those receiving effective antiretroviral treatment, in whom bacterial pneumonia remains the major cause of respiratory disorders, as previously discussed; and 4) those receiving ineffective antiretroviral treatment (because of acquired resistance to HIV), in whom typical, opportunistic and nosocomial bacterial pneumonia still occur.
If it is borne in mind that bacterial infection remains the leading cause of death at the pre-acquired immune deficiency syndrome stage 138 and the most frequent terminal event in human immunodeficiency virus-infected patients 100, it is clear that their prevention remains a major goal. Thus, pneumococcal vaccination might be given to the patient on highly active antiretroviral therapy with a CD4 cell count reaching 200 cells·mm−3.
The authors thank P. Yvernault and M. Dabancourt for their technical assistance.
- Received February 5, 2002.
- Accepted March 18, 2002.
- acquired immune deficiency syndrome
- bacterial bronchitis
- bacterial pneumonia
- human immunodeficiency virus infection
- © ERS Journals Ltd