Pneumocystis carinii pneumonia: a review of current issues in diagnosis and management.

Abstract

Before the advent of antiretroviral therapy and prophylaxis against Pneumocystis carinii, pneumonia due to this organism eventually occurred in 60–80% of HIV-infected adults in America and Western Europe[1]. Longitudinal studies have recorded a dramatic decline in opportunistic infections including Pneumocystis carinii pneumonia (PCP) since the late 1980s[2, 3]. This has been attributed to effective antiretroviral therapy, increasing use of anti-PCP prophylaxis and improved awareness of the infection amongst health care professionals. Nevertheless, PCP remains a significant cause of morbidity and mortality in the highly active antiretroviral therapy (HAART) era, occurring predominantly in those who are undiagnosed or who are noncompliant with preventative medication. A review of 15 cases of bronchoscopically proven PCP admitted to the Royal Free Hospital between January 1998 and August 2000 revealed that seven patients were unaware of their HIV serostatus. For a further four patients, who had a mean CD4 count of 37 cells/μL (range 2–126 cells/μL), no antipneumocystis prophylaxis was being taken. This review will cover the diagnosis, treatment and prophylaxis of PCP as well as addressing some of the current advances in understanding the pathogenesis and transmissibility of PCP. Molecular biological techniques have shown that P. carinii has closer sequence homology to a fungus than to a protozoan[4, 5]. More recently, similar techniques have been used to study the natural history of this infection and have also raised intriguing questions about its transmissibility and pathogenesis. Infection with P. carinii is generally acquired in early life, with 75% of children in the USA and Europe acquiring antibodies to P. carinii by 4 years of age[6]. Infection with P. carinii is not confined to the West, however, as studies have indicated similar high rates of seroprevalence in adults in Kenya, Zaire, South Africa, Mexico and Korea, as in the USA[7]. Whilst PCP has been a common AIDS-defining opportunistic infection in the west, many studies have recorded lower frequencies of this infection in HIV-infected adults (but not children) in Africa[8]. The reasons for this may include such factors as early death from tuberculosis before a sufficient drop in CD4 count is reached and underdiagnosis of PCP due to lack of diagnostic facilities. Another possibility is that strains of P. carinii found in Africa may be less pathogenic. Conventionally, it has been thought that PCP is due to reactivation of latent organisms in the lung. However, increasing evidence suggests that P. carinii may be cleared in the immunocompetent host and that re-infection may be an important cause of PCP in the immunocompromised[9]. Studies examining mitochondrial subunit ribosomal (r)RNA genotypic differences from bronchoalveolar lavages (BAL) of patients with separate episodes of PCP showed that different sequences were found in half, suggesting re-infection with a new strain of P. carinii[10, 11]. Recently, the issue of transmissibility has been addressed by examining the genotypes of the internal transcribed spacer (ITS) regions of P. carinii from several clusters of PCP occurring among patients with haematological malignancies and HIV infection. Nine different ITS sequences were detected among eight haematological patients, making nosocomial infection unlikely. In contrast, a common ITS sequence was found in two haematology patients who had shared a hospital room and two HIV-infected patients who had had close contact on the ward[12]. Nevertheless, different ITS sequences were detected from an HIV-infected couple that shared a household, suggesting that other factors such as variable infectivity of different P. carinii strains, or host factors may be equally important in determining the transmission of PCP. A notable recent report has described infection of an immunocompetent health worker from a patient with PCP[13]. The issue of the transmissibility of P. carinii is made potentially more serious by the discovery of dihydropteroate synthetase gene mutations that confer resistance against cotrimoxazole and are associated with a poorer outcome from episodes of PCP[14]. Thus, the possibility of the spread of cotrimoxazole-resistant strains of P. carinii exists, but its significance remains to be determined. In the light of these findings, prudent clinical practice may therefore be to limit the contact of new cases of PCP with other immunocompromised patients during the early phase of treatment. It has been demonstrated that P. carinii DNA sequences can be detected from the air of half the rooms of patients with PCP as compared with a quarter of rooms from patients without the infection[15]. Therefore, placing infected individuals in a side room may be desirable, although currently respiratory isolation is not deemed to be necessary. The fact that different P. carinii genotypes exist has led to speculation that some genotypes may be more pathogenic or more transmissible than others[16]. These authors compared the severity of episodes of PCP with the P. carinii genotype as determined by the ITS sequence. They found a preponderance of one genotype in the moderate/severe PCP group and another in the mild group. These intriguing studies raise the possibility that genotypic analysis of PCP could be used as an adjunct to clinical decision making in the management of PCP. Whilst different genotypes of P. carinii may be important in the pathogenesis of PCP, host factors such as the CD4 lymphocyte count are known to be significantly associated with morbidity and mortality. In one study of AIDS patients with PCP who required mechanical ventilation, the mortality rate was 100% in the subgroup of patients with CD4 counts less than 10 cells/μL, but only 25% in those whose CD4 count was greater than 100 cells/μL[17]. However, the blood CD4/CD8 counts may not adequately reflect the lymphocyte subsets found in the relevant site of pathology, the lung. Data from this institution have demonstrated a more profound lack of CD4 T lymphocytes and a greater predominance of CD8 T lymphocytes in the BAL fluid of HIV-infected patients beyond that demonstrated in the peripheral blood (unpublished data). The demonstration of a decrease in mortality in severe cases of PCP by the concomitant use of corticosteroids[18] implies an immunopathological component to the disease. It is therefore interesting to note that murine studies have implicated CD8 T lymphocytes as an important component of this pathogenesis[19]. These authors demonstrated that depletion of CD4 and CD8 T lymphocytes in mice resulted in infection, but not disease, whereas depletion of CD4 T lymphocytes alone resulted in severe pneumonitis, implying that the CD8 lymphocytes were important in PCP pathogenesis, or that CD4 T-cell help was required for adequate anti-PCP CD8 T-cell function. It is intriguing to consider whether in humans, BAL CD8 lymphocytosis is an adverse prognostic factor in PCP. The pathogenetic mechanisms in human infection may include the release of cytokines from BAL CD8 lymphoytes such as tumour necrosis factor-alpha and interferon-gamma that orchestrate further damage to the alveolar space in which P. carinii proliferates as an extracellular parasite. Other investigators have noted that BAL neutrophilia is associated with increased severity of PCP[20, 21] and that the neutrophil chemotactic factor IL-8 is raised in the BAL of patients with PCP[22, 23]. It is not clear from these studies what role the neutrophils played in the pathogenesis of PCP. Pathological evidence of increased severity of diffuse alveolar damage was not found on transbronchial biopsies of patients with PCP and BAL neutrophilia in one study[20]. It seems likely that questions regarding the pathogenesis of PCP will soon be answered by the application of immunological techniques such as antigen-specific analysis of BAL lymphocytes and the determination of intracellular cytokine release from lymphocytes, macrophages and neutrophils in the BAL of patients with PCP. The probability of developing PCP in individuals with HIV infection rises dramatically as the CD4 count drops below 200 cells/μL[24]. The clinical features of PCP typically involve the onset of a dry cough and breathlessness with or without fever over a period of 1 to 3 weeks. Physical findings include tachypnoea, tachycardia and occasionally cyanosis, but auscultation of the chest is generally unremarkable, although PCP can mimic asthma[25]. The discovery of oxygen desaturation on exercise in the context of known or suspected HIV infection should greatly raise the suspicion of PCP, even in the context of a normal chest radiograph[26]. The classic radiological features of PCP are fine, bilateral, perihilar interstitial shadowing. However, the chest radiograph may be normal[27], reveal pneumothorax[28], upper lobe shadowing[29], or even rarely solitary nodules[30]. The use of inhaled pentamidine as prophylaxis against PCP has been particularly associated with atypical radiological features of PCP such as cavities, upper zone changes and even pleural effusion[31], probably reflecting the patchy deposition of this drug within the lung. In one study comparing the radiographic findings with clinical severity, a normal chest radiograph was associated with less severe PCP and a low mortality rate[32]. High resolution computerized tomography (CT) is more sensitive than chest radiography in the diagnosis of PCP with the typical findings being mosaic pattern ground-glass shadowing. One small study correlated high resolution CT (HRCT) findings with bronchoscopically proven PCP and found a sensitivity of 100% with a specificity of 89%[33] for HRCT. It has been suggested that HRCT findings are diagnostically reliable enough to obviate the need for invasive tests such as bronchoscopy. However, many clinicians prefer to detect the organism, either by induced sputum, or by BAL, rather than relying on radiological and clinical evidence alone. A study that examined the radiological features of 43 patients following the initiation of anti-PCP therapy found that 36% had early radiological deterioration and in 13% there was no change after 2 weeks of therapy. However, most radiographic abnormalities resolved within 45 days[34]. The gold standard for the diagnosis of PCP is the discovery of P. carinii in the sputum or BAL fluid. Some centres, particularly in the USA, favour the use of induced sputum as an initial diagnostic test for PCP. Whilst this technique is cheaper and noninvasive, it is less sensitive than BAL[35, 36]. The use of monoclonal antibody techniques may improve the sensitivity of induced sputum[37]. Although the induced sputum technique is generally safe, it can lead to reductions in oxygen saturations[38]. The advantages of bronchoscopy include the ability to perform site-directed washings and to increase the chance of detecting co-pathogens in the lung[39]. It has been demonstrated that the yield of P. carinii was increased with washings from the upper lobes, rather than the normally lavaged right middle lobe or lingula and that this was independent of whether inhaled pentamidine was given as PCP prophylaxis[40]. The development of the polymerase chain reaction (PCR) has reopened the question of whether induced sputum may be adequate for the diagnosis of PCP. PCR may be performed on a variety of genetic substrates such as the internal transcribed spacers (ITS) of rRNA, mitochondrial rRNA and the major surface glycoprotein genes. Most studies have confirmed a high sensitivity and variably high specificity of PCR techniques when compared to morphological detection of P. carinii in BAL[41-43]. The sensitivity for PCR detection of PCP on oropharyngeal samples has been lower than for BAL, with results between 50 and 70% in two studies[41, 42]. However, one small study using a nested PCR of mitochondrial rRNA was fully sensitive when compared to BAL. Using a dilutional technique, the authors were also able to demonstrate a reduction in the amount of amplification product detected from sequential oropharyngeal samples after the patients were started on anti-P. carinii therapy[44]. At present, PCR remains predominantly a research tool for diagnostic purposes and it remains to be determined whether it will prove more cost-effective than BAL. However, it may play a role in the diagnosis of patients who are too sick to bronchoscope, to improve detection rates from sputum samples. An alternative role for PCR may be for the surveillance of oropharyngeal samples from patients at high risk of developing PCP, such as those who have had previous episodes and have low CD4 counts. Whether this would be a cost-effective alternative when compared to early conventional investigation of patients who develop symptoms of PCP remains to be determined. Prophylaxis against PCP should be initiated in HIV-infected subjects with CD4 counts that drop below 200 cells/μL or who have had a previous episode of PCP[45]. Following the realization that effective and durable increases in CD4 cell counts could be achieved with HAART, it appears to be safe to stop primary prophylaxis against PCP when the CD4 count has risen to more than 200 cells/μL on HAART[46-49]. These data are summarized in Table 1. Combining all these trials of stopping primary prophylaxis gives a total number of person-years of follow-up (PYFU) of 660, with one documented case of PCP in whom the CD4 count at the time of diagnosis of PCP was only 143 cells/μL[49]. As yet it is not clear how soon after the CD4 count has risen above 200 cells/μL prophylaxis should be stopped. Some workers have cautiously suggested waiting for 6 months[49], although 3 months may be reasonable. The risk of developing PCP is greater in those who have had a previous episode of PCP than in those who have never had one. To date, several studies examining the safety of stopping secondary prophylaxis in patients with CD4 reconstitution to greater than 200 cells/μL after HAART have revealed no episodes of PCP after a combined total of 347 PYFU of follow-up[46, 50-52] (see Table 1). These results strongly suggest that stopping secondary prophylaxis is safe, although the duration of follow-up after stopping prophylaxis in these studies was less than a year and therefore longer follow-up may be required. Despite apparently adequate T-cell reconstitution with HAART, it is important that CD4 counts be watched carefully following discontinuation of prophylaxis, as failure of therapy resulting in a decline in CD4 counts may occur for a variety of reasons, necessitating the reintroduction of prophylaxis. Furthermore, the adoption of structured treatment interruptions as a therapeutic strategy may also result in significant fluctuations of CD4 cells that could re-expose the individuals to the risk of PCP and other opportunistic infections. There have been a few published cases of PCP developing despite an increase in CD4 count beyond 200 cells/μL on HAART[53, 54]. In two cases, the nadir CD4 count before the institution of HAART was very low (13 cells/μL and 5 cells/μL)[54]. It is possible that these two patients had deletion of their P. carinii specific immune repertoires that were not reconstituted on HAART and therefore HAART did not provide immune protection despite a dramatic increase in CD4 counts. Antigen-specific clonal deletion is a recognized complication of severe immunodeficiency due to HIV infection[55] and has accounted for rare cases of cytomegalovirus retinitis despite effective HAART[56]. Various drugs are available for prophylaxis against PCP, including cotrimoxazole, inhaled pentamidine, dapsone and atovaquone. A meta-analysis of trials of PCP prophylaxis revealed that cotimoxazole was significantly superior to inhaled pentamidine[57]. Cotrimoxazole has the added benefit of providing protection against toxoplasmosis and bacterial respiratory tract infections[58] and it is also inexpensive. Dapsone has been shown to be equal to inhaled pentamidine or atovaquone, but less effective than cotrimoxazole for prophylaxis[59]. Atovaquone is better tolerated than dapsone[60]. Although cotrimoxazole is superior as a prophylactic drug, it is associated with a significant incidence of side effects. When compared with inhaled pentamidine, discontinuation because of toxic effects was seven times more frequent with cotrimoxazole and four times more frequent with dapsone[57]. One randomized study comparing high dose cotrimoxazole (960 mg), low dose cotrimoxazole (480 mg) and inhaled pentamidine revealed that both cotrimoxazole groups were equally effective as prophylaxis and significantly better than inhaled pentamidine. Adverse effects occurred earlier in the higher dose than the lower dose cotrimoxazole group[61]. A randomized comparison of thrice weekly and daily cotrimoxazole favoured the daily dosing schedule, with small reductions in the relative risk of developing PCP, but this was associated with more adverse events[62]. Thrice weekly cotrimoxazole should therefore be given to patients who develop side effects on daily dosage in an effort to improve compliance. Hypersensitivity is a relatively common side effect of cotrimoxazole therapy. Several strategies have been developed to overcome hypersensitivity reactions to cotrimoxazole. Desensitization is generally the preferred method rather than re-challenge[63]. The use of a 6-hour cotrimoxazole graded challenge in patients with hypersensitivity reactions was found to be highly effective[64]. A graded challenge with cotrimoxazole has also been found to be successful in patients who had previously had an episode of Stevens–Johnson syndrome associated with the drug[65]. Despite the decreasing incidence of PCP in the HAART era, this infection remains an important cause of morbidity and mortality[17, 66]. Treatment for PCP can be with high dose cotrimoxazole, intravenous pentamidine, clindamycin and primaquine, dapsone and trimethoprim or atovaquone. One randomized trial found that there were no significant differences in the outcome of mild to moderate episodes of PCP treated with high dose cotrimoxazole, dapsone and trimethoprim or clindamycin and primaquine[67]. However, it is important to note that this study excluded those with severe PCP. One large study that compared intravenous cotrimoxazole with intravenous pentamidine for all severities of PCP determined that there was no significant difference between both treatment arms[68]. Importantly, 80% of those started on cotrimoxazole and 69% of those started on pentamidine failed to complete the starting therapy. Reasons for failure to complete initial therapy were due to drug toxicity (34% in the cotrimoxazole group and 25% in the pentamidine group) and failure to respond to therapy (42% cotrimoxazole group and 40% pentamidine group). However, it must be noted that this study was performed before the use of corticosteroids for severe episodes of PCP. Corticosteroids have been shown to reduce the incidence of hypersensitivity reactions to cotrimoxazole[69] and also to increase its effectiveness if hypoxia is present[18]. Other smaller comparisons of cotrimoxazole with pentamidine have determined that both are effective treatments for PCP[70, 71]. Inhaled, rather than intravenous, pentamidine is less effective than cotrimoxazole[72]. Atovaquone has been shown to be an effective alternative to cotrimoxazole in mild–moderate PCP, and one that is associated with a lower incidence of adverse events[73]. Whilst high dose cotrimoxazole for PCP is the initial treatment of choice in many centres, the high incidence of intolerance, in particular hypersensitivity reactions, make this drug less than an ideal first choice agent. A further important clinical issue is that therapeutic failure may occur with cotrimoxazole despite adequate serum concentrations of the drug[74]. Studies have demonstrated that therapeutic failure of PCP treatment with cotrimoxazole may be partly attributable to P. carinii dihydropteroate synthetase gene mutations[14, 75]. These findings emphasize the need to consider changing to an alternative anti- P. carinii regime in the context of clinical deterioration in a patient with PCP. In the event of an adverse drug reaction, or failure to respond to cotrimoxazole, then the second-line agent should be intravenous pentamidine for episodes of moderate–severe PCP. An alternative to cotrimoxazole is trimetrexate, a potent dihydrofolate reductase inhibitor[76] that should be given with the folate compound leucovorin to protect mammalian cells. Trimetrexate–leucovorin has been shown to be an effective salvage therapy in those who fail to respond to cotrimoxazole[77], although a comparison of cotrimoxazole with trimetrexate–leucovorin revealed the latter drug to be inferior, albeit better tolerated[78]. A further salvage drug that has received little attention in the literature is eflornithine[79]. Drug treatment strategies are detailed in Tables 2 and 3. One issue in the treatment of PCP is that sulpha-containing regimes such as cotrimoxazole, trimethoprim/dapsone and primaquine/clindamycin are contra-indicated in patients with G6PD deficiency. Despite these recommendations, we suspect that patients are rarely tested for G6PD deficiency before the initiation of sulpha-containing anti-P. carinii regimes, and it remains to be determined as to whether significant haemolysis occurs in G6PD deficient patients treated for PCP with such drugs. Indeed, cotrimoxazole prophylaxis appears to be well tolerated by HIV-infected patients in Africa where G6PD deficiency is common[80, 81]. Nevertheless, it has been suggested that atovaquone should be considered as a therapy in G6PD deficient patients[82]. In HIV-infected patients with moderate–severe PCP, classified by resting arterial oxygen tensions of less than 9.3 kPa on breathing air, corticosteroids should be given. Adjunctive corticosteroids have been shown to reduce death and respiratory distress associated with moderate–severe PCP[18]. There is no benefit from corticosteroid usage in episodes of mild PCP[83]. Late presentation of PCP is often associated with respiratory failure, necessitating ventilation. Several studies have demonstrated a poor outcome from the use of mechanical ventilation[84, 85], but more recently it has been suggested that the outcome from mechanical ventilation is not as poor as initially thought[66]. In this study, factors that were associated with raised mortality 3 months after intensive care unit admission included delay in starting mechanical ventilation of more than 3 days, a duration of mechanical ventilation of ≥ 5 days, pneumothorax and nosocomial infection. Therefore, whilst noninvasive ventilation such as continuous positive airways pressure ventilation (CPAP) has been shown to have a generally good outcome for PCP[86-88], patients that require mechanical ventilation should not have this treatment option delayed. PCP remains a serious opportunistic infection occurring predominantly in those who are undiagnosed, or noncompliant with prophylactic medication. Advances in the understanding of the molecular biology of P. carinii have raised interesting questions regarding the possibility that certain strains may be more pathogenic than others, and that cotrimoxazole resistant strains may spread between individuals. The clinical relevance of these findings remains to be determined, but it is suggested here that new cases of PCP should be managed in a side room during the early phase of treatment. The plain radiographic manifestations of PCP are protean and a normal chest radiograph is not incompatible with the diagnosis. CT scanning is far more sensitive as a diagnostic tool. We favour BAL over induced sputum as a diagnostic test, due to its increased sensitivity, potential for site-directed washings and detection of co-pathogens in the lung, whilst also recognizing the value of induced sputum in those who are too sick to bronchoscope. The value and cost-effectiveness of PCR techniques as a diagnostic strategy have yet to be determined. Evidence suggests that daily cotrimoxazole (480 mg or 960 mg) is the optimum prophylactic regime for HIV positive individuals with CD4 counts less than 200 cells/μL. Primary and probably secondary prophylaxis may be safely stopped when the CD4 counts rise consistently above 200 cells/μL on HAART. Prophylaxis should not be stopped until this rise has been sustained for 3–6 months after the introduction of successful HAART. The treatment of choice for established PCP is high dose cotrimoxazole. Although this therapy is associated with a high incidence of adverse reactions, skin hypersensitivity reactions may be reduced by the concomitant use of corticosteroids which should also be given if hypoxia is present. Adverse reactions to cotrimoxazole may be successfully treated by the use of a desensitization regime. Alternatives to cotrimoxazole include pentamidine for severe disease, and atovaquone or clindamycin/primaquine or dapsone/trimethoprim for mild-to-moderate disease. Patients with severe PCP who fail to respond to cotrimoxazole should be switched to intravenous pentamidine as the first choice salvage therapy and trimetrexate/leucovorin as the second choice.