American Journal of Respiratory and Critical Care Medicine

Bronchoalveolar lavage (BAL) and transbronchial biopsies from 351 human immunodeficiency virus (HIV)-positive patients with presumed Pneumocystis pneumonia were analyzed to determine the spectrum and frequency of intersitital lung disease mimicking Pneumocystis pneumonia. Among 67 patients without Pneumocystis, nonspecific interstitial pneumonitis (NSIP) was the most common histologic diagnosis (n = 16). Tissue sections from patients with NSIP were tested by in situ hybridization for Epstein–Barr virus, cytomegalovirus (CMV), and HIV; sections were also tested with the polymerase chain reaction (PCR) for HIV env and gag protein DNA. In patients with NSIP, Epstein–Barr virus and CMV could not be detected by in situ hybridization; HIV nucleic acid was amplifiable with PCR in 10 of 15 formalin-fixed, paraffin-embedded tissue sections. Symptoms, physical findings, and blood gas values were similar in patients with NSIP and matched controls with Pneumocystis. Patients with NSIP presented earlier in the course of HIV, with higher weight, serum albumin levels, and CD4+ T-lymphocyte counts (492 ± 828 cells/mm3 versus 57 ± 60 cells/mm3), and more normal lactate dehydrogenase (LDH) levels (280 ± 113 IU/L versus 432 ± 141 IU/L; means ± SD). Seven to 10 d later, improvement in blood gas values was of similar magnitude for the two groups. Only one other unequivocal, treatable infection was diagnosed only with transbronchial biopsy. These results indicate that NSIP may be the most common diagnosis mimicking Pneumocystis pneumonia in acquired immune deficiency syndrome (AIDS), and that NSIP may improve during empiric therapy.

In most developed countries, Pneumocystis pneumonia remains the most common serious initial infection in human immunodeficiency virus (HIV)-positive persons, and occurs in up to 50% of patients with acquired immune deficiency syndrome (AIDS) in the last 6 mo of their lives (1). The initial diagnostic maneuver for individuals presenting with possible Pneumocystis pneumonia is usually sputum induction, which is associated with a variable sensitivity of 50 to 95% (2-5). Because the test can have a negative predictive value as low as 39% (2), the absence of Pneumocystis carinii in induced sputum may not reliably exclude the diagnosis. Other patients are too ill to undergo sputum induction or bronchoscopy. Thus, a sizable number of patients are initially treated without a confirmed diagnosis.

It is important to know what other causes of interstitial pneumonia mimic Pneumocystis when patients are treated empirically for P. carinii. This is of considerable importance if a sizable proportion of these patients have other treatable conditions, such as tuberculosis, fungal infection, or Kaposi's sarcoma. The problem is further complicated because in severe cases of Pneumocystis pneumonia, adjunctive corticosteroids should be prescribed (6). Improvement might occur initially with infections other than Pneumocystis if corticosteroids attenuate the inflammatory response in the lungs, but this would delay the diagnosis and specific treatment of these other disorders. Worse yet, pulmonary deterioration may occur rapidly when corticosteroids are administered to patients with pulmonary Kaposi's sarcoma (7, 8). Moreover, patients without Pneumocystis pneumonia may experience unnecessary drug toxicity during empiric therapy.

To ascertain the spectrum and frequency of causes of interstitial pneumonia in patients in whom Pneumocystis was not confirmed, we reviewed our experience with 351 patients who satisfied criteria for presumptive Pneumocystis pneumonia and who entered studies of experimental therapies. These patients were treated at a single institution at a time when sputum induction was not readily available. Thus, all patients subsequently underwent bronchoscopy with both bronchoalveolar lavage (BAL) and transbronchial biopsy. We report the histologic diagnoses in patients in whom P. carinii was not found with these procedures. We also investigated nonspecific interstitial pneumonitis (NSIP), the most common histologic diagnosis, in HIV-positive patients without Pneumocystis, in order to determine its etiology and whether there were clinical features that distinguished patients with NSIP from those with Pneumocystis.

Three hundred and fifty one patients with clinical features deemed highly suggestive of Pneumocystis pneumonia (fever, nonproductive cough, dyspnea, interstitial radiographic infiltrates, elevated serum lactate dehydrogenase [LDH] levels, and hypoxemia) were empirically enrolled in studies at Los Angeles County–University of Southern California Medical Center to test new therapies for Pneumocystis pneumonia. These studies were conducted between November 1986 and October 1990.

To reduce the likelihood of including patients with mycobacterial or fungal infections or pulmonary Kaposi's sarcoma, potential study candidates whose chest radiographs showed miliary patterns, hilar adenopathy, or pleural effusions were excluded from enrollment. None of the study patients had received aerosolized pentamidine for prophylaxis of Pneumocystis pneumonia or chemotherapy for Kaposi's sarcoma or lymphoma.

Experimental therapies for Pneumocystis pneumonia that were evaluated in the studies included trimetrexate plus leucovorin (AIDS Clinical Trials Group [ACTG] protocol 029/031), aerosolized pentamidine (ACTG protocol 040), adjunctive corticosteroids (California Collaborative Treatment Group protocol 509), and primaquine plus clindamycin (ACTG 044). These studies were not limited to patients with first episodes of Pneumocystis pneumonia; rather, each evaluated therapies based on severity of Pneumocystis pneumonia at presentation. ACTG 029/031 and CCTG 509 selected patients with moderate-to-severe Pneumocystis pneumonia (i.e., baseline [A–a]Do 2 > 30 mm Hg), whereas ACTG 040 and 044 selected patients with mild-to-moderate pneumocystis pneumonia (i.e., baseline [A–a]Do 2 < 40 mm Hg).

Each patient provided written informed consent, and was scheduled for bronchoscopy with BAL and to have four to six transbronchial biopsies. Bronchoscopy was performed within 5 d after the start of empiric therapy for Pneumocystis pneumonia. Transbronchial biopsies were done in all cases unless there was a coagulopathy or endobronchial Kaposi's sarcoma. The BAL sample from each patient was routinely processed for bacterial, fungal, and mycobacterial cultures, and was stained with Gomori's methenamine silver for cytologic demonstration of P. carinii and other fungi.

Cytologic diagnosis of Pneumocystis pneumonia was confirmed in 284 (81%) of the patients. In the remaining 67 (19%) patients, BAL and transbronchial biopsies revealed no evidence of P. carinii. Experimental therapy was continued for 21 ± 3 d, as mandated by the protocols for patients with confirmed Pneumocystis pneumonia. For subjects without documentation of P. carinni, experimental therapy for Pneumocystis pneumonia was discontinued as soon as the results of examination of BAL and transbronchial specimens were complete.

Clinical information gathered prospectively, culture results, and histopathology of biopsy specimens from the patients without Pneumocystis pneumonia served as the primary data set for this study.


The histopathology of transbronchial biopsies not showing P. carinii was independently reviewed by an AIDS pathologist (L.N.) and a pulmonary pathologist (M.K.), who were blinded to the clinical aspects of the study cases. Biopsy specimens were considered adequate for review if there were at least 30 total alveoli in sections for each patient. All tissue sections were stained with methenamine silver (Gomori method) for P. carinii and other fungi, and with the Ziehl– Neelsen method for mycobacteria.

NSIP was defined by the presence of chronic inflammation in the interalveolar septa in the absence of a specific etiologic agent identified through standard histopathologic, histochemical, cytologic, and culture techniques. Specific histologic abnormalities of NSIP, namely interstitial inflammation, interstitial fibrosis, interstitial edema, pneumocyte hyperplasia, alveolar edema, alveolar hyaline membranes, alveolar inflammation, bronchial inflammation, and vasculitis were sought. The extent of each finding was graded numerically as 0 (absent), 1 (plus/minus, < 10% involvement), 2 (mild, 10 to 25% involvement), 3 (moderate, 26 to 50% involvement), or 4 (severe, > 50% involvement). Cytomegalovirus (CMV) pneumonia was diagnosed only if sections showed typical nuclear inclusions with surrounding halos, plus adjacent interstitial inflammation. Bronchitis was defined by acute or chronic inflammatory infiltrates in the bronchial wall, with normal lung parenchyma.

In situ Hybridization and Tissue Polymerase Chain Reaction Studies

In 15 of the 16 biopsies showing NSIP, Epstein–Barr virus (EBV), CMV, and HIV nucleic acid was sought through in situ hybridization. Biopsies from two HIV-positive patients with pulmonary lymphoid hyperplasia (PLH), and biopsies from three HIV-positive patients with normal lung tissue were also tested. For EBV and CMV, biotinized probes were used (Enzo Biochem, New York, NY). The in situ probes from CMV, EBV, and HIV have a limit sensitivity of 10 to 20 copies of target per cell.

Formalin-fixed, paraffin-embedded tissue sections were deparaffinized in successive rinses with xylene and alcohol, followed by blocking of endogenous peroxidase activity with 3% H2O2 in 100% methanol for 15 min at room temperature. A solution of probe was then applied to the tissue sections and heated to 92° C for 10 min. The tissue sections were then incubated at 37° C for 25 min, covered with posthybridization solution for 10 min at 37° C, and then rinsed with wash buffer. Avidin–biotin–peroxidase complex was then applied for 15 min, followed by aminoethyl carbazole solution for 10 min. The sections were then rinsed and counterstained with Mayer's hematoxylin for 3 min. Positive control cells provided by the manufacturer, as well as negative and positive control tissue sections, were processed with the transbronchial biopsies.

For in situ hybridization of HIV nucleic acid, an alkaline phosphatase-linked RNA probe (DuPont, Boston, MA) was used. Briefly, the tissue sections were rehydrated in 100 nM MgCl2 for 10 min at room temperature. They were then prehybridized in formamide for 10 min at 70° C, rinsed in 5× standard citrate saline (SSC), heated to 55° C, incubated with probe solution for 60 min at 55° C, and treated with detection buffer for 4 h in the dark at 37° C; the development was stopped by rinsing the slides in distilled water. The sections were counterstained with methyl green for 25 s, rinsed with distilled water, and allowed to air-dry in the dark. Positive control cells infected with HIV, as well as negative and positive control retinal tissue sections, were processed with the transbronchial biopsies.

The HIV gag and env genes were each sought separately in these same formalin-fixed, paraffin-embedded tissues with the polymerase chain reaction (PCR) (9). In brief, 10-μm-thick sections were suspended in 90 μl PCR buffer with 0.25% Tween 20 and 10 μl proteinase K (10 mg/ml) prior to incubation at 56° C overnight. After boiling, DNA templates were amplified with 2.5 units Taq polymerase (Perkin Elmer-Cetus) in a standard 100-μl PCR reaction for 40 cycles at 94° C for 30 s, 42° C for 15 s, and 72° C for 1 min. Primer pairs (1 μM each) were SK38 and SK39 (HIV gag) and SK68 and SK39 (HIV env). The PCR products were denatured, fixed to nylon membranes, and subsequently hybridized to 32P-labeled probes (SK19, SK70, or K12, ∼ 1 × 109 cpm/μg). Washed dot blots (50° C, 1× SSC, 0.2% sodium dedecylsulfate [SDS]-[final stringency]) were exposed on XAR-5 film (Kodak, Rochester, NY). All experiments included positive and negative control templates, deoxyribonucleic acid (DNA) control reactions, and reagent and water negative controls, with the expected results.

The PCR technique is capable of identifying less than 10 DNA templates per reaction. Diluted positive control templates at the level of 100 copies per PCR reaction were demonstrably positive in this system. Therefore, the limit sensitivity was considered to be less than 100 copies per PCR reaction under the described experimental conditions.

Extended formalin fixation can decrease the availability of targets for PCR amplification, by cross-linking. PCR experiments in this study were done on surgical specimens that were routinely formalin fixed for less than 24 h. Comparisons between DNA extracted from fresh and formalin-fixed, paraffin-embedded sections were not done to assess yield or variation in priming efficacy.

Flow Cytometry for Lymphocyte Phenotyping

All 16 patients with NSIP and 32 case controls had lymphocyte phenotyping done within three days of initiating experimental therapies for Pneumocystis pneumonia. The tests were done in the University of Southern California Flow Cytometry Laboratory, which is certified by the College of American Pathologists and the National Institutes of Health AIDS Clinical Trails Group.

Case Control Study

For each patient with NSIP, two patients with Pneumocystis pneumonia were selected from the same treatment study and matched with the index case for age to within 3 yr, and for prior use of zidovudine (the only licensed antiretroviral therapy at the time these studies were conducted). The age of control subjects was determined from their medical record identification cards, and prior use of zidovudine was determined from a computerized list of medications for patients that was generated by the hospital pharmacy. Case controls were matched for these variables and selected from the same treatment study in order to provide control subjects with pneumonia of similar severity to that of patients with NSIP.

Thirty-nine clinical descriptors present when patients entered the treatment studies were evaluated and compared for patients with NSIP and Pneumocystis pneumonia. Differences between arterial blood gas values at baseline and 7 to 10 d later were also compared in the two groups. Data was not collected for subjects without Pneumocystis pneumonia after they were discharged from the hospital. Continuous variables in the two groups were compared with Student's t test, and categorical variables were compared through contingency tables (chi-square and Fisher's exact tests). Differences were considered significant if for values of p < 0.05.

Pathologic Diagnoses

Table 1 shows the pathologic diagnoses in the 67 patients who did not have Pneumocystis pneumonia based on the absence of P. carinii by silver staining and absence to the telltale intraalveolar eosinophilic exudate from the cell-wall matrix of P. carinii.


Number of Cases (n = 67)Histologic Description
16Nonspecific interstitial pneumonitis
16Normal histology
13Inadequate histology (< 30 alveoli)
 7Bronchial abnormalities:
 Acute or chronic bronchitis
 No alveolar or interstitial disease
 2Pyogenic pneumonia
 2Pulmonary lymphoid hyperplasia
 2Caseating granulomas*
 2Talc pneumonitis
 1Eosinophilic pneumonia
 1Pulmonary anthracosis
 1Squamous-cell dysplasia
 1Pulmonary hemosiderosis
 1Non-caseating granulomas*
 1Caseating granulomas with acid-fast bacilli
 1Cytomegalovirus pneumonia§

*Negative stains and cultures for fungi and acid-fast bacilli.

Cultures of lavage fluid grew Mycobacterium tuberculosis.

§Intranuclear inclusions and interstitial inflammatory changes were present; culture of biopsy grew cytomegalovirus.


NSIP, as strictly defined by the absence of histologic evidence of any disease that could result in an intersitital inflammatory reaction, was detected in 16 (24%) cases. This was the most common pathologic diagnosis and was found in 5% of the 351 patients with presumed Pneumocystis pneumonia. All cases of NSIP showed chronic interstitial inflammation, ranging from mild to severe. On a scale of 0 to 4, the average score for interstitial inflammation was 2.5. The second most consistent histologic feature of NSIP was type II pneumocyte hyperplasia, which was demonstrable in all cases and averaged 2.4 on a scale of 0 to 4. The inflammation of NSIP generally “spilled over” into the alveoli, with a severity ranging from plus/minus to moderate, and with an average score of 2.3. All but one of the cases of NSIP showed interstitial edema, ranging from plus/minus to moderate, with an average numerical grading score of 2.1. Ten of the cases of NSIP showed bronchial inflammation, which was generally mild, but was moderate in two cases and severe in one case. Seven cases showed interstitial fibrosis, which was plus/minus in one case and mild in the other six. Four cases showed alveolar edema, ranging from plus/minus to moderate. Only one of the cases showed alveolar hyaline membranes, and none showed vasculitis.

Normal histology was present in 16 other cases; this does not entirely exclude pulmonary disease, since the limited size of biopsies may have allowed diagnoses to be missed. The inadequate-histology category contained 13 cases. Although inadequacy was based on the presence of fewer than 30 total alveoli, there were in fact no diagnoses that could have been made from the limited amount of tissue available from slides of the sections in these cases.

The next most common diagnosis was acute and/or chronic bronchitis with normal lung parenchyma. Since all patients had either abnormal blood gas values or chest radiographs that showed infiltrates, the explanation for this finding is not clear. The “normal” parenchyma may represent sampling errors away from areas of ventilation–perfusion mismatch.

Sections from the remaining 15 cases showed a variety of abnormalities. The only biopsies for which the histologic diagnosis had unequivocal therapeutic implications involved a case of tuberculosis with caseating granulomas containing acid-fast bacteria (AFB) and a case that met the ACTG definition of CMV pneumonia. The patient with tuberculosis also had AFB in sputum, and thus the diagnosis of tuberculosis would have been established without bronchoscopy. The patient with clinical and pathologic evidence of CMV pneumonia had no retinitis on dilated funduscopic examination. Thus, transbronchial biopsy resulted in the diagnosis of only one treatable infection other than P. carinni that would have been diagnosed by noninvasive testing. However, long-term follow-up of the two patients with caseating and one patient with noncaseating granulomas without AFB and fungi was not available to permit a more confident assessment of whether these three patients also had pulmonary infections.

In situ Hybridization and Tissue PCR Results

In situ hybridization studies of the tissues from patients with NSIP showed no cells positive for EBV or CMV in any of the transbronchial biopsies studied. In situ hybridization showed rare cells positive for HIV in four of the 15 cases of NSIP, in neither of the biopsies from HIV-positive patients with PLH, and in one of three biopsies from HIV-positive patients with normal histology. The few cells positive for HIV generally appeared to be alveolar macrophages or alveolar lining cells (pneumocytes).

PCRs for the HIV gag and env genes showed variable degrees of positivity in 10 of the 15 cases of NSIP, weak reactivity in sections from both patients with PLH, and weak reactivity in one and strong reactivity in two of the three samples from HIV-positive patients with normal lung histology.

Case Control Analysis

Table 2 shows that patients with NSIP were generally comparable with patients who had Pneumocystis pneumonia. Risk factors for AIDS, prior AIDS-case-defining conditions, the proportion of patients with pulmonary symptoms and the duration of these symptoms, and physical findings were similar in the two groups. However, evaluation of other clinical descriptors did reveal significant differences (p < 0.05) between the two groups (Tables 2 and 3). At entry into the treatment studies, patients with NSIP weighed more (70.9 kg versus 63.2 kg), had lower serum LDH values (280 IU/L versus 432 IU/L), more often had normal LDH (69% versus 19% of patients), had higher serum albumin concentrations (3.8 g/dl versus 3.4 g/dl), had higher white blood cell counts (9.5 × 109/L versus 5.4 × 109/L), had appreciably higher CD4+ lymphocyte counts (492 cells/μl versus 57 cells/μl), and were less likely to have CD4+ cell counts < 200/μL (31% versus 97% of patients), with p < 0.05 for each comparison. Moreover, patients with NSIP as a group had less extensive infiltration on chest radiographs, (i.e., they more often had only one or two quadrants involved, as compared with diffuse involvement of multiple lobes in patients with Pneumocystis pneumonia; p = 0.003).


FeaturesNSIP (n = 16) Pneumocystis Pneumonia (n = 32)
Sex, male/female16/032/0
Age, yr + 1 SD33.9 ± 7.3       33.5 ± 7.6
 White, non-Hispanic 815
 Hispanic 5 8
 African–American 3 8
 Asian 0 1
Risk factors for AIDS
 IVDU 1 2
 Heterosexual 0 1
 Transfusion 1 0
Prior AIDS diagnosis
Pneumocystis pneumonia 1 7
 Kaposi's sarcoma 2 7
 Esophageal candidiasis 2 1
 CMV retinitis 0 1
 CMV colitis 1 1
 Cryptococcal meningitis 0 1
History of fever11 20
Days of fever 16.5 ± 44.0         13.0 ± 16.0
Days of cough 12.8 ± 13.6        18.8 ± 16.5
Days of dyspnea 14.9 ± 18.3        15.7 ± 14.9
Days of chest pressure  9.1 ± 14.9        5.0 ± 6.5
Weight, kg 70.9 ± 13.4        63.2 ± 10.0
Admission temperature100.9 ± 2.1100.5 ± 1.8
Respiratory rate, per min26 ± 7         27 ± 10
Oral thrush 922
Lymphadenopathy 3 5
Rales 816

Definition of abbreviation: IVDU = intravenous drug use.


Test ResultNSIP (n = 16) Pneumocystis Pneumonia (n = 32)
Lactate dehydrogenase, IU/L   280 ± 113432 ± 141*
LDH abnormal, > 300 IU/L526
Serum albumin, g/dl  3.8 ± 0.63.4 ± 0.5
Serum hemoglobin, g/dl11.9 ± 2.1        11.9 ± 1.8
Total WBC count, ×103/μl  9.5 ± 5.8         5.4 ± 2.5
Neutrophil count, ×103/μl  6.3 ± 4.0         4.0 ± 2.1
Platelet count, ×103/μl   273 ± 79     277 ± 112
CD4 lymphocyte count, per μl   492 ± 82857 ± 60§
No. with CD4+ counts, < 200/mm3 531
Alkaline phosphatase, IU/L  135 ± 123           101 ± 64
Alanine aminotransferase, IU/L  32 ± 15      40 ± 35
PaO2 , mm Hg73.3 ± 7.0        69.2 ± 14.9
Pco 2, mm Hg32.1 ± 3.9        33.7 ± 4.4
(A–a)Do 2, mm Hg35.0 ± 7.8        38.6 ± 14.0
Chest radiographic abnormalities**
 Diffuse interstitial625
 Diffuse alveolar1 0
 < Two quadrants involved8 6
 Apical infiltrates0 1
 Pleural effusions0 0
 Hilar adneopathy0 0
 None    1 0

*p = 0.0002.

p = 0.0001.

p = 0.0048.

§p = 0.006.

p = 0.000001.

**p = 0.003.

Response to Therapy

Of the 16 patients with NSIP, eight were initially randomized to receive treatment with “blinded” trimetrexate or intravenous trimethoprim–sulfamethoxazole (ACTG 029/031), six to open label therapy with clindamycin plus primaquine (ACTG 040), one to blinded therapy with aerosolized pentamidine or intravenous trimethoprim–sulfamethoxazole (ACTG 044), and one to open label trimethoprim–sulfamethoxazole without prednisone (CCTG 509). Once the results of stains of BAL and transbronchial biopsy specimens were reported to have been negative for P. carinii, experimental therapy was discontinued. This occurred between treatment Days 3 and 10. In none of the 16 cases was open-label, nonstudy therapy for Pneumocystis pneumonia initiated during the respective hospitalization.

Of the 16 patients with NSIP, eight were sufficiently ill to have a blood-gas analysis done while breathing room air at 7 to 10 d after their baseline evaluation. The other eight patients were discharged from the hospital without repeat blood gas analyses. Thirty-one of 32 patients with Pneumocystis pneumonia had blood gas analysis done at Days 7 to 10 as directed by study design. Table 4 shows that both the PaO2 and (A–a)Do 2 improved after 7 to 10 d in patients with NSIP. Although these improvements appeared greater than for patients with Pneumocystis pneumonia, the differences in response between the two groups were of similar magnitude (p > 0.05 for both comparisons).


NSIP Pneumocystis Pneumonia
BaselineFollow-upDifferenceBaselineFollow-upDifferencep Value*
Number of patients
 with tests1683231
Blood gases
 PaO2 , mm Hg73 ± 7.085 ± 148.569 ± 15    75 ± 167.7> 0.05
(A–a)Do 2, mm Hg 35 ± 8.1     20 ± 14 −13.638 ± 14    30 ± 17     −8.1> 0.05

*Comparison of differences between baseline and follow-up at Days 7 to 10.

All tests were done with patients breathing room air for at least 20 min.

Normal value is ⩽ 14 mm Hg.

It is important to know the frequency of various conditions that may mimic Pneumocystis pneumonia, since treatment for P. carinii is often begun empirically, especially when tachypnea or hypoxemia is present. Results of our evaluation, at an institution where more than 200 cases of Pneumocystis pneumonia have been confirmed annually for almost a decade, and where practitioners have had considerable experience in diagnosing this infection, indicate that almost 20% of patients with typical features of Pneumocystis pneumonia do not have P. carinii confirmed by BAL and transbronchial biopsy. It would be desirable to avoid the expense and toxicities of therapies for Pneumocystis in such a sizable number of patients in whom the diagnosis cannot be confirmed.

The most common histologic diagnosis in patients with clinical and radiographic features mimicking Pneumocystis was NSIP. NSIP was detected in 24% of the 67 patients without Pneumocystis and 5% of the original 351 patients presumed to have Pneumocystis. The prevalence of NSIP in other studies of HIV-positive patients with pulmonary disease has varied from 11 to 38% (10-14). Selection of patient populations could explain these differences. In a study in which 41 (38%) of 110 patients with AIDS had NSIP, either biopsies showed concurrent pulmonary Kaposi's sarcoma or patients had received previous experimental therapies or had a history of intravenous drug use (in 28 of the 41 cases) (10). Diffuse alveolar damage and chronic interstitial disease are reported to occur in patients who use intravenous drugs (15). In addition, histologic changes in pulmonary Kaposi's sarcoma may resemble those of NSIP (16). By contrast, our patients were selected for evidence highly suggestive of Pneumocystis pneumonia. They had not received prior experimental therapy, and the diagnosis of NSIP was not made if there was histologic evidence of other diseases such as Kaposi's sarcoma. Thus, differences in study populations and a more stringent definition of NSIP may explain why the occurrence of NSIP in our patients was rarer than in several other reports (10, 11, 13).

Our case control analysis indicated that certain clinical descriptors might be helpful in distinguishing NSIP from Pneumocystis pneumonia. In particular, patients with NSIP were more likely to have less advanced HIV with better weight, serum albumin levels, and CD4+ lymphocyte counts. In addition, patients with NSIP appeared to have less lung inflammation, since they were more likely to have normal LDH values and fewer lobes with interstitial disease on chest radiographs. It is unlikely that these differences were caused by patient-selection bias, since controls were matched by enrollment in treatment studies based on severity of hypoxemia and ventilatory abnormality. In fact, PaO2 and (A–a)Do 2 at baseline were similar in the patients with NSIP and Pneumocystis pneumonia. Unfortunately, there was considerable overlap of features between the groups, and thus none of the clinical descriptors was sufficiently predictive to be of clinical value in diagnosis.

The natural history of NSIP is yet to be determined. However, in our study, results of follow-up blood-gas determinations done 7 to 10 d after the baseline evaluation showed appreciable improvement in at least half the patients with NSIP during empiric therapy for Pneumocystis pneumonia. The improvement was of similar magnitude to that which occurred in patients with documented Pneumocystis pneumonia. The other patients with NSIP were sufficiently improved clinically to have therapy for Pneumocystis discontinued, and their primary physicians did not feel it was necessary to document improvement in their blood gas values. In a series of seven patients with NSIP reported from Europe, a subacute course and spontaneous improvement was also noted (17). The reason for improvement in patients with NSIP is not clear. It is unlikely that more than a few patients with NSIP in our study could have had Pneumocystis pneumonia that was not detected by bronchoscopy, since the combined yield of BAL and transbronchial biopsy for identifying P. carinii is generally in excess of 95% (18). In addition, most patients had anti-Pneumocystis therapy discontinued after 3 to 5 d when the results of BAL and biopsies were reported not to show P. carinii. It is likely that most patients with Pneumocystis pneumonia would relapse quickly after such abbreviated therapy.

Currently, the etiology of NSIP is unknown. The cytokine profile of BAL fluid (BALF) samples from patients with NSIP differs from that of those with Pneumocystis pneumonia, suggesting that NSIP is a distinct entity and not merely a diagnosis of exclusion in cases in which P. carinii or other typical pathogens cannot be detected (19, 20). Whether HIV per se is the cause of NSIP in some patients is uncertain. HIV RNA in lung tissue was detected by in situ hybridization in rare cells of four of our 15 patients with NSIP, and in four of 46 patients with NSIP in another study (21). In our investigation, the more sensitive PCR testing for HIV gag and env DNA in tissue sections was positive in 10 of the 15 patients, but was also positive in the HIV control sections from patients with PLH and normal lung histology. One interpretation of these results is that HIV per se may, in some cases, result in an immunologic pulmonary reaction associated with histologic findings of NSIP.

We also sought evidence for EBV and CMV by in situ hybridization, but all 15 specimens from our patients with NSIP were negative for these viruses. Another herpes virus, human herpes virus-6 (HHV-6), has been associated with interstitial pneumonitis in some patients with AIDS (22), and DNA of this virus has been detected in high concentrations in the lungs of bone-marrow transplant recipients with pneumonia (23). Whether pneumonia due to HHV-6 would improve spontaneously and give the false impression that a patient was responding to therapy for Pneumocystis is unknown. It remains to be determined whether some cases of NSIP are due to HHV-6 or to other, yet undefined viruses.

There was a remarkable near absence of unequivocal treatable disorders (other than Pneumocystis pneumonia) diagnosed solely by transbronchial biopsy. Only the single case of CMV pneumonia would have been missed without transbronchial biopsy, although mycobacterial or fungal infection could not be entirely excluded in the three cases with lung granulomas. However, the study population was highly selected, since subjects with chest radiographs showing miliary patterns, hilar adenopathy, or pleural effusions were excluded from enrollment in the treatment studies. Because these findings may be associated with pulmonary tuberculosis, fungal infection, or Kaposi's sarcoma, it is likely that many patients with granulomatous infections or Kaposi's sarcoma were excluded from our study.

These results should not be construed as justification for not doing bronchoscopy in less selected patients, especially those failing to improve with empiric therapy. In a retrospective study of patients with negative sputum examinations for P. carinii, bronchoscopy resulted in a diagnosis of P. carinii pneumonia in more than half of the cases; the most frequent diagnoses were Pneumocystis pneumonia (192 cases), Kaposi's sarcoma (93 cases), tuberculosis (28 cases), and cryptococcosis (9 cases) (24). Thus, treatable neoplastic and granulomatous conditions may be detected more commonly in less selected populations than in our study subjects. In addition, the argument can be made that a negative bronchoscopy for P. carinii in many cases allows potentially toxic drugs to be discontinued. We therefore believe that the use of bronchoscopy still needs to be considered on a case-by-case basis, but our results suggest that the diagnostic yield of treatable causes of interstitial lung disease other than P. carinii in patients with HIV will be low in the absence of radiographic changes more suggestive of other diseases.

In summary, our results suggest that NSIP may be the most common diagnosis mimicking Pneumocystis pneumonia in patients with HIV, and may improve during empiric therapy for P. carinii pneumonia. Although NSIP more often occurred earlier in the course of HIV, when patients had better immunologic and nutritional function, and was more often associated with normal LDH levels, the overlap in presentations was too great for differentiating NSIP from Pneumocystis pneumonia on the basis of these clinical parameters alone. The study also demonstrated that other treatable conditions are unlikely to be detected by transbronchial biopsy when radiographs are carefully screened to exclude patients with findings more consistent with granulomatous causes of interstitial lung disease or pulmonary Kaposi's sarcoma.

1. Chan I. S. F., Neaton J. D., Saravolatz L. D., Crane L. R., Osterberger J.for the Community Programs for Clinical Research for AIDSFrequencies of opportunistic diseases prior to death among HIV-infected persons. AIDS9199511451151
2. Bigby T. D., Margolskee D., Curtis J. L., Michael P. F., Sheppard D., Hadley W. K., Hopewell P. C.The usefulness of induced sputum in the diagnosis of Pneumocystis carinii pneumonia in patients with the acquired immunodeficiency syndrome. Am. Rev. Respir. Dis.1331986515518
3. Kovacs J. A., Ng V. L., Masur H., Leoung G., Hadley W. K., Evans G., Lane H. C., Ognibene F. P., Shelhamer J., Parrillo J. E.Diagnosis of Pneumocystis carinii pneumonia: improved detection in sputum with use of monoclonal antibodies. N. Engl. J. Med.3181988589593
4. Leigh T. R., Hume C., Gazzard B., Parsons P., Husain O. A. N., Collins J. V.Sputum induction for diagnosis of Pneumocystis carinii pneumonia. Lancet21989205206
5. O'Brien R. F., Quinn J. L., Miyahara B. T., Lepoff R. B., Cohn D. L.Diagnosis of Pneumocystis carinii pneumonia by induced sputum in a city with a moderate incidence of AIDS. Chest951989136138
6. The National Institutes of Health—University of California Expert panel for Corticosteroids as Adjunctive Therapy for Pneumocystis pneumonia Consensus statement on the use of corticosteroids as adjunctive therapy for Pneumocystis pneumonia in the acquired immunodeficiency syndrome. N. Engl. J. Med.323199014511457
7. Gill P., Loureiro C., Bernstein-Singer M., Rarick M., Sattler R., Levine A.Clinical effect of glucocorticoids on Kaposi's sarcoma related to the acquired immunodeficiency syndrome (AIDS). Ann. Intern. Med.1101989937940
8. Salahuddin S. Z., Nakamura S., Biberfield P., Kaplan M. H., Markham P. D., Larrson L., Gallo R. C.Antigenic properties of Kaposi's sarcoma-derived cells after long-term culture in vitro. Science2421988430433
9. Saiki R. K., Bugawan T. L., Horn G. T., Mullis K. B., Erlich H. A.Analysis of enzymatically amplified β-globulin and HLA-DQα DNA with allele-specific oligonucleotide probes. Nature3241986163166
10. Suffredini A. F., Ognibene F. P., Lack E. E., Simmons J. T., Brenner M., Gill V. J., Lane H. C., Fauci A. S., Parrillo J. E.Nonspecific interstitial pneumonitis: a common cause of pulmonary disease in the acquired immunodeficiency syndrome. Ann. Intern. Med.1071987713
11. Batungwanayo J., Taelman H., Lucas S., Bogaerts J., Alard D., Kagame A., Blance P., Clerinx J., van de Perre P., Allen S.Pulmonary disease associated with human immunodeficiency virus in Kigali, Rwanda. Am. J. Respir. Crit. Care Med.149199415911596
12. Stover D. E., White D. A., Romano P. A., Gellene R. A., Robeson W. A.Spectrum of pulmonary disease associated with the acquired immune deficiency syndrome. Am. J. Med.781985429437
13. Barrio J. L., Harcup C., Horst J. B., Pitchenik A. E.Value of repeat fiberoptic bronchoscopies and significance of nondiagnostic bronchoscopic results in patients with the acquired immunodeficiency syndrome. Am. Rev. Respir. Dis.1351987422425
14. Prober C. G., Whyte H., Smith C. R.Open lung biopsy in immunocompromised children with pulmonary infiltrates. Am. J. Dis. Child.13819846063
15. Ramaswamy G., Jagadha V., Tchertkoff V.Diffuse alveolar damage and interstitial fibrosis in acquired immunodeficiency syndrome patients without concurrent pulmonary infection. Arch. Pathol. Lab. Med.1091985408412
16. Im-Hof V., Cerny T., Burkhardt A.Involvement of the trachea and lungs with Kaposi's sarcoma. J. Suisse Med.1181988134138
17. Griffiths M. H., Miller R. F., Semple S. J. G.Interstitial pneumonitis in patients infected with human immunodeficiency virus. Thorax50199511411146
18. Hopewell P. C.Pneumocystis carinii pneumonia: diagnosis. J. Infect. Dis.157198811151119
19. Denis M., Ghadirian E.Dysregulation of interleukin 8, interleukin 10, and interleukin 12 release by alveolar macrophages from HIV type 1-infected subjects. AIDS Res. Hum. Retrovir.10199416191627
20. Lipschik G. Y., Doerfler M. E., Kovacs J. A., Travis W. D., Andrawis V. A., Lawrence M. G., Dichter J. R., Ognibene F. P., Shelhamer J. H.Leukotriene B4 and interleukin-8 in human immunodeficiency virus-related pulmonary disease. Chest1041993763769
21. Travis W. D., Fox C. H., Devaney K. O., Weiss L. M., O'Leary T. J., Ognibene F. P.Lymphoid pneumonitis in 50 adult patients infected with the human immunodeficiency virus: lymphocytic interstitial pneumonitis versus nonspecific interstitial pneumonitis. Hum. Pathol.231992529541
22. Knox K. K., Carrigan D. R.Disseminated active HHV-6 infections in patients with AIDS. Lancet3431994557558
23. Cone R. W., Hackman R. C., Huang M. W., Bowden R. A., Meyers J. D., Metcalf M., Zeh J., Ashley R., Corey L.Human herpes virus 6 in lung tissue from patients with pneumonitis after bone marrow transplantation. N. Engl. J. Med.3291993156161
24. Huang L., Hecht F. M., Stansell J. D., Montanti R., Hadley W. K., Hopewell P. C.Suspected Pneumocystis carinii pneumonia with a negative induced sputum examination. Am. J. Respir. Crit. Care Med.151199518661871
Correspondence and requests for reprints should be addressed to Dr. Fred Sattler, LAC–USC Medical Center, Rand Schrader Clinic, Room 351, 1300 North Mission Road, Los Angeles, CA 90033.


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record