Comprehensive, up-to-date information on HIV/AIDS treatment, prevention, and policy from the University of California San Francisco

Complete blood count and sedimentation rate show no characteristic pattern in patients with PCP. Serum chemistries are not particularly helpful; however, the serum LDH concentration is frequently increased.(49,84-87) Results of a study of 62 patients (54 with advanced HIV disease) showed that only 7% of those with documented PCP had normal serum LDH levels; the mean LDH level in patients with PCP was 362 IU. The mean initial level in surviving patients was 340 IU, vs 447 IU for nonsurvivors. With treatment, 75% of the survivors showed a consistent decrease in the LDH level; 19% showed an increase before the levels eventually decreased. In contrast, 75% of the nonsurvivors showed increasing levels before death.(88) In one report of 78 patients with advanced HIV disease and PCP, a mean initial LDH level of 355 IU was found in patients with a good outcome (survival without respiratory failure) vs 710 IU in patients with a poor outcome (respiratory failure or death).(89) All survivors showed a decline in LDH level after 1 week of anti-Pneumocystis therapy, whereas nonsurvivors showed progressive increases. Therefore, although serum LDH concentration is a nonspecific indicator of lung parenchymal damage and thus is of limited diagnostic use, it may serve as a prognostic indicator in PCP.

Arterial blood gas measurements in PCP generally show increases in alveolar-arterial oxygen gradient, P(A-a)O₂, although arterial oxygen pressure (PaO₂) values vary widely depending on the severity of the disease. Up to 25% of patients with PCP may have a PaO₂ of 80 mm Hg or above while breathing room air.(47) Similar to the serum LDH concentration, the PaO₂ and P(A-a)O₂ correlate with prognosis. A P(A-a)O₂ of >29 mm Hg or a PaO₂ of <71 mm Hg is associated with poorer outcome,(46-48,90) and provides an indication for adjunctive corticosteroid therapy, as discussed below.

Serologic testing for P jiroveci has been useful in epidemiologic studies showing that infection with the organism is widely prevalent in the general population. However, serology has not shown utility as a diagnostic test for PCP. In an examination of patients with PCP using acute and convalescent sera and a "normal" titer up to 1:32, 18 of 21 PCP patients were found to undergo seroconversion (4-fold or greater rise in P jiroveci antibody titer). Eight of these patients, however, had a convalescent titer less than the "normal" titer of 1:32, and 56% of control subjects had titers up to 1:32.(91) Data from other centers show similar results, with inconsistent increases in both immunoglobulin (Ig) M and G titers during acute PCP disease.(92,93) A study of the usefulness of an enzyme-linked immunosorbent assay (ELISA) for IgG antibody to P jiroveci and a latex agglutination test for P jiroveci antigen demonstrated that, although the mean IgG antibody titers of patients with and without acute PCP were not statistically different, antigen titers were both sensitive and specific in identifying patients with acute PCP. Antigen testing thus shows potential as an adjunct to other tests in current use, but requires further evaluation before it can be recommended as a routine diagnostic test.

Recent advances in molecular typing that may be helpful in epidemiologic analyses include direct analysis of various genes and ribosomal subunits of P jiroveci, but this technique is not available for routine diagnostic purposes.(94)

The classic appearance of PCP on chest radiography in patients with AIDS is a diffuse interstitial infiltrate, although all of the following presentations have occurred: abscesses, cavitation or cystic lesions, lobar consolidation, nodular lesions, effusions, pneumothorax, pneumomediastinum, and a normal chest radiograph (see Figure 1,Figure 2,Figure 3,Figure 4,Figure 5,Figure 6).(97,45,98-110)

Rare radiographic presentations include patchy upper lobe consolidation imitating tuberculosis, a miliary pattern, and mediastinal and hilar enlargement.(102,111-113) In one report, PCP occurred as an endobronchial mass in a patient with concomitant tuberculosis.(114) Another group of investigators described local patchy interstitial infiltrates, including unilateral disease.(102)

Use of high-resolution computed tomography (HRCT)(see Figure 7), also referred to as thin-section CT, may be helpful for those patients who have normal chest radiographic findings at initial presentation.(115) However, it is clear that some patients with PCP also will have false-negative HRCT findings.(116) In a small study of 14 patients with PCP, including 3 patients with AIDS, investigators were unable to show significant differences in HRCT when compared with conventional CT findings.(117) Although there are some studies that favorably compare HRCT with chest x-rays,(118) rigorous studies to assess the sensitivity and specificity of HRCT for PCP have not been completed. This may be due to the fact that relatively few cases of PCP are now seen in the United States.

Prophylaxis for PCP with AP often changes the presentation of acute PCP. Patients who develop PCP while using AP are more likely to have predominantly upper lobe infiltrates,(72,73,104,119-121) although this pattern also occurs in patients who have not received AP prophylaxis.(122) Approximately 60% of patients with PCP recurrence despite receiving AP for secondary prophylaxis presented with infiltrates in a predominantly upper lobe distribution, possibly as a result of poor deposition of the aerosol into the upper lobes (as measured by radionuclide studies).(123) AP also may be cleared more rapidly from the upper lung fields, resulting in less protection in those areas.

Pleural effusions are uncommon in patients with HIV disease and PCP as the only opportunistic infection. Such effusions, when present, particularly if they are large, should raise suspicions of pulmonary KS, lymphoma, tuberculosis, or other bacterial or fungal infections.(45) Chest radiographs are normal in approximately 5-10% of patients with PCP,(95,124) and, in some small series, up to 25% of patients had normal radiographic findings.(91)

Pneumatoceles (also called cavities, cysts, blebs, or bullae) and spontaneous pneumothoraces in patients with PCP were uncommon in the early stage of the HIV epidemic, but became more commonly recognized as part of the acute presentation of PCP during the course of PCP treatment or as residual findings after successful completion of treatment.(72,73,102,125-136) Clinicians have reported cases of recurrent pneumothoraces, which may require mechanical intervention with pleurodesis, surgical stapling, or resection of bullae.

In a study of 100 consecutive patients with PCP at San Francisco General Hospital, 10 had pneumatoceles with no demonstrated predilection for particular lobe of the lung.(137,138) In 3 of these 10 patients, the pneumatoceles were associated with spontaneous pneumothoraces. Most HIV-infected patients who present with pneumatoceles have had prior episodes of PCP and many have been treated prophylactically with AP. Several authors have noted that the majority of the lesions have been in the apical or peripheral areas of the lung, suggesting a predilection for those areas less accessible to AP.(72,73,130-133,136,139,140)

In a study of 408 patients receiving AP for prophylaxis, however, only 21 had pneumothoraces. All but 1 of these 21 patients had a prior or concomitant episode of PCP. Patients with multiple episodes of PCP were at greater risk for pneumothoraces.(141,142) For patients without other predisposing factors (eg, mechanical ventilation, transbronchial biopsy, transthoracic needle puncture), spontaneous pneumothorax occurs in 2-6% of those with PCP.(143-146)

AP may be a precipitating or contributing factor in the etiology of pneumothorax, yet at least one third of all patients with PCP and pneumothorax have never been treated with AP. The causes of the pneumothoraces probably are multifactorial, and the contribution of each factor is difficult to assess. Possible mechanisms leading to pneumatocele formation and pneumothorax include chronic low-grade infection with PCP, erosion of the parenchymal tissue secondary to the release of protease or elastase or both, and alteration in the lung connective tissue by HIV infection itself.(129-131,144,145)

The radiographic appearance of PCP commonly worsens over the first few days of therapy; in more severe cases, early PCP may progress to air space consolidation. However, deterioration continuing beyond 7-10 days of treatment for PCP likely represents a failure of therapy, and change of treatment regimen should be considered.(48,147)

See also "Radiographic Assessment" chapter

A diffuse interstitial pattern can occur in other infections common in advanced HIV disease, including cytomegalovirus (CMV), coccidioidomycosis, histoplasmosis, tuberculosis, and Mycobacterium avium complex infections.(45) Interstitial reticular and reticulonodular infiltrates also can be present in patients with noninfectious complications (such as lymphoid interstitial pneumonitis).(45) In 2 studies, investigators found interstitial pneumonitis without a specific etiology in 32% of patients.(148,149) Those cases of pneumonitis were clinically indistinguishable from cases of PCP, although the radiographic abnormalities were generally less severe (50% of the patients with nonspecific pneumonitis had normal chest radiographic findings), and histologic study showed less alveolar damage in those patients than in patients with PCP.

The least invasive means of definitively diagnosing PCP is the examination of sputum induced by inhalation of hypertonic saline solution. The procedure involves inhalation of a 3-5% saline mist generated by an ultrasonic nebulizer and collection of the subsequent expectorated sputum. Early results from 2 centers showed that sputum induction had a sensitivity of 55%, enabling 1 center to reduce the need for bronchoscopy by 44%.(173-175) However, the negative predictive value of sputum induction was only 39%. Since then, further experience at San Francisco General Hospital has shown the sensitivity of induced sputum examination for PCP to be 74-77% and the negative predictive value to be 58-64%.(176) One report described a concentration technique that raises the sensitivity to 78% with a concomitant negative predictive value of 71%.(177) However, because of the relatively low negative predictive value of sputum induction, confirming a diagnosis of PCP in a patient whose sputum tests negative requires an invasive diagnostic procedure.

Advantages of sputum induction compared with bronchoscopic methods include lower cost, lower morbidity, and less patient discomfort. The main disadvantage of sputum induction is relatively low sensitivity in patients with a lower organism load who may expectorate fewer organisms. Moreover, evaluating the slides requires more time and more experienced laboratory personnel than does evaluating specimens obtained by bronchoscopy.

Before the HIV epidemic, the diagnosis of PCP typically was made by open lung biopsy.(178,179) With the large number of cases seen in association with HIV, transbronchial biopsy (TBBx) and BAL performed with the fiberoptic bronchoscope have become the "gold standard" by which other diagnostic procedures are evaluated. The reported sensitivity of TBBx for PCP ranges from 66% to 98% using both fixed tissue and touch preparations; the sensitivity of touch preparations alone has ranged from 74% to 88%.(180-184) The sensitivity of BAL ranges from 55% to 98%, with most investigators reporting sensitivities around 90%.(185,186) AP prophylaxis may decrease the sensitivity of BAL for diagnosing PCP.(187)

To perform BAL, the bronchoscope is wedged into a distal bronchus and 100-120 mL of saline solution in 20-mL aliquots is instilled. After each bolus, the solution is aspirated to retrieve a total of 40-60 mL. One study found that a BAL cell block (made from the sediment of the aspirated material) yields a slightly higher sensitivity than BAL fluid (95% vs 88%), and that bronchial brushings are the least sensitive of all the bronchoscopic specimens (78%).(188) Combined BAL and TBBx have a sensitivity that approaches 100%.

A 1995 review of 1,716 cases of suspected PCP in 1,259 patients at San Francisco General Hospital showed that induced sputum was positive in 800 cases (46.6%).(189) Of the remaining 916 patients, 314 (34.3%) did not undergo evaluation with bronchoscopy. BAL was performed in the other 602 (65.7%), including 50 TBBx procedures, resulting in the diagnosis of PCP in 31.1%, other diagnoses in 18.6%, and no etiologic agent found in 50.3% of the bronchoscopic procedures. Of the 50 initial bronchoscopies in which TBBx was performed, 8 provided a diagnosis of PCP: 3 by BAL specimen only, 2 by TBBx specimen only, and 3 by both. Of the 303 cases in which no diagnosis was made, 45 repeat bronchoscopies with both BAL and TBBx were done, resulting in an additional 5 patients (11.1%) found to have PCP. Thus, of the 192 cases requiring bronchoscopy to diagnose PCP, BAL was sufficient to detect the organisms in 190 (99%); however, 5 patients did undergo second bronchoscopies with both BAL and TBBx (although PCP was found in the second BAL specimen in all 5 instances). Thus, in this center, where induced sputum has a good yield, bronchoscopy is still required to secure a diagnosis in up to one third of those patients screened by induced sputum.

Bronchoscopic procedures, although generally safe, do carry a small risk of complications. In one review, approximately 9% of patients undergoing TBBx developed a pneumothorax, and 5% required placement of a chest tube.(181) An obvious advantage of BAL is that it avoids this complication entirely and may be performed in patients with bleeding diatheses.

One report described nonbronchoscopic BAL performed with a control-tipped reusable catheter.(190) This technique is less costly and appears promising, but more experience is needed before its full value can be assessed.

Bronchoscopy procedures also entail risks to the operator for potential exposures to hepatitis B, HIV (via mucous membrane exposure to blood), and tuberculosis (via aerosol transmission), and to subsequent patients (via incompletely sterilized equipment). Appropriate use of personal protective equipment and quality control measures are necessary to minimize the risk of nosocomial infection with these agents.

Because bronchoscopic procedures and sputum examination are usually diagnostic, few additional diagnoses of PCP can be obtained by open lung biopsy,(191) which carries significant morbidity risk. Open lung biopsy should therefore be reserved for progressive pulmonary disease in which the less invasive procedures are nondiagnostic or contraindicated.

TMP-SMX is most often used as initial therapy for PCP. Initially, the standard dosage used in patients with advanced HIV disease was 20 mg/kg/day of TMP plus 100 mg/kg/day of SMX given either orally or intravenously (divided every 6 or 8 hours daily). Subsequent reports described similar efficacy and a lower frequency of adverse reactions with a dosage of 15 mg/kg/day of TMP and 75 mg/kg/day of SMX for 14-21 days.(200,201) This has become the standard dosing regimen for acute PCP in adults.

For unknown reasons, the rate of adverse reactions to TMP-SMX is higher in HIV-infected patients than in non-HIV-infected patients.(196) Because of the high incidence and severity of adverse effects, only 45-50% of patients who start therapy with TMP-SMX are able to complete a 21-day course.(202-207)

One report described adverse effects in all patients treated with TMP-SMX, including rash (33%), elevation of liver enzyme levels (44%), nausea and vomiting (50%), anemia (40%), creatinine elevation (33%), and hyponatremia (94%).(204) Common adverse reactions that necessitated a change in therapy included neutropenia (15%) and severe rash (15%). These occurred after a mean of 11.5 days of treatment (range, 6-18 days). Two other reports described a lower rate of overall adverse effects, but a larger number of patients requiring a change of therapy for neutropenia (28%) and rash (33%).(202,203) Drug-induced neutropenia generally does not occur before 1 week of therapy and usually responds to discontinuation of the medication. Fortunately, neutropenia induced by TMP-SMX does not appear to be further affected by subsequently administered pentamidine, and vice versa.(208) Less common adverse reactions to TMP-SMX include tremor, ataxia, aseptic eosinophilic meningitis, and renal tubular acidosis with or without sodium and potassium disturbances.(209-213)

In one study, patients received TMP-SMX at a lower dosage of 15 mg of TMP/kg/day and 75 mg of SMX/kg/day, and the dosages were tailored to maintain a TMP serum level of 5-8 mg/mL.(200) Despite the lower initial dose, 25 of 36 patients required further reductions to keep within the prescribed TMP serum level (average final dosage, 12 mg/kg/day). There was an 86% survival rate, but adverse reactions were common (rash, 44%; anemia, 39%; neutropenia, 72%; elevations in liver function tests, 22%). Neutrophil counts rose when the TMP-SMX dosage was decreased. One report noted that lowering the SMX dose did not affect the frequency of adverse reactions in 11 patients, implying that TMP is more commonly the cause of the reactions.(201)

In a study of cross-reactions to various sulfa medications, 3 patients with advanced HIV disease and previously documented allergic reactions to TMP-SMX during PCP therapy were rechallenged during a subsequent episode of PCP and were able to complete 21-day courses, although 2 of the patients had the same adverse reactions (rash and fever) during the subsequent course.(214) A variety of investigators have tried various regimens to desensitize patients to TMP-SMX with variable success. The proposed schedules have ranged from rapid reintroduction of TMP-SMX over a 2-day period to regimens lasting up to 42 days.(215,216) Some investigators have even used direct rechallenge in patients who had prior adverse reactions but had not been exposed to TMP-SMX recently. However, most of the studies have the following drawbacks: short follow-up periods that evaluate success based on achieving a full-strength TMP-SMX dosage without subsequent evaluation, small sample sizes, or retrospective design with potential for bias.(217-222) (See "Prophylaxis" section below and Table 3 for more information on TMP-SMX desensitization.)

Pentamidine is the most common alternative drug for treating PCP in patients who have adverse reactions or fail to respond to TMP-SMX. Early in the HIV epidemic, the usual dosage of pentamidine was 4 mg/kg/day given intravenously once daily. A dosage of 3 mg/kg/day was also effective in several studies (see below) and is now the widely accepted starting dosage with a usual duration of therapy of 14-21 days. Approximately 45% of patients receiving pentamidine require a change to another agent because of adverse effects.(202,204,223) As with TMP-SMX, pentamidine fails as initial treatment in up to 33% of patients.(196) Contrary to findings reported in 1986,(204) a 1988 study (200) found that patients begun on pentamidine had a lower survival rate than those started on TMP-SMX (61% vs 86%); this was true for both first and repeat episodes of PCP.

The 1986 study reported adverse effects in all patients treated with pentamidine, including anemia (33%), creatinine elevation (60%), elevation in liver function tests (63%), and hyponatremia (56%).(204) The most common adverse effect requiring a change in therapy was neutropenia (32%). Adverse effects requiring a change in therapy occurred after a mean of 10.4 days of therapy (range, 6-16 days). Other investigators reported similar adverse effects but at somewhat different rates.(200,202,203) Uncommon reactions included renal failure associated with myoglobinemia and myoglobinuria. Reports also described elevations in serum potassium and creatinine kinase levels.(224-226) As the use of pentamidine has become more widespread, reports more frequently describe cardiac toxicity, including bradycardia, prolongation of the QT interval, and ventricular arrhythmia (including torsades de pointes), occasionally resulting in death.(208,216,227,228)

Pancreatitis, hypoglycemia, and hyperglycemia are common adverse effects of pentamidine. A review of patients with advanced HIV disease for whom the CDC dispensed pentamidine early in the epidemic noted that 57% of all patients became hypoglycemic or hyperglycemic.(229) There have been at least 3 cases of fatal pancreatitis attributed to pentamidine in which the autopsy revealed extensive pancreatic necrosis.(230,231) Prior studies reported that pentamidine directly affects the pancreatic islet cells, resulting in both hyperglycemia and hypoglycemia. The risk of hypoglycemia increases with duration and dosage of therapy,(229) although it may occur precipitously early in the course of therapy or after completion of therapy. Several cases of fatal hypoglycemia with pentamidine have been reported.(200,229) Hyperglycemia may not become evident until several months after treatment is completed; diabetes has occurred as late as 150 days after therapy.(229)

Parenteral pentamidine is distributed widely in the body, particularly in the kidneys, adrenal glands, spleen, and liver.(223,232,233) Lung concentrations are lower, and at a dosing interval of 24 hours, the drug is not detectable in the lung until the fourth dose, that is, until patients have received approximately 1 g.(233,234) Based on animal data, one investigator suggested that treatment of PCP requires a concentration of 30 mg/g of lung tissue. Parenteral administration (in humans) requires 5 doses to achieve this level.(233) One report described peak plasma concentrations of pentamidine after intramuscular and intravenous administration as 209 and 612 ng/mL, respectively. The plasma half-life of pentamidine after intravenous administration is approximately 6.5 hours.(235,236) Dosage adjustments are not required in patients with creatinine clearance rates above 35 mL/min. Neither peritoneal dialysis nor hemodialysis appears to alter plasma concentrations. Use of pentamidine in pregnancy is not well studied, but the drug clearly crosses the placenta. In addition, pregnancy may alter the usual pharmacokinetics of pentamidine, raising the levels above those predicted by pharmacokinetic modeling.

Oral TMP-dapsone is a widely used outpatient therapy for mild-to-moderate PCP. Most studies have used a regimen consisting of TMP (20 mg/kg/day) orally in 4 divided doses plus dapsone (100 mg/day). A 60-patient, double-blind, randomized trial at San Francisco General Hospital compared TMP-dapsone (TMP 15 mg/kg/day plus dapsone sulfate 100 mg/day) with TMP-SMX and confirmed that the efficacy was equal to that of TMP-SMX (2 and 3 failures, respectively); there was only 1 death (of a patient in the TMP-SMX treatment arm).(237) Fifty-seven of the 60 patients experienced some adverse effect, with 9 (30%) and 17 (57%) patients in the TMP-dapsone and TMP-SMX arms, respectively, requiring a change of therapy. In the TMP-SMX group, liver enzyme elevation and neutropenia were more frequent and more severe. Asymptomatic methemoglobinemia occurred in 20 of 30 patients treated with TMP-dapsone, but only 1 patient required discontinuation of therapy (due to a methemoglobin level of 21% at day 9 of therapy). The methemoglobinemia responded promptly to administration of methylene blue. Asymptomatic hyperkalemia was significantly more frequent in the TMP-dapsone group. The mean time to onset of adverse reactions was 10.3 days for the TMP-dapsone group and 12.5 days for the TMP-SMX group. Therapy continued after occurrence of a rash unless there was desquamation or mucous membrane involvement.

Because dapsone can cause oxidative damage to erythrocytes leading to hemolysis in patients with genetic deficiencies in the enzyme glucose-6-phosphate dehydrogenase (G6PD), patients being considered for dapsone treatment should be tested for G6PD deficiency. Patients who are G6PD deficient should not be treated with dapsone-containing regimens. If G6PD-deficient patients are unable to tolerate other regimens and are treated with TMP-dapsone, physicians should be prepared to support them with transfusion if severe hemolysis occurs.

Although it was previously recommended that clinicians should routinely monitor methemoglobin level during PCP therapy with TMP-dapsone, this step is probably not necessary unless the patient exhibits signs of cyanosis, anemia, or hypoxemia. Unfortunately, routine oximetry can be misleading because it will not distinguish methemoglobin from oxyhemoglobin. Thus, if significant methemoglobinemia is suspected, a methemoglobin level must be specifically requested. For patients with methemoglobin levels between 10% and 20%, discontinuation of dapsone generally results in resolution of the methemoglobinemia in 1-2 days. If the methemoglobin saturation reaches 20% or if the patient is symptomatic (or both), clinicians should consider changing therapy for PCP and administering methylene blue (1-2 mg/kg intravenously) to reverse the methemoglobinemia. Because dapsone has a long half-life, a repeat dose of methylene blue may be necessary or, alternatively, continuous infusion (0.1 mg/kg/hr) in those cases in which the methemoglobin remains persistently elevated may be considered.(238)

TMP and dapsone have a pharmacokinetic interaction, with each drug causing increased levels of the other. Compared with patients treated with dapsone alone, patients treated with TMP-dapsone had significantly higher dapsone levels (2.1 vs 1.5 mg/mL), and significantly higher rates of methemoglobinemia (67% vs 11%), adverse reactions requiring a change in therapy (30% vs 0%), and rash (40% vs 17%).(239) Compared with 30 patients treated with TMP-SMX, 30 patients on dapsone and the same dose of TMP had 48% higher TMP levels, and fewer required a change in therapy due to adverse reactions (30% vs 57%). Although acetylation and oxidative metabolism eliminate dapsone, acetylation status (rapid vs slow) has no relationship to toxicity.

Patients allergic to another sulfa drug are not necessarily allergic to dapsone.(240) In a series of 21 patients who had adverse reactions to TMP-SMX during initial treatment for PCP and were then treated with TMP-dapsone for their second episodes, only 6 had adverse reactions. Conversely, of 9 patients who had prior reactions to dapsone, none had problems when taking TMP-SMX for their second episodes.(214)

In an open study of dapsone (100 mg/day orally) without TMP, 11 (61%) of 18 patients with first-episode PCP responded, and 7 (39%) did not.(241) Of the 11 patients who responded, 6 (55%) had adverse reactions, although none was severe enough to require changing therapy. Given the availability of other more effective oral regimens, dapsone alone is not recommended as a therapy for PCP.

Two 21-day regimens of clindamycin and primaquine have been used to treat PCP: 1) primaquine, 30 mg (26.3-mg base) given orally once daily, plus clindamycin, 900 mg given intravenously every 8 hours, then lowered to 450 mg orally every 6 hours once improvement occurs; and 2) an orally administered regimen using primaquine (same dose) plus clindamycin, 600 mg every 6 hours.(242)

Clinical experience suggests that rash and diarrhea, attributed mainly to clindamycin, are the main adverse reactions. Patients otherwise doing well are able to continue the regimen despite the rash. Methemoglobinemia induced by primaquine is a concern, but methemoglobin levels rarely exceed 20% (see methemoglobinemia comments above in "TMP-Dapsone" section). Primaquine should be avoided in patients with G6PD deficiency.

A multicenter trial in the United States randomized 181 patients with initial P(A-a)O₂ <45 mm Hg to receive either oral TMP-SMX, TMP-dapsone, or clindamycin-primaquine.(243) The dapsone dosage was 100 mg daily; the clindamycin dosage was 600 mg 3 times daily; the primaquine dosage was 30 mg (base) daily; and the TMP and TMP-SMX dosage was approximately 5 mg/kg (of TMP) 3 times daily. Subjects with a P(A-a)O₂ of 35-45 mm Hg received concomitant corticosteroid therapy. Fifty-four percent of patients completed the full regimen of the initially assigned therapy (50% of TMP-SMX, 59% of TMP-dapsone, 52% on clindamycin-primaquine. At day 7, therapeutic failure was declared in 8% of patients treated with TMP-SMX, 5% with TMP-dapsone, and 5% with clindamycin-primaquine. At day 21, the failure rate overall was 9% (9%, 12%, and 7%, respectively).

No clear differences in mortality or relapse rates were detected among treatment arms. Rash was the most frequent dose-limiting adverse reaction (19% of TMP-SMX patients, 10% on TMP-dapsone, 21% on clindamycin-primaquine), with severe rash occurring most often in the clindamycin-primaquine arm (16%). Serious hematologic toxicities (marrow depression or methemoglobinemia >15%) occurred in 11% of TMP-SMX patients, 5% of TMP-dapsone patients, and 28% of clindamycin-primaquine patients. No patient receiving TMP-SMX, 1 patient (2%) receiving TMP-dapsone, and 4 patients (7%) receiving clindamycin-primaquine had methemoglobinemia >15%. Elevation of liver function tests above 5 times the upper limit of normal occurred in 19% of patients receiving treatment with TMP-SMX, 3% with TMP-dapsone, and 7% with clindamycin-primaquine. Overall, the 3 treatment regimens were comparable, with differences mainly in the toxicities.

Trimetrexate (TMTX) is a DHFR inhibitor that is active against P jiroveci. TMTX is administered with folinic acid (leucovorin, LVN) (20 mg/m² intravenously or orally every 6 hours for 23 days) to rescue hematopoietic cells that depend on DHFR activity. Adverse reactions include fever, rash, severe abdominal pain, anemia, neutropenia, thrombocytopenia, and transaminase and alkaline phosphatase elevations. Neutropenia and thrombocytopenia are the most frequent adverse reactions that require dosage modification or short-term suspension of TMTX therapy.(244) The usual dose is TMTX 30-45 mg/m² intravenously once daily for 21 days, plus LVN 20 mg/m² intravenously or orally every 6 hours for 23 days (ie, continuing for 2 days after cessation of TMTX).(245)

TMTX-LVN is most commonly used as salvage therapy for individuals who fail to respond to, or cannot tolerate, standard PCP therapies. Such patients may have up to a 50% survival rate when treated with TMTX-LVN. For those patients who have been refractory to conventional therapy, however, and have been supported with mechanical ventilation, the survival rates range from 6% to 16%.(246-248) One study determined response rates to TMTX (30 mg/m²) among 49 patients in 3 groups: patients failing or intolerant of conventional therapy (response rate 63%), untreated patients with sulfa intolerance (response rate 70%), and untreated patients also receiving 1 g of sulfadiazine orally every 6 hours (response rate 70%).(249) Neutropenia or thrombocytopenia occurred in 12 patients (25%), requiring dosage reduction in 9 patients. Of concern is that 7 of 38 survivors had a documented relapse within 3 months after therapy was completed. In a multicenter trial in the United States, 215 patients with baseline P(A-a)O₂ >30 mm Hg were randomly selected to receive TMTX-LVN (45 mg/m² once daily plus 20 mg/m²) or TMP-SMX (dosed at 5 mg/kg of TMP) for 21 days, improvement in the P(A-a)O₂ at 10 days of therapy was significantly better in patients receiving TMP-SMX, although treatment failure rates at day 10 and day 21 did not differ significantly between treatments.(250) However, although the TMTX-LVN was better tolerated, the mortality at 4 weeks after end of therapy was significantly higher. Therefore, TMTX-LVN should not be used as first-line therapy over TMP-SMX. No direct comparison between TMTX-LVN and pentamidine has been performed.

Atovaquone (Mepron) is an antimalarial drug that is relatively well tolerated.(251) In a phase I study using escalating daily doses from 100 to 3,000 mg/day,(252,253) the only drug-related adverse reaction was a maculopapular rash that resolved without discontinuation of the drug. Pharmacokinetics of the drug suggest that the half-life is up to 77 hours (mean, 55 hours); when it is administered with food, the drug's limited absorption is enhanced.(253,254) In a study of atovaquone treatment in 34 patients with mild to moderately severe PCP (PaO₂ >60 mm Hg on room air), all patients survived, 85% improved, and 79% were considered successfully treated; 5 patients did not respond to therapy.(255) Adverse reactions included rash (7 patients), neutropenia (3 patients), anemia (6 patients), and liver enzyme elevations (9 patients). Only 4 patients were withdrawn from treatment because of adverse reactions (2 each for rash and drug fever). At 6 months of follow-up, 7 relapses were recorded among the 27 patients who were successfully treated (26%). Because the patients withdrawn for toxicity all were in the highest dosage regimen and mean steady-state concentrations were not significantly different among the 3 dosages, the middle dosing regimen (750 mg 3 times daily) was deemed the most appropriate.

Of 408 patients in a subsequent multicenter, double-blind, randomized comparison of atovaquone (750 mg) against TMP-SMX (320-1,600 mg), each given 3 times daily as an oral treatment for mild to moderately severe PCP, 322 had histologically documented PCP and 284 were evaluable.(256-258) Of the evaluable patients, the success rates were similar: 62% for the atovaquone group and 64% for TMP-SMX group; treatment-limiting adverse reactions occurred in 7% and 20% of patients, respectively. Thirty-four percent of both groups required treatment with alternative medications to complete the courses of therapy. Twelve deaths were recorded (atovaquone, 11; TMP-SMX, 1; 7% vs 0.6%); none were attributed to PCP. A significantly higher percentage of the total enrolled patients treated with TMP-SMX had treatment-limiting adverse reactions (24% vs 9%), with significant differences occurring in the frequency of fevers, nausea and vomiting, neutropenia, and liver dysfunction. Rash was reported in both groups, but the incidence was not significantly higher in the TMP-SMX arm. Therapeutic success correlated with a higher steady-state concentration of atovaquone; preexisting diarrhea, which may affect drug levels, was associated with increased mortality in the atovaquone group. Rash appears to be correlated to the plasma concentration of the atovaquone.

In a comparison of oral atovaquone (750 mg 3 times daily) with pentamidine (3 to 4 mg/kg intravenously), 109 patients with P(A-a)O₂ <45 mm Hg and a PaO₂ >60 mm Hg on room air were treated and found to have a successful completion rate of 57% vs 40% in favor of the atovaquone.(259) A higher failure rate in the atovaquone arm (29% vs 17% for pentamidine) was counterbalanced by a significantly higher rate of treatment-limiting adverse events in the pentamidine arm (36% vs 4%). Eight weeks after completing therapy, the mortality rates were essentially the same (16% vs 17%).

In a summary of patients with both mild-to-moderate and severe PCP who were treated with atovaquone, the response rate was 77.8% for the mild-to-moderate PCP patients and 56.4% for the severe PCP patients. However, 23.3% and 34.4%, respectively, required discontinuation for adverse events. Mortality attributable to PCP was 2.1% and 32%, respectively.(260)

These studies were performed prior to approval of the liquid suspension formulation of atovaquone. The superior bioavailability of the liquid formulation allows use of a 750-mg twice-daily dosing regimen instead of thrice-daily dosing.

Clinicians have used daily administration of AP (either a fixed dose of 600 mg or a dose of 4 mg/kg) to treat PCP.(261,262-272) This regimen may allow early relapse more frequently than TMP-SMX or intravenous pentamidine and should be reserved for patients with mild disease who cannot tolerate more effective regimens. This route of administration has fewer adverse effects and provides high levels of drug in the lung, but essentially no systemic levels--making it ineffective in the treatment of extrapulmonary pneumocystosis.

Corticosteroids are beneficial as adjunctive therapy when begun early in the treatment of moderate to severe cases of PCP. Respiratory failure in PCP usually occurs early in the course of therapy, so the benefit of corticosteroids may lie in preventing the deterioration of pulmonary function. Corticosteroids are recommended as an adjunct to anti-PCP therapy in patients with a PaO₂ <70 mm Hg, P(A-a)O₂ >35 mm Hg, or oxygen saturation <90% on room air, provided there are no specific contraindications (eg, suspected tuberculosis, disseminated fungal infection).(273,274) The usual corticosteroid regimen consists of prednisone 40 mg twice daily for 5 days, then 40 mg daily for 5 days, and then 20 mg daily until day 21 of therapy, based on the regimen used in one study.(275) Corticosteroid treatment ideally should begin at the same time as anti-PCP therapy and no later than 72 hours after anti-PCP treatment has begun. Many clinicians use an alternative regimen of 40 mg of prednisone twice daily for 5 days, 40 mg daily for 5 days, and 20 mg daily for 5 days.

The view that corticosteroids might be of benefit developed because of sporadic reports in which patients with PCP inadvertently received corticosteroids and improved.(276-278) A number of small series of patients compared corticosteroids and routine PCP therapy with routine therapy without corticosteroids. Although the steroid-treated groups usually were more severely ill and often required mechanical ventilation, their mortality, time to clinical and radiographic improvement, number of adverse effects, and relapse rate were generally improved over those for the non-steroid-treated groups.(279-284) Survival of steroid-treated patients was increased to levels occurring in less severely ill patients. Most reports, however, described small numbers of patients studied retrospectively without standardized steroid regimens.

The largest study of corticosteroids randomized 251 patients with acute PCP (225 confirmed, 26 presumed) in an unblinded trial comparing adjunctive prednisone treatment with conventional treatment without corticosteroids and found less frequent respiratory failure during therapy (30 patients on conventional therapy vs 13 on prednisone) and lower risk of death at days 21, 31, and 84.(275) Forty-one patients (16%) died--28 while receiving standard therapy and 13 while receiving prednisone. There were 4 relapses in each arm (5% on standard therapy and 4% on prednisone therapy). Corticosteroids did not reduce the rate of adverse reactions that required a change in therapy, but patients were more likely to complete a course of initial therapy because treatment failures were less frequent. Adverse events associated with corticosteroid use included an increase in episodes of thrush and reactivation of localized herpetic lesions. It is difficult to compare this study with others because of its unusual definitions of respiratory failure: death, need for mechanical ventilation, or hypoxemia ratio <75 (defined as partial pressure of arterial oxygen divided by fraction of inspired oxygen [PaO₂/FIO₂]).

Despite the beneficial effects of corticosteroids in patients with PCP, when given early in the course of PCP, there is no clear evidence that corticosteroids improve outcome in patients treated with corticosteroids only after their clinical condition has deteriorated.(285) Moreover, in such cases corticosteroid treatment has been associated with an increased rate of gram-negative bacilli infections.(286)

In contrast to studies mentioned above, a multicenter study of patients with confirmed, severe PCP (room air PaO₂ <70 mm Hg or P(A-a)O₂ >40 mm Hg) found no difference in survival, respiratory failure requiring mechanical ventilation, superinfection, or development of KS or lymphoma in patients randomized to receive prednisone (40 mg twice daily for 10 days) compared with placebo.(285) Thus, in this group of patients with severe PCP, the value of adjunctive steroids appears less clear than it does in those patients with moderate PCP. (This study did find a decrease in the incidence of hypersensitivity reactions to TMP-SMX in the steroid treatment group.)(287,288) In a separate review of 173 patients with PCP (31% with adjunctive steroids and 69% without), no difference was noted in length of survival after PCP, increase in mortality, or infectious complications associated with AIDS or KS.(289) In addition, evaluation of 148 patients with confirmed PCP showed that those patients who had CMV cultured from their BAL specimens and received adjunctive steroid therapy had a significantly worse survival at 3 months than those without CMV, adjunctive steroids, or both.(290)

In summary, although the benefits of corticosteroid use in moderate to severe PCP are well supported by clinical trials, the use of corticosteroids must be tempered by prudence in patients who may have concomitant active, untreated infections--particularly tuberculosis, but also viral, fungal, parasitic, or other bacterial pathogens.(286)

TMP-SMX is the preferred regimen for PCP prophylaxis because of low cost, efficacy, and simultaneous efficacy in prophylaxis against toxoplasmosis. The high frequency of adverse effects, however, prevents many patients from remaining on the drug.

In the first completed and published study of TMP-SMX prophylaxis in HIV-infected patients, 60 patients with KS were randomized to receive either TMP-SMX twice daily plus leucovorin, or no prophylaxis.(341) No patient randomized to receive TMP-SMX developed PCP while on therapy, 4 of 5 developed PCP after stopping the drug because of adverse reactions, and 16 of 30 in the control group developed PCP. Fifty percent of the patients on TMP-SMX had an adverse reaction; 4 of those who stopped had severe erythroderma, and the last had persistent neutropenia.

In an open-label study of primary prophylaxis, 215 patients were randomized to receive either AP at 300 mg once monthly, a single-strength tablet of TMP-SMX (80 mg of TMP per 400 mg of SMX) daily, or a double-strength tablet of TMP-SMX (160 mg of TMP per 800 mg of SMX) daily.(342,343) All patients had <200 CD4 cells/µL. The trial was aborted after a mean of 264 days of follow-up because 6 (11%) of 71 patients in the AP group developed PCP, compared with no patients in either of the TMP-SMX groups. Only 2 patients in the AP group were discontinued from treatment because of adverse reactions (cough and bronchospasm) compared with 17 in the group taking the single-strength tablet of TMP-SMX and 18 in the group taking the double-strength tablet of TMP-SMX (fever, rash, and nausea and vomiting). The mean time to onset of the adverse reactions was 57 days in the single-strength tablet group compared with 16 days in the double-strength tablet group. Three patients in the AP group and none in the TMP-SMX groups were diagnosed with Toxoplasma encephalitis.

In a multicenter trial of secondary prophylaxis sponsored by the National Institute of Allergy and Infectious Diseases, 310 patients who had recovered from their first episode of PCP were randomized to receive a double-strength tablet of TMP-SMX (160 mg of TMP per 800 mg of SMX) once daily or AP (300 mg) once monthly.(344) After a median follow-up of 17.4 months, 14 (9.1%) of 154 patients in the TMP-SMX arm and 36 (23.1%) of 156 patients in the AP arm developed PCP. The estimated 1-year recurrence rates were 3.5% in the TMP-SMX group and 18.5% in the AP group; at 18 months, the estimated recurrence rates were 11.4% and 27.6%, respectively. Forty-seven percent of the patients in the TMP-SMX arm and 44% of the patients in the AP arm experienced hematologic toxicity, and 3 patients in the TMP arm and 2 in the AP arm experienced pancreatitis. Forty-two patients (27%) in the TMP-SMX arm switched to AP, primarily because of rash or hematologic toxicity. Of note, only 7 of the 14 recurrences in patients assigned to the TMP-SMX arm actually occurred while the patients were still receiving TMP-SMX, including 3 patients who had PCP recurrences within 3 weeks of entry into the study. In contrast, 34 of 36 patients in the AP arm were still receiving AP when the PCP recurrence occurred (the remaining 2 were awaiting initial treatment). Toxoplasmosis occurred in 4 patients in the TMP-SMX group and 6 in the AP group.

Several studies evaluating the efficacy of intermittent TMP-SMX compared with daily dosing showed that intermittent dosing (usually 1 double-strength tablet 3 times weekly) has a very low failure rate.(345,346-348) The adverse reaction rate in one study was reportedly 2.4 times higher with the daily regimen compared with the thrice-weekly dosing regimen.(348) Shortcomings of these studies, however, include a short mean follow-up period, a retrospective approach, a nonrandomized scheme, or an insufficient number of patients.

Desensitization to TMP-SMX may allow some patients with non-life-threatening prior reactions to take TMP-SMX for continuing prophylaxis. Proposed desensitization regimens vary in length from as short as 5 hours to as long as several weeks. The desensitization success rate also varies, but can be as high as 86%.(349-356)

In the only completed randomized, double-blind study of desensitization, 191 patients who had prior documented rash or fever to TMP-SMX were randomized to receive either direct rechallenge with a single-strength TMP-SMX tablet or an escalating dosage schedule over 5 days culminating with a single-strength TMP-SMX tablet on day 6 (see Table 3).(357) Although the 5-day schedule could be extended to a maximum of 10 days if necessary, patients and physicians were blinded to study treatment until the patients actually reached the single-strength tablet dosage. Success was defined as being able to stay on TMP-SMX for 6 months, but patients could be treated symptomatically for rash or fever while participating in the study. The results show that 75.3% (73 of 97) of patients who underwent dosage escalation, compared with 57.4% (54 of 94) of patients who were directly rechallenged, were able to continue single-strength TMP-SMX for 6 months. The only significant adverse reactions reported were rash and fever. Both arms reported similar rates of rash (approximately 15%), but fever occurred more frequently in the direct rechallenge arm (18% vs 7%) and was seen more frequently in the rechallenge arm as a cause of discontinuation (12% vs 4%). Although the observation that 57% of all patients with prior rash or fever were able to tolerate TMP-SMX after direct rechallenge offers hope that patients with prior reactions may once again be able to take TMP-SMX, it remains difficult to predict which patients will require desensitization and in whom it will ultimately be successful.

In a study of primary prophylaxis, 213 patients were evenly randomized to AP 300 mg monthly, vs single-strength TMP-SMX once daily, vs double strength TMP-SMX.(358) Two patients on the AP arm developed PCP (mean follow-up, 264 days) compared with no patients on either of the TMP-SMX arms (mean follow-up 288 days). Interestingly, it appears that the higher dosage of TMP-SMX did not lead to a significantly higher rate of adverse reactions, but they seemed to appear earlier in the course of the follow-up period. These data suggest that single-strength TMP-SMX may have efficacy similar to that of double-strength TMP-SMX prophylaxis against PCP.

Widespread use of TMP-SMX at dosages used for prophylaxis may lead to resistance, not only in P jiroveci (see "Drug Resistance" section), but also among bacteria against which the drug has been useful. Seventy-five (42%) of 179 Streptococcus pneumoniae isolated from blood of HIV-positive patients were found to be penicillin resistant, in contrast to only 12% in HIV-negative patients. Thirty-four of those isolates were further tested, and 15 (44%) were found to be resistant also to TMP-SMX.(359) TMP-SMX resistance correlated with length of time on PCP prophylaxis using TMP-SMX. Although the broader use of TMP-SMX so far has not resulted in widespread reports of infections with TMP-SMX-resistant S pneumoniae, prolonged use of agents with broad spectra for prophylaxis may result in resistance by the untargeted organisms indigenous in the target population. In this regard, other agents with more limited spectra may be more appropriate (see below).

Because the efficacy of dapsone is lower than that of TMP-SMX for PCP prophylaxis, dapsone is recommended only if first-line prophylaxis with TMP-SMX is not tolerated. Dapsone is tolerated by about two thirds of patients who have experienced side effects on TMP-SMX,(360,361) but dapsone itself may cause serious side effects, including methemoglobinemia and hemolysis.

Dapsone has been studied at a variety of dosing regimens as prophylaxis against PCP. One study used dapsone sulfate (25 mg by mouth 4 times a day) in 221 HIV-infected patients as both primary and secondary prophylaxis.(362) Only 2 patients receiving prophylaxis developed PCP, in contrast to 19 of 26 patients who refused prophylaxis. Adverse reactions to dapsone included anemia, LDH elevation, methemoglobinemia, nausea, and skin rash. Because the patient population was mixed, quantitative efficacy of dapsone was difficult to assess.

Another report described 61 patients with advanced HIV disease who were treated with dapsone at 100 mg (7 patients), 200 mg (50 patients), or 300 mg (4 patients) weekly for primary (48 patients) and secondary (13 patients) prophylaxis.(363) Thirteen of 17 patients who had previously received prophylaxis with TMP-SMX had an adverse reaction, as had 3 of 12 who had previously received prophylaxis with AP. At a mean of 9 months of follow-up, only 1 (2%) of 50 patients taking 200 mg of dapsone developed PCP, in contrast to the 14 cases expected to develop if they had not received prophylaxis. Eight patients (13%) experienced adverse reactions, including rash (2 patients), anemia (2 patients), and nausea, neutropenia, headache, and elevated LFT (1 each). Only 4 of the 13 patients who previously had an adverse reaction to TMP-SMX developed an adverse reaction while on dapsone. Studies using 50-100 mg of dapsone 3 times weekly in patients who had previously exhibited adverse reactions to TMP-SMX suggested that dapsone sulfate alone may have a failure rate between those of TMP-SMX and AP.

In a study comparing dapsone sulfate (50 mg/day) directly with AP (300 mg monthly) for secondary prophylaxis, 5 (5.5%) of 92 patients on dapsone developed PCP compared with 12 (11.8%) of 102 patients on AP.(364,365) In contrast, when the dosage regimens of dapsone (100 mg twice weekly) and AP (100 mg every 2 weeks) were studied in 278 patients, 10% of patients in each group developed PCP at 1 year of follow-up.(366,367) Three pneumothoraces occurred (in 2 patients on AP, 1 on dapsone).(366)

In the first completed and published study evaluating dapsone alone for prophylaxis, 843 patients with CD4 counts <200 cells/µL were randomly selected to receive TMP-SMX double strength twice daily), dapsone sulfate (50 mg twice daily), or AP (300 mg every 4 weeks) for primary prophylaxis.(368) The 36-month cumulative risk for PCP was 18% in the TMP-SMX group, 17% in the dapsone group, and 21% in the AP group. Among those patients who had CD4 counts <100 cells/µL, the cumulative risk was 19%, 22%, and 33%, respectively. Only 21% of TMP-SMX patients and 25% of dapsone patients were able to complete the study on the originally assigned drug regimen, compared with 88% of the AP patients. Of the 24 cases of cerebral toxoplasmosis noted, approximately 60% of those occurred while the patients were receiving AP.

Because of concerns about the suboptimal efficacy of dapsone alone, the addition of pyrimethamine has been studied and appears to be effective.(369) Pyrimethamine is generally given once a week, at a dose of 50-100 mg, with folinic acid (LVN) to prevent myelosuppression. Studies comparing pyrimethamine-dapsone with AP for PCP prophylaxis did not show a significant difference in efficacy, but a large proportion of patients in these studies switched treatment during follow-up.(370,371) A study comparing dapsone (100 mg) plus pyrimethamine (50 mg), each given weekly, with TMP-SMX (double strength, twice a day, 3 times weekly) for primary prophylaxis found a higher incidence of PCP in the pyrimethamine-dapsone arm. At a mean follow-up of 430 days, 6 (6.3%) of 96 evaluable patients in the pyrimethamine-dapsone arm developed PCP compared with none of 104 patients in the TMP-SMX arm, but because only 1 patient was actually still taking the medications when PCP occurred, actual efficacy could not be determined. The rates of adverse events that required switching to another medication were 9.6% for pyrimethamine-dapsone and 9.3% for TMP-SMX. Of 131 patients seropositive for Toxoplasma gondii at the onset of the study, only 2 patients in the pyrimethamine-dapsone arm and 1 patient in the TMP-SMX arm developed toxoplasmosis.(372) Other studies have compared pyrimethamine-dapsone at various doses with TMP-SMX, AP, or both, and all show similar or minimal decrease in efficacy in prophylaxis against PCP.(373,374) Thus, adding weekly pyrimethamine plus folinic acid to daily dapsone for PCP prophylaxis is a reasonable alternative in those patients at risk of developing toxoplasmosis (eg, CD4 count <100 cells/µL and positive toxoplasma antibody test).

The side effects of dapsone are discussed under "Treatment Regimens" above. Patients who have experienced fevers and rash while on dapsone may benefit from a desensitization regimen starting at 0.01 mg and progressing over 6 weeks to 50 mg twice daily.(375) In addition to reports of methemoglobinemia, and anemia in patients with G6PD deficiency, a sulfone syndrome also has been attributed to dapsone.(376) The patient had previously experienced fevers, sweats, and skin lesions thought to have been caused by TMP-SMX, and was enrolled into a dapsone prophylaxis study (50 mg twice daily). Within 2 months, the patient developed fever, jaundice, hepatomegaly, lymphadenopathy, atypical lymphocytosis, hemolytic anemia, and elevated transaminases, and a liver biopsy specimen showed granulomas and a panlobar eosinophilic infiltrate. All symptoms resolved with cessation of the dapsone therapy.

Although the systemic toxicities of dapsone may be problematic, when TMP-SMX is not tolerated, dapsone offers efficacy as prophylaxis against PCP. In addition, unlike AP, dapsone provides some protection against toxoplasmosis, and potentially against extrapulmonary pneumocystosis.(368,371,372)

Atovaquone was shown to be active in treatment of both PCP and central nervous system toxoplasmosis prior to demonstration of its efficacy in prophylaxis. Atovaquone (1,500 mg daily) was compared with dapsone (100 mg daily) in 1,057 patients with the accepted indications for primary or secondary PCP prophylaxis plus a prior history of intolerance to TMP-SMX.(377) For patients randomized to dapsone who also had CD4 counts <100 cells/µL and positive toxoplasma serology, pyrimethamine (50 mg) and folinic acid (15 mg) were both recommended once weekly for improved toxoplasma prophylaxis. After a mean follow-up of 24 months, 121 confirmed and probable PCP events occurred in patients assigned to atovaquone (15.5/100 patient years) and 134 events in patients on dapsone (18.3/100 patient years). There were only 7 cases of central nervous system toxoplasmosis; 4 on atovaquone and 3 on dapsone (no significant difference). There was no significant difference in the overall rate of adverse reactions between the 2 arms, but those in the atovaquone arm mainly had gastrointestinal symptoms, whereas those on dapsone had more hypersensitivity reactions (rash, fever, pruritus, and dermatitis) and anemia.

Of interest, however, was the observation that patients on dapsone at enrollment and subsequently randomized to atovaquone had a 3.55 higher rate of adverse reactions requiring discontinuation as compared with those patients randomized to continue dapsone (29.9 vs 8.0/100 patient years). In contrast, among patients not on dapsone at the time of randomization, those randomized to atovaquone had a lower rate of treatment-limiting adverse reactions compared with those randomized to dapsone (23.2 vs 56.2/100 patient years). Accordingly, those patients already receiving dapsone at baseline were able to remain on the drug substantially longer than those patients who were not originally on dapsone but then randomized to receive it.

The conclusion from this study is that the efficacy of atovaquone for PCP prophylaxis is similar to that of dapsone, and that patients who have already demonstrated the ability to tolerate 1 of these 2 drugs are more likely to have side effects if switched to the other drug. Thus, if the patient who had an adverse reaction to TMP-SMX is already on dapsone, dapsone therapy should continue, but if the patient is not on dapsone, atovaquone may be preferred, given its more manageable side effect profile, for PCP prophylaxis.

Although it is not the most effective, AP using pentamidine isethionate was the first and, for several years, only FDA-approved agent for prophylaxis of PCP in HIV-infected patients.(330) On the basis of 2 studies, the FDA approved the use of AP (300 mg every 4 weeks via the Respirgard II nebulizer or equivalent) for PCP prophylaxis in patients with a prior episode of PCP or with CD4 counts <200 cells/µL.(327,378)

A dose-ranging study in 408 HIV-infected patients in San Francisco established the superiority of the 300 mg dose every 4 weeks over lower doses. The most common adverse effects were cough (36%) and mild bronchospasm (11%); these were easily treated with bronchodilators and resulted in interruption of <1% of all treatments. No cases of extrapulmonary PCP occurred. Twenty-one patients had 46 pneumothoraces, 9 of whom had multiple events, and 6 had bilateral events. Patients with a higher number of previous PCP episodes had a higher risk of pneumothorax. All episodes of pneumothorax but 1 were associated with an acute episode of PCP, pulmonary parenchymal disease from prior PCP, or an invasive procedure.(379) Several placebo-controlled or observational studies confirmed the efficacy of AP in different formulations or delivery systems.(380-382,383-385) In a randomized open-label study of primary PCP prophylaxis in 66 patients, TMP-SMX (1 double-strength tablet once daily), was associated with significantly improved mortality compared with AP.(386)

An investigation of the possible reasons for AP failures determined that failure is not solely related to the total dose of pentamidine delivered to the lung.(387) Other potential factors include regional differential deposition of pentamidine, inadequate deposition in alveoli, individual pentamidine clearance rates, and nebulizer variability. The optimum regimen and delivery system may never be determined. Aside from the adverse pulmonary reactions, hypoglycemia has been the only significant systemic reaction reported, but this is rare.

Although not used as PCP therapy, the antimalarial combination drug pyrimethamine-sulfadoxine (25 mg of pyrimethamine, 500 mg of sulfadoxine; brand name Fansidar) once weekly has been used for PCP prophylaxis.(388,389) Pyrimethamine-sulfadoxine has shown efficacy as primary and secondary PCP prophylaxis.(390-394) However, reports of severe cutaneous adverse reactions, including Stevens-Johnson syndrome and toxic epidermal necrolysis, have prevented its widespread use in HIV-infected individuals.(390,391,395)

Pentamidine (4 mg/kg monthly, intramuscularly or intravenously) has been studied as secondary prophylaxis in a small number of patients who had previously received pentamidine for treatment of PCP.(396) Only 2 of 10 patients had a recurrence, compared with 7 of 12 patients who had received no prophylaxis.

The question of whether to discontinue PCP prophylaxis arose in the late 1990s as it became clear that a large proportion of patients treated with effective ART would have increases in their CD4 counts to levels well above 200 cells/µL, the threshold for initiating PCP prophylaxis. The major studies that have looked at this question in either primary or secondary prophylaxis are summarized below.

In an uncontrolled, observational Swiss cohort study, 262 patients on primary PCP prophylaxis and having CD4 counts >200 cells/µL (and at least 14% of total lymphocytes) over at least 3 months discontinued primary prophylaxis.(397) Median nadir CD4 count was 110 cells/µL; median CD4 count on entry was 325 cells/µL; and median HIV viral load at entry was 500 copies/mL. Median follow-up was 11 months, and no episodes of PCP were found. The expected number of cases based on historical data from the cohort was 6.1 cases per patient year, or 15 cases. Seven patients required restarting prophylaxis due to a loss of CD4 cells to <200 cells/µL.

In a randomized clinical trial, 708 patients on combination ART with improvements in their CD4 counts to >200 cells/µL on 2 occasions 3 months apart were randomized to continue (353 patients) or to discontinue (355 patients) primary prophylaxis.(398) Median nadir CD4 counts were 90 and 89 cells/µL, respectively. At enrollment, patients in the continuation arm had a median CD4 count of 336 cells/µL and patients in the discontinuation arm had a median CD4 count of 324 cells/µL; median plasma HIV viral load was <2.7 log₁₀ copies/mL in both arms. The mean duration of follow-up was 6.12 months (209 patient years) and 6.36 months (210 patient years), respectively. No cases of PCP were found in any patient in either arm.

In a study of discontinuation of secondary prophylaxis for PCP, 325 patients in 8 European cohorts were followed for a median of 13 months (374 patient years) after recording at least 1 CD4 count of >200 cells/µL.(399) The median nadir CD4 count was 50 cells/µL but the median CD4 count at discontinuation of prophylaxis was 350 cells/µL; 76% of patients had plasma viral load copies <500 copies/mL. No patients were found to have PCP during the study.

EuroSIDA, a prospective observational cohort study in 52 centers across Europe, studied discontinuation in 319 patients on primary prophylaxis and 59 patients on secondary prophylaxis. The median nadir CD4 count and CD4 count at discontinuation was 123 cells/µL and 274 cells/µL, respectively, for the primary prophylaxis patients, and 60 cells/µL and 270 cells/µL, respectively, for secondary prophylaxis patients. Median time of CD4 count to >200 cells/µL before discontinuation was 8 and 5 months, respectively. Over a 3-year period, there were no cases of PCP found in the study participants.(400) These findings are mirrored in other studies that also have examined discontinuation of both primary and secondary prophylaxis.(401,402)

Certainly discontinuation of prophylaxis for PCP has several benefits. A reduction in the number of pills, a lower likelihood of resistance not only in P jiroveci but also in bacterial species, a reduction in the risk of adverse effects and drug interactions, and lesser drug costs are all welcome. Factors to be considered when deciding whether to discontinue prophylaxis include: current CD4 cell count, length of time above the 200 cells/µL threshold, and plasma HIV viral load. Most experienced practitioners would maintain prophylaxis until the CD4 count is >200 cells/µL for at least 3 months; some have preferred 6 months. Although having an undetectable plasma viral load is certainly desirable, it is clear that not all patients will be able to achieve that, and available data suggest that patients with lowered, but not undetectable, viral loads also have a lowered risk of PCP as long as other criteria are fulfilled. Thus, effective ART can reduce the need for both primary and secondary prophylaxis.(403,404)