Fallow-up Bronchoalveolar Lavage in AIDS Patients with Pneumocystis carinii Pneumonia Pneumocystis carin;; Burden Predicts Early Relapse1 - 3
GREGORY COLANGELO, ROBERT R BAUGHMAN, MICHAEL N. DOHN, and PETER T. FRAME
Introduction
Pneumocystis carinii remains an important pathogen in patients with human immunodeficiency virus (HIV) infection. It is estimated that 80% of HIV-infected patients develop R cariniipneumonia at least once during the course of their disease (1-3). Many patients develop recurrent infections, and treatment failures are seen in 10 to 40070 of these episodes. Bronchoalveolar lavage (BAL) is a well-established technique to diagnose R carinii pneumonia (4-7). It is also known that R carinii can be recovered from patients with prior R carinii pneumonia during disease-free periods up to 60 days after an episode of acute pneumonia (8, 9). In one study examining repeat bronchoscopy in patients with HIV infection and worsening pulmonary symptoms, no benefit was noted from repeat bronchoscopy (10). In this study no attempt was made to assess whether there was clearance of the R carinii pneumonia. Quantitative techniques have been useful in assessing the response to therapy in animal models of R cariniipneumonia (11). Several groups have recently attempted to assess the amount of R carinii in the lung by examining BAL specimens (12, 13). We have developed a simple, reproducible method of estimating the amount, the burden of R cariniiin the BAL specimen (14). Examination of the BAL fluid also allows one to estimate the inflammatory response to R cariniiinfection. It has been found that increased neutrophils in the initial BAL are associated with increased mortality (IS, 16). Information regarding the changes in inflammatory response has not been previously reported. We studied patients with R carinii pneumonia after 21 days of therapy. The BAL findings were compared to outcome. There were three groups of patients: those who responded to therapy
SUMMARY We performed an analysis of the value of repeat bronchoalveolar lavage (BAL) at 21 days to identify patients at risk for early relapse with Pneumocystis car/nll pneumonia. Patients with P. car/nil pneumonia and the acquired Immunodeficiency syndrome (AIDS) were asked to participate In this study. All patients had P. car/nll Identified on methenamine sliver stain of BAL fluid. BAL fluid was also stained with a modified Wrlght-Glemsa technique. The Wrlght-Glemsa stain was done to determine the cell differential count, and the number of P. car/nil clusters associated with 500 nucleated cells was used as an estimate of P. car/nll burden in the BAL. Initial and follow-up lavage was performed In 56 patients. Patients were classified based on their clinical response to antl-P. car/nil therapy at 21 days. Nonresponders were patients with persistent or worsening symptoms. Responders were patients who improved and had therapy discontinued. Responders were further classified as responders with relapse If P. car/nil pneumonia recurred within 6 months of the initial episode or responders without relapse if they remained disease free during the follow-up period. Responders without relapse reduced P. car/nll cluster counts more than 50% In 24 of 25 cases. In responders with relapse P. carlnll cluster counts were unchanged. The responders as a group had a significant decrease In the percentage of neutrophlls In the BAL, with only 2 of 32 stili having Increased neutrophlls In the follow-up lavage compared to 17 of 24 nonresponders (p < 0.0001). Nonresponders were classified as those in whom an alternative reason for respiratory failure was found at follOW-Up lavage (14 patients) and those who only had P. car/nil (10 patients). Of the 10patients with P. car/nll alone all had more than 50% of the initial amount of P. car/nil found in the follow-up lavage. We conclude that BAL at 21 days can be useful In Identifying patients at Increased risk for early relapse and In evaluating patients not responding to antl-P. car/nll therapy. AM REV RESPIR DIS 1991; 143:1067-1071
and did well for the 6-month study period, those who responded to therapy but developed relapse of R carinii within 6 months, and those who failed to respond to therapy, some of whom had a new pathogen identified in their lavage.There were distinct differences in the amount of R cariniifound in the BAL of patients who responded to therapy versus those failing therapy or who would subsequently relapse. Methods At our institution all HIV-infected patients with pulmonary symptoms suggestive of R carinii pneumonia undergo bronchoscopy with HAL. The first part of the study was an 8-month period during which all 40 patients who had R carinii pneumonia documented by methenamine silver stain of bronchoscopy specimen were enrolled in the study. All patients who were still alive 21 days after initial diagnosis were approached regarding repeat bronchoscopy 3 to 4 wk after initial bronchoscopy. In the second part of the study, all
patients who underwent repeat bronchoscopy and BAL 21 to 28 days after an initial diagnostic lavage were analyzed. Patients underwent repeat bronchoscopy because of failure to respond to therapy or because they were part of an investigational study of anti-P cariniitherapy. Those patients who agreed to repeat bronchoscopy gave written consent of a protocol approved by our institutional review board. Patients underwent bronchoscopy initially and at follow-up using a standardized protocol, and the lavage specimen was handled as previously described (17).All initial speci-
(Received in original form June 8, 1990 and in revised form November 26, 1990) 1 From the Department of Medicine, University of Cincinnati Medical Center, Cincinnati, Ohio. 2 Supported in part by Grant AI-25897 from the National Institutes of Health. 3 Correspondence and requests for reprints should be addressed to Dr. Robert P. Baughman, 231 Bethesda Avenue, Mail Location 564, Cincinnati, OH 45267-0564.
1067
1068
mens were sent for cytology and fungal, mycobacterial, microbiologic, and viral culture. Patients who were nonresponders had follow-up BAL fluid sent for the same studies. For the responders the follow-up BAL was studied for infection using only the WrightGiemsa stain. Lavage was performed in the right middle lobe unless there was a localized infiltrate on chest roentgenogram. Follow-up lavages were done in the same area as the initial lavage. Aliquots (200 to 400 JlI) from the initial and follow-up BAL specimens were spun onto glass slides using a cytocentrifuge (Cytospin'" II; Shandon, Sewickley,PAl. The slides were air dried and then stained using a modified Wright-Giemsa stain (Diff-Quiks; American Scientific Products, McGaw Park, IL). Differntial cell counts were performed on 200 nucleated cellsper slide using standard criteria (18). The number of clusters of R carinii in relation to 500 nucleated cells was noted for each slide by observers blinded to the patient's clinical status (14, 19). Patients received a total of 14 to 21 days of one of the following regimens: trimethoprim-sulfamethoxazole, trimetrexate with leukovorin rescue, or inhaled or intravenously administered pentamidine. In some cases patient regimens were changed because of the toxicity of the antimicrobial agent or failure to respond to therapy within the first 2 wk of treatment. All surviving patients were reevaluated at 21 days with chest roentgenogram, arterial blood gas (if patient agreed), and clinical evaluation. All patients were followed for at least 6 months after the repeat BAL. All patients who survived the episode of R carinii pneumonia were treated with aerosolized pentamidine or oral trimethoprim-sulfamethoxazole prophylaxis. Patients on trimethoprim-sulfamethoxazole prophylaxis were usually changed to aerosol pentamide after 1to 3 months of treatment because of drug toxicity. For part 1,patients who wereenrolled in the study but who refused repeat lavage also were followed as outpatients and wereincluded in the analysis. Patients were classified at the time of the repeat lavage on the basis of symptoms, chest roentgenogram, and arterial blood gases into one of the following groups: (1) nonresponders, patients with clinical and laboratory signs of persistent or new pulmonary disease; and (2)responders, patients with clinical and laboratory signs of resolution of their pneumonia. After completion of therapy the patients werefollowedfor 6 months, and these responders fell into two categories: (1) responders with relapse patients who developed relapse, within the next 6 months; and (2) responders without relapse, patients who remained disease free for at least 6 months after therapy. There was also a group who died within 21 days of beginning therapy and did not have repeat lavage; they were considered early deaths.
Statistical Analysis Data were analyzed using a Mann-Whitney
COLANGELO, BAUGHMAN, DOHN, AND FRAME
U test for comparisons between groups, as the number of clusters were not normally distributed. In some cases chi-square analysis with Yare's correction factor for small numbers was used. A p value of less than 0.05 was considered statistically significant.
Results
Part 1 A total of 40 HIV-infected patients were diagnosed as having P. carinii pneumonia over an 8-month period. Of the 40, 15patients did not undergo follow-up lavage at 21 days. Of these, 10 patients refused the procedure and all were alive without disease at the 6-month followup. Additionally, 5 patients died before the 21-day follow-up lavage could be performed. The remaining 25 patients underwent follow-up BAL at 21 days. There were7 nonresponders and 18responders. Of the 18 responders 6 relapsed within 6 months of initial therapy. There was no significant difference in the study groups with respect to age, chest roentgenogram, or A-a gradient at the time of initial BAL, although there was a tendency for the nonresponders and early deaths to have a larger A-a gradient (table 1). The results of P. carinii cluster counting are shown in table 2. There was no significant difference between the number of clusters of P. carinii seen in the different groups, except for the five patients who died within 21 days of their initial lavage. In this group four of five had more than 20 clusters per 500 nucleated cells, and the group as a whole had significantly more P. cariniithan the other four groups. The results of the 25 repeat BAL are also shown in table 2. For the nonresponders and the responders without relapse there was a significant fall in the number of clusters of P. carinii found at repeat bronchoscopy. For the responders with relapse, however, the number of clusters of P. carinii did not fall. For the responders without relapse, the
P. carinii clusters decreased by more than 50070 in all but one case. In responders with relapse, however, no patient had a reduction in R carinii clusters of more than 50070. The difference betweenthe two groups was significant using chi-square analysis (chi-square = 10.4, p < 0.01). Of the sevennonresponders, three died during hospitalization for P. carinii pneumonia more than 3 wk after initial bronchoscopy. In five patients, another pathogen was thought to be the cause for persistent disease (one each of Cryptococcus
neoformans, Histoplasma capsulatum, Klebsiella pneumoniae, Nocardiaasteroides, and cytology-positive cytomegalovirus (CMV). In one case the patient was thought to have only congestiveheart failure. This patient responded to diuretics without further antimicrobials and was discharged from the hospital. In only one case was worsening of the P. carinii believed to be the exclusive cause for treatment failure. In this patient there was a significant rise in the number of P. carinii clusters from 5 in 500 nucleated cells before therapy to 20 in 500 nucleated cells after therapy.
Part 2 Over a 4-yr period 237 episodes of P. carinii pneumonia have been diagnosed in AIDS patients at our institution. A total of 56 patients have undergone repeat BAL 21 to 28 days after initial diagnosis of P. carinii. The 30 responders were treated with a single agent for the whole course of therapy (trimethoprim-sulfamethoxazole, 15 patients; intravenously administered pentamidine, 5 patients; aerosol pentamidine, 3 patients; and trimetrexate, 7 patients). The remaining responders and nonresponders weretreated with more than one agent. Based on the 6-month follow-up the 32 responders were divided into two groups: responders without relapse (25 patients) and responders with relapse (7
TABLE 1 CHARACTERISTICS OF PATIENTS STUDIED IN PART 1
Follow-up Study With Nonresponders Responders Without relapse With relapse Without Alive Early death
Age·
A-a Gradient
Deathst (%)
7
35 (26-55)
61 (21-80)
3 (43)
12 6
36 (24-48) 32 (25-43)
36 (26-52) 30 (12-51)
0(0) 1 (17)
10 5
32 (24-45) 39 (25-63)
46 (1-65) 67 (31-71)
0(0) 5 (100)
Number
• Median (range). Number (percentage) patients who died within 6 months of episode.
t
1069
FOLLOW-UP HAL IN P. CARINII PNEUMONIA
TABLE 2
500
NUMBER OF P. carinii CLUSTERS ASSOCIATED WITH 500 NUCLEATED CELLS Number of Patients
Follow-up Study
Follow-up BAL 50
With Nonresponders Responders Without relapse With relapse Without Alive Early death
..•.
100
Initial BAL *
7
5 (1-24)
1 (0-21)t
12
o (0-11)t
6
3 (1-9) 4 (1-31)
10 5
4 (0-10)* 32 (1-44)§
4 (1-20)
.....
••
10
* Median (range).
less than initial BAL, p < 0.01. P. carin;; seen on all slides when entire slide reviewed. § Significantly different from other groups (ANOVA), p < 0.001.
t Significantly
III.
*
patients). The initial and follow-up P. cariniicluster counts for both groups are shown in figure 1. Patients were considered nonresponders based on the clinical situation, including persistent fever, hypoxemia, or chest roentgenogram pattern. The nonresponders were divided into two groups based on their lavage findings. A total of 14patients had a new pathogen or condition (congestive heart failure) found at repeat bronchoscopy. These included fungal pneumonia (2 patients), routine bacterial infection (7 patients), nocardiosis (1 patient), CMV (3 patients), and congestive heart failure (1 patient). The fungal, routine bacterial, and nocardial infections were all determined by culture of the bronchoalveolar fluid, with only 1 case of fungal pneumonia detected by culture from the first lavage. The 3 cases of CMV were patients with cytologic evi-
Without Relapse
dence of CMV in the BAL fluid. In 10 nonresponders P. carinii alone was found in follow-up BAL fluid. The initial and follow-up cluster counts of the two nonresponder groups are shown in figure 2. The percentage of P. carinii in the initial BAL found in the follow-up BAL is shown in figure 3. Of the 18patients who had more than 50070 of the initial P. carinii number in the follow-up lavage, 17 were either nonresponders with P. carinii alone (10 patients) or responders with relapse (7 patients). Shown in table 3 is the cellular differential count of the lavage cells. In the current study, an increased percentage of lymphocytes was seen in all groups, with the upper limit of normal in our laboratory of 10070 (16). The percentage of neutrophils of the initial BAL was elevated in all groups but not significantly differ-
~
~
~
~
I
'2 U :::J
U
:::J
Z
o
30
~:::J
20
~
(3
40
j
40
Z
30
Is
20
(3
~ 10
~110 0:1
0:1 =II:
=II:
Initial Initial
Follow-up
Without Relapse
Initial
Follow-up
With Relapse
Fig. 1. The 32 patients who were clinical responders were divided into two categories based on whether they relapsed or did not relapse in the next 6 months. The number of P. carinii clusters seen with 500 nucleated cells initially and after 3 wk of antipneumocystis therapy are shown. The number of P. carinil clusters fell for the without relapse group (p < 0.01), but it did not fall to less than 50% of the initial counts for any of the with relapse group.
f...CI.!lJlIl Alone New Pathogen
Responders
50 50
With Relapse
Follow-up
.f. kIr.1nI.I Alone
Initial
Follow-up
New Pathogen
Fig. 2. The 24 patients who underwent repeat SAL because they were not responding to their antipneumocystis medication were divided into two groups, those with a new pathogen identified and those in whom only P. carinii was found. The number of P. carinii clusters associated with 500 nucleated cells for the initial and followup SAL are shown. Although the number of P. carinil clusters significantly fell in the new pathogen group (p < 0.01), it did not significantly change for the group with P. carinii alone.
Nonresponders
Fig. 3. The percentages of the initial cluster count of P. carinii found in the follow-up SAL are shown. Of the 18 patients who did not have a 50% or more reduction in their SAL cluster count, 17 were either responders with relapse or non responders with P. carinii alone.
ent between the four groups despite their different outcomes. There was no difference in the follow-up lavages of the two groups of responders. There was normalization of the percentage of neutrophils in those patients who were responders, with only 2 of 32patients stillhaving more than 5070 neutrophils, but 17 of 24 nonresponders still had more than 5070 neutrophils (chi-square = 25.5, p < 0.0001). Discussion
This study was designed to evaluate the utility of follow-up BAL in HIV-infected patients with cytologically proven P. carinii pneumonia. A semiquantitative estimate of P. carinii burden was used to compare the response to therapy between the groups. Our technique for P. carinii cluster counting has been reported elsewhere(14) and has been found to be simple and reproducible. Other investigators have used different techniques to relate P. carinii burden to outcome in patients with P. carinii pneumonia. Greco and coworkers have related P. carinii cluster counts to A-a gradient in patients with P. carinii pneumonia and found a significant correlation between A-a gradient and P. carinii counts (12). On the other hand, Shivaram and colleagues (20) studied 14 patients with AIDS and P. cariniipneumonia and found no relationship between P. carinii casts in the BAL and hypoxemia. No follow-up data were presented for either of these studies. Limper and coworkers report the number of cysts per milliliter lavage fluid (13). They did not find a correlation between cyst number and survival. There was a
1070
COLANGELO, BAUGHMAN, DOHN, AND FRAME
ly to have persistent P. carinii on transbronchial biopsy and cleared P. carinii more slowly than patients with lower P. % Lymphocytes % Neutrophils· Number of cariniiscores. However, the relationship Initial Follow-up Initial Follow-up Patients of P. carinii burden to survival failed to reach statistical significance. Responders 1 (0-10)t 9 (1-23) 13 (0-46) 12 (1-5) 25 Without relapse It is known that patients with P. carinii 23 (3-45) 16 (8-4) 2 (1-42) 2 (1-45) 7 With relapse pneumonia are at risk for repeated epiNonresponders sodes. In this study, all patients who 25 (0-45) 14 (1-35) 11 (1-4) 12 (8-36) P. carinii alone 8 relapsed were on aerosol prophylaxis at 14 (1-93) 6 (1-23) 7 (1-3) 3 (1-78) 16 Other pathogens the time of relapse. However, there have • Mean (range). been no previous reports of P. cariniioust Significantly differs from initial, p < 0.01. den after clinically successful therapy as an indicator of risk for early relapse. Our weak correlation between increased cyst separation of the groups difficult (10). data suggest that patients at risk for earnumber and a higher arterial oxygen ten- Our cohort study used a prospective de- ly relapse with P. carinii pneumonia can sion in a group that included both AIDS sign and evaluated only cases of cytolog- be identified by comparing P. carinii clusand non-AIDS patients with P. carinii ically proved P. carinii pneumonia, which ter counts at initial and follow-up lavage. Patients who did not respond to therpneumonia. They also found that pa- allowed a well-defined population of patients with AIDS had significantly more tients and subgroups. The study by Al- apy clinically for P. carinii pneumonia P. cariniicystsand fewerneutrophils than legra and coworkers was designed to had reduced P. carinii cluster counts by non-AIDS patients with P. carinii evaluate the efficacy of trimetrexate, and more than 50% in 14 of 24 cases. In the the authors did not comment on repeat 14 patients who had reduced P. carinii pneumonia. In our study there was no relationship lavage or P. carinii burden in prognosis clusters by more than 50070, an additionbetween P. carinii burden and A-a gra- (21). There was no attempt by the Allegra al, treatable pathogen was identified in dient at entry into the study. Neither group to identify alternative pathogens 10 cases. Only one of these pathogens previous study related the P. carinii bur- or relate P. carinii burden to the risk of was found on the initial lavage. Of the remaining 4 cases 1 had congestive heart den to the risk of relapse or evaluated relapse. The first part of this study was to con- failure responding to diuretics and 3 had the response to therapy with follow-up firm that P. carinii cluster counts could cytologicevidence of CMV.Although the BAL. Repeat BAL also has been used by oth- be used when examining a nonselected presence of CMV in the culture of BAL er investigators. Barrio and colleagues group of patients. Based on this cohort specimens is common in AIDS patients retrospectively examined HIV-infected study, in which the outcome of all pa- (22, 23), cytologic changes caused by patients who underwent serial bronchos- tients with P. cariniipneumonia was fol- CMV can be associated with hypoxemia copy in the evaluation of persistent or lowed, we used the 50% decrease in P. and fever (23), criteria for considering worsening pulmonary infiltrates or wor- carinii cluster counts as a cutoff to patients in this study as nonresponders. sening clinical status (10). In the group differentiate between the various groups There is no clear-cut effective treatment reexamined between 1 and 30 days, P. of P. carinii pneumonia patients under- for CMV pneumonitis; therefore 11 of carinii pneumonia was present in only 9 going repeat lavage. 14(79%) patients with more than a 50070 Patients who clinically responded to decrease in P. carinii were found to have of 21 (43%) of the initial lavages. In follow-up bronchoscopy only one patient therapy for P. carinii pneumonia fell in- treatable cause of nonresponse. (5070) had a new, treatable pathogen. In to two groups: those with and without Previous data have demonstrated a this group, Barrio could not demonstrate relapse with P. carinii pneumonia dur- high mortality in patients with P. carinii a benefit to early, repeat bronchoscopy. ing the first 6 months from the time of pneumonia not responsive to initial therAllegra and coworkers used serial lavage therapy. The group of patients who de- apy (24). All 10 nonresponders with onin the assessment of the safety and effi- veloped early relapse were unable to sig- ly P. carinii identified in the follow-up cacy of trimetrexate in the treatment of nificantly reduce their P. carinii burden lavage were treated with a new antipneuP. carinii pneumonia (21). A gross esti- as determined by P. carinii cluster count- mocystis therapy (usually intravenously mate of P. carinii burden was made by ing. The group of patients who did not administered pentamidine). However, 6 a blinded observer and recorded on a relapse had reduced P. carinii cluster of these patients died from this episode scale of 1 to 4. There were a significant counts by more than 50% in 24 of 25 of P. carinii pneumonia. number of relapses. However, the pa- cases. Shelhamer and colleagues have In the original cohort of 40 patients tients were not specifically characterized demonstrated persistent P. carinii in pa- there were5 patients who died of P. carinii tients with a history of P. carinii pneu- pneumonia within 3 wk of presentation with respect to P. carinii burden. Both studies differ significantly from monia but did not relate this to the risk and did undergo repeat BAL. These 5 pathe current investigation. In only nine of of relapse (8). Brenner and coworkers tients had significantly higher initial P. Barrios' patients was P. carinii pneumo- demonstrated worse long-term survival carinii cluster counts than patients who nia found on the initial lavage. They did in the group of patients with persistent survived their episode of P. cariniipneunot use BAL in either initial or follow-up P. carinii on transbronchial biopsy 3 wk monia. Although the number of patients examinations. Rather, bronchial wash- after therapy for P. carinii pneumonia (9). is small, this trend suggests that P. carinii ings, brushings, and trans bronchial biop- The cause of death in this group was not burden may be an indicator of prognosies wereused. No bacterial cultures were reported, however. They also noted that sis in patients with P. carinii pneumonia. performed. Also, the retrospective design patients with higher estimates of P. carinii It has been found that patients with eliminatespatients who improved, making burden (graded 0 to 3 +) were more like- HIV infection with P. carinii pneumoTABLE 3
SUMMARY OF BAL FINDINGS IN PATIENTS UNDERGOING FOLLOW-UP LAVAGE
1071
FOLLOW·UP SAL IN P. CARIN" PNEUMONIA
nia routinely have an increased percentage of lymphocytes in their BAL fluid (25, 26). It has also been shown that an increased percentage of neutrophils in the BAL fluid, as was seen in the nonresponders, is associated with an increased mortality (15, 16),although some believe this can be reversed by aggressive therapy (27). Information on the followup lavages has not been previously reported. However, it is known that an increased percentage of lymphocytes can be found in the BAL fluid of HIVinfected patients without R carinii (28). We found that the percentage of neutrophils fell significantly in those patients treated successfully, but the nonresponders still had a high percentage of neutrophils at follow-up lavage. This suggests that persistent neutrophilia in a follow-up BAL supports nonresponse. For the responders, neither the neutrophil nor lymphocyte percentage could identify those patients at high risk for early relapse. It should be noted that only three patients in this study weretreated with aerosol pentamidine for acute infection. It has recently been reported that lavage is less sensitive for R carinii pneumonia in patients on aerosol pentamidine (29). The conclusions of this study in assessing the response to therapy may not be accurate for all patients with R carinii pneumonia, including those patients treated with aerosol pentamidine for acute infection. In summary, we have used follow-up BAL and a semiquantitative clustercounting technique in HIV-infected patients with R cariniipneumonia to identify patients at high risk for developing early relapse. This may be useful in evaluating new antipneumocystis agents or deciding about various prophylactic regimens. It should be pointed out that all our early relapsers were on aerosol pentamidine at the time of relapse. We also found follow-up lavage useful to identify previously unsuspected pathogens in patients not responding to treatment for R carinii pneumonia. In a small number of patients who died early in the course of R carinii pneumonia, a significantly higher R carinii cluster count was found. Acknowledgment
The writers thank Dr. Robert Loudon for his
review and Mackie Price for her assistance in preparing this manuscript. References 1. Masur H, Lane C, Kovacs J, Allegra C, Edman J. Pneumocystis pneumonia: from bench to clinic: NIH conference. Ann Intern Med 1989; 111:813-26. 2. Murray JF, Garay SM, Hopewell PC, Mills J, Snider GL, Stover DE. Pulmonary complications of the acquired immunodeficiency syndrome: an update. Report of the Second National Heart, Lung and Blood Institute Workshop. Am Rev Respir Dis 1987; 135:504-9. 3. Murray JF, Felton CP, Garay SM, et al. Pulmonary complications of the acquired immunodeficiency syndrome. N Engl J Med 1984; 310:1682-8. 4. Broaddus C, Dake MD, Stulbarg MS, et al. Bronchoalveolar lavage and transbronchial biopsy for the diagnosis of pulmonary infections in the acquired immunodeficiency syndrome. Ann Intern Med 1985; 102:747-52. 5. Golden JA, Hollander H, Stulbarg MS, Gamsu G. Bronchoalveolar lavage as the exclusive diagnostic modality for Pneumocystis carinii pneumonia: a prospective study among patients with acquired immunodeficiency syndrome. Chest 1986; 90:18-22. 6. Stover DE, White DA, Romano PA, Gellene RA. Diagnosis of pulmonary disease in acquired immune deficiency syndrome (AIDS): role of bronchoscopy and bronchoalveolar lavage. Am Rev Respir Dis 1984; 130:659-62. 7. Stover DE, Zaman MB, Hajdi SI, Lange M, Gold J, Armstrong D. Bronchoalveolar lavage in the diagnosis of diffuse pulmonary infiltrates in the immunosuppressed host. Ann Intern Med 1984; 101:1-7. 8. Shelhammer JH, Ognibene FP, Macher AM, et al. Persistence of Pneumocystis carinii in lung tissue of acquired immunodeficiency syndrome patients treated for pneumocystis pneumonia. Am Rev Respir Dis 1984; 130:1161-5. 9. Brenner M, Ognibene FP, Lack EE, et al. Prognostic factors and life expectancy of patients with acquired immunodeficiency syndrome and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1987; 136:1199-206. 10. Barrio JL, Harcup C, Baier JH, Pitchenik AB. Value of repeat fiberoptic bronchoscopies and significance of non diagnostic bronchoscopic results in patients with the acquired immunodeficiency syndrome. Am Rev Respir Dis 1987; 135:422-5. 11. Kim CK, Foy JM, Cushion MT, et al. Comparison of histologic and quantitative techniques in evaluation of therapy for experimental Pneumocystis carinii pneumonia. Antimicrob Agents Chemother 1987; 31:197-201. 12. Greco MJ, Heimann A, Steibigel RT, Smaldone GC. Severity of Pneumocystis carinii pneumonia (PCP) correlates with number of organisms recovered from lavage (abstract). Am Rev Respir Dis 1989; 139:AI49. 13. Limper AH, Offord KD, Smith TF, Martin WJ. Pneumocystis cariniipneumonia: differences in lung parasite number and inflammation in patients with and without AIDS. Am Rev Respir Dis 1989; 140:1204-9. 14. Baughman RP, Strohofer S, Colangelo G,
Frame PT. A semi-quantitative technique for estimating Pneumocystis cariniiburden in the lung. J Clin Microbiol 1990; 28:1425-7. 15. Mason GR, Hashimoto CH, Dickman PS, Foutty LF, Cobb CJ. Prognostic implications of bronchoalveolar lavage neutrophilia in patients with Pneumocystis carinii pneumonia and AIDS. Am Rev Respir Dis 1989; 139:1336-42. 16. Smith RL, El-Sadr WM, Lewis ML. Correlation of bronchoalveolar lavage cell populations with clinical severity of Pneumocystis carinii pneumonia. Chest 1988; 92:60-4. 17. Thorpe JE, Baughman RP, Frame PT, WesseIer TA, Staneck JL. Bronchoalveolar lavage for diagnosing acute bacterial pneumonia. J Infect Dis 1987; 155:855-61. 18. Baughman RP, Strohofer S, Kim CK. Variation of differential cell counts of bronchoalveolar lavage fluid. Arch Pathol Lab Med 1986; 110:341-3. 19. Tollerud OJ, Kim CK, Weseler TA, Baughman RP. Use of a rapid differential stain for identifying Pneumocystis carinii in bronchoalveolar lavage fluid. Chest 1988; 95:493-7. 20. Shivaram I, Wooten 0, Klapper P, Demetis S, Korlipara 0, Zaman M. Relationship of inflammatory cellsand organism load in lung lavage (BAL) to severity of Pneumocystis carinii pneumonia (PCP) (abstract). Am Rev Respir Dis 1990; 141:A270. 21. Allegra CJ, Chabner BA, Tuazon CA, et al. Trimetrexate for the treatment of Pneumocystis cariniipneumonia in patients with the acquired immunodeficiency syndrome. N Engl J Med 1987; 317:978-85. 22. Millar AB, Patou G, Miller RF, et al. Cytomegalovirus in the lungs of patients with AIDS. Respiratory pathogen or passenger? Am Rev Respir Dis 1990; 141:1474-7. 23. Miles PR, Baughman RP, Linnemamm CC. Cytomegalovirus in the bronchoalveolar lavage fluid of patients with AIDS. Chest 1990; 97:1072-6. 24. Haverkos H. PCP therapy project group. Assessment of therapy for Pneumocystiscarinii pneumonia. Am J Med 1984; 76:501-8. 25. Wallace JM, Barbers RG, Oishi JS, Prince H. Cellular and T-Iymphocyte subpopulation profiles in bronchoalveolar lavage fluid from patients with acquired immunodeficiency syndrome and pneumonitis. Am Rev Respir Dis 1984; 130:786-90. 26. White DA, Gellene RA, Gupta S, Cunningham-Rundles C, Stover DE. Pulmonary cell populations in the immunosuppressed patient: bronchoalveolar lavage findings during episodes of pneumonitis. Chest 1985; 88:352-9. 27. Savleda J, Gea J, Aran X, et al. Usefulness of neutrophilia in the bronchoalveolar lavage, effect of an aggressive therapeutic approach in patients with AIDS and Pneumocystis carinii pneumonia (abstract). Am Rev Respir Dis 1990; 141:A270. 28. Venet A, Clavel F, Israel-Biet D, et al. Lung in acquired immune deficiency syndrome: infectious and immunologic status assessed by bronchoalveolar lavage. Bull Eur Physiopathol Respir 1985; 21:535-43. 29. Jules-Elusee KM, Stover DE, Zaman MB, Bernard EM, White DA. Aerosolized pentamidine: effect on diagnosis and presentation of Pneumocystis carinii pneumonia. Ann Intern Med 1990; 112:750-7.
GREGORY COLANGELO, ROBERT R BAUGHMAN, MICHAEL N. DOHN, and PETER T. FRAME
Introduction
Pneumocystis carinii remains an important pathogen in patients with human immunodeficiency virus (HIV) infection. It is estimated that 80% of HIV-infected patients develop R cariniipneumonia at least once during the course of their disease (1-3). Many patients develop recurrent infections, and treatment failures are seen in 10 to 40070 of these episodes. Bronchoalveolar lavage (BAL) is a well-established technique to diagnose R carinii pneumonia (4-7). It is also known that R carinii can be recovered from patients with prior R carinii pneumonia during disease-free periods up to 60 days after an episode of acute pneumonia (8, 9). In one study examining repeat bronchoscopy in patients with HIV infection and worsening pulmonary symptoms, no benefit was noted from repeat bronchoscopy (10). In this study no attempt was made to assess whether there was clearance of the R carinii pneumonia. Quantitative techniques have been useful in assessing the response to therapy in animal models of R cariniipneumonia (11). Several groups have recently attempted to assess the amount of R carinii in the lung by examining BAL specimens (12, 13). We have developed a simple, reproducible method of estimating the amount, the burden of R cariniiin the BAL specimen (14). Examination of the BAL fluid also allows one to estimate the inflammatory response to R cariniiinfection. It has been found that increased neutrophils in the initial BAL are associated with increased mortality (IS, 16). Information regarding the changes in inflammatory response has not been previously reported. We studied patients with R carinii pneumonia after 21 days of therapy. The BAL findings were compared to outcome. There were three groups of patients: those who responded to therapy
SUMMARY We performed an analysis of the value of repeat bronchoalveolar lavage (BAL) at 21 days to identify patients at risk for early relapse with Pneumocystis car/nll pneumonia. Patients with P. car/nil pneumonia and the acquired Immunodeficiency syndrome (AIDS) were asked to participate In this study. All patients had P. car/nll Identified on methenamine sliver stain of BAL fluid. BAL fluid was also stained with a modified Wrlght-Glemsa technique. The Wrlght-Glemsa stain was done to determine the cell differential count, and the number of P. car/nil clusters associated with 500 nucleated cells was used as an estimate of P. car/nll burden in the BAL. Initial and follow-up lavage was performed In 56 patients. Patients were classified based on their clinical response to antl-P. car/nil therapy at 21 days. Nonresponders were patients with persistent or worsening symptoms. Responders were patients who improved and had therapy discontinued. Responders were further classified as responders with relapse If P. car/nil pneumonia recurred within 6 months of the initial episode or responders without relapse if they remained disease free during the follow-up period. Responders without relapse reduced P. car/nll cluster counts more than 50% In 24 of 25 cases. In responders with relapse P. carlnll cluster counts were unchanged. The responders as a group had a significant decrease In the percentage of neutrophlls In the BAL, with only 2 of 32 stili having Increased neutrophlls In the follow-up lavage compared to 17 of 24 nonresponders (p < 0.0001). Nonresponders were classified as those in whom an alternative reason for respiratory failure was found at follOW-Up lavage (14 patients) and those who only had P. car/nil (10 patients). Of the 10patients with P. car/nll alone all had more than 50% of the initial amount of P. car/nil found in the follow-up lavage. We conclude that BAL at 21 days can be useful In Identifying patients at Increased risk for early relapse and In evaluating patients not responding to antl-P. car/nll therapy. AM REV RESPIR DIS 1991; 143:1067-1071
and did well for the 6-month study period, those who responded to therapy but developed relapse of R carinii within 6 months, and those who failed to respond to therapy, some of whom had a new pathogen identified in their lavage.There were distinct differences in the amount of R cariniifound in the BAL of patients who responded to therapy versus those failing therapy or who would subsequently relapse. Methods At our institution all HIV-infected patients with pulmonary symptoms suggestive of R carinii pneumonia undergo bronchoscopy with HAL. The first part of the study was an 8-month period during which all 40 patients who had R carinii pneumonia documented by methenamine silver stain of bronchoscopy specimen were enrolled in the study. All patients who were still alive 21 days after initial diagnosis were approached regarding repeat bronchoscopy 3 to 4 wk after initial bronchoscopy. In the second part of the study, all
patients who underwent repeat bronchoscopy and BAL 21 to 28 days after an initial diagnostic lavage were analyzed. Patients underwent repeat bronchoscopy because of failure to respond to therapy or because they were part of an investigational study of anti-P cariniitherapy. Those patients who agreed to repeat bronchoscopy gave written consent of a protocol approved by our institutional review board. Patients underwent bronchoscopy initially and at follow-up using a standardized protocol, and the lavage specimen was handled as previously described (17).All initial speci-
(Received in original form June 8, 1990 and in revised form November 26, 1990) 1 From the Department of Medicine, University of Cincinnati Medical Center, Cincinnati, Ohio. 2 Supported in part by Grant AI-25897 from the National Institutes of Health. 3 Correspondence and requests for reprints should be addressed to Dr. Robert P. Baughman, 231 Bethesda Avenue, Mail Location 564, Cincinnati, OH 45267-0564.
1067
1068
mens were sent for cytology and fungal, mycobacterial, microbiologic, and viral culture. Patients who were nonresponders had follow-up BAL fluid sent for the same studies. For the responders the follow-up BAL was studied for infection using only the WrightGiemsa stain. Lavage was performed in the right middle lobe unless there was a localized infiltrate on chest roentgenogram. Follow-up lavages were done in the same area as the initial lavage. Aliquots (200 to 400 JlI) from the initial and follow-up BAL specimens were spun onto glass slides using a cytocentrifuge (Cytospin'" II; Shandon, Sewickley,PAl. The slides were air dried and then stained using a modified Wright-Giemsa stain (Diff-Quiks; American Scientific Products, McGaw Park, IL). Differntial cell counts were performed on 200 nucleated cellsper slide using standard criteria (18). The number of clusters of R carinii in relation to 500 nucleated cells was noted for each slide by observers blinded to the patient's clinical status (14, 19). Patients received a total of 14 to 21 days of one of the following regimens: trimethoprim-sulfamethoxazole, trimetrexate with leukovorin rescue, or inhaled or intravenously administered pentamidine. In some cases patient regimens were changed because of the toxicity of the antimicrobial agent or failure to respond to therapy within the first 2 wk of treatment. All surviving patients were reevaluated at 21 days with chest roentgenogram, arterial blood gas (if patient agreed), and clinical evaluation. All patients were followed for at least 6 months after the repeat BAL. All patients who survived the episode of R carinii pneumonia were treated with aerosolized pentamidine or oral trimethoprim-sulfamethoxazole prophylaxis. Patients on trimethoprim-sulfamethoxazole prophylaxis were usually changed to aerosol pentamide after 1to 3 months of treatment because of drug toxicity. For part 1,patients who wereenrolled in the study but who refused repeat lavage also were followed as outpatients and wereincluded in the analysis. Patients were classified at the time of the repeat lavage on the basis of symptoms, chest roentgenogram, and arterial blood gases into one of the following groups: (1) nonresponders, patients with clinical and laboratory signs of persistent or new pulmonary disease; and (2)responders, patients with clinical and laboratory signs of resolution of their pneumonia. After completion of therapy the patients werefollowedfor 6 months, and these responders fell into two categories: (1) responders with relapse patients who developed relapse, within the next 6 months; and (2) responders without relapse, patients who remained disease free for at least 6 months after therapy. There was also a group who died within 21 days of beginning therapy and did not have repeat lavage; they were considered early deaths.
Statistical Analysis Data were analyzed using a Mann-Whitney
COLANGELO, BAUGHMAN, DOHN, AND FRAME
U test for comparisons between groups, as the number of clusters were not normally distributed. In some cases chi-square analysis with Yare's correction factor for small numbers was used. A p value of less than 0.05 was considered statistically significant.
Results
Part 1 A total of 40 HIV-infected patients were diagnosed as having P. carinii pneumonia over an 8-month period. Of the 40, 15patients did not undergo follow-up lavage at 21 days. Of these, 10 patients refused the procedure and all were alive without disease at the 6-month followup. Additionally, 5 patients died before the 21-day follow-up lavage could be performed. The remaining 25 patients underwent follow-up BAL at 21 days. There were7 nonresponders and 18responders. Of the 18 responders 6 relapsed within 6 months of initial therapy. There was no significant difference in the study groups with respect to age, chest roentgenogram, or A-a gradient at the time of initial BAL, although there was a tendency for the nonresponders and early deaths to have a larger A-a gradient (table 1). The results of P. carinii cluster counting are shown in table 2. There was no significant difference between the number of clusters of P. carinii seen in the different groups, except for the five patients who died within 21 days of their initial lavage. In this group four of five had more than 20 clusters per 500 nucleated cells, and the group as a whole had significantly more P. cariniithan the other four groups. The results of the 25 repeat BAL are also shown in table 2. For the nonresponders and the responders without relapse there was a significant fall in the number of clusters of P. carinii found at repeat bronchoscopy. For the responders with relapse, however, the number of clusters of P. carinii did not fall. For the responders without relapse, the
P. carinii clusters decreased by more than 50070 in all but one case. In responders with relapse, however, no patient had a reduction in R carinii clusters of more than 50070. The difference betweenthe two groups was significant using chi-square analysis (chi-square = 10.4, p < 0.01). Of the sevennonresponders, three died during hospitalization for P. carinii pneumonia more than 3 wk after initial bronchoscopy. In five patients, another pathogen was thought to be the cause for persistent disease (one each of Cryptococcus
neoformans, Histoplasma capsulatum, Klebsiella pneumoniae, Nocardiaasteroides, and cytology-positive cytomegalovirus (CMV). In one case the patient was thought to have only congestiveheart failure. This patient responded to diuretics without further antimicrobials and was discharged from the hospital. In only one case was worsening of the P. carinii believed to be the exclusive cause for treatment failure. In this patient there was a significant rise in the number of P. carinii clusters from 5 in 500 nucleated cells before therapy to 20 in 500 nucleated cells after therapy.
Part 2 Over a 4-yr period 237 episodes of P. carinii pneumonia have been diagnosed in AIDS patients at our institution. A total of 56 patients have undergone repeat BAL 21 to 28 days after initial diagnosis of P. carinii. The 30 responders were treated with a single agent for the whole course of therapy (trimethoprim-sulfamethoxazole, 15 patients; intravenously administered pentamidine, 5 patients; aerosol pentamidine, 3 patients; and trimetrexate, 7 patients). The remaining responders and nonresponders weretreated with more than one agent. Based on the 6-month follow-up the 32 responders were divided into two groups: responders without relapse (25 patients) and responders with relapse (7
TABLE 1 CHARACTERISTICS OF PATIENTS STUDIED IN PART 1
Follow-up Study With Nonresponders Responders Without relapse With relapse Without Alive Early death
Age·
A-a Gradient
Deathst (%)
7
35 (26-55)
61 (21-80)
3 (43)
12 6
36 (24-48) 32 (25-43)
36 (26-52) 30 (12-51)
0(0) 1 (17)
10 5
32 (24-45) 39 (25-63)
46 (1-65) 67 (31-71)
0(0) 5 (100)
Number
• Median (range). Number (percentage) patients who died within 6 months of episode.
t
1069
FOLLOW-UP HAL IN P. CARINII PNEUMONIA
TABLE 2
500
NUMBER OF P. carinii CLUSTERS ASSOCIATED WITH 500 NUCLEATED CELLS Number of Patients
Follow-up Study
Follow-up BAL 50
With Nonresponders Responders Without relapse With relapse Without Alive Early death
..•.
100
Initial BAL *
7
5 (1-24)
1 (0-21)t
12
o (0-11)t
6
3 (1-9) 4 (1-31)
10 5
4 (0-10)* 32 (1-44)§
4 (1-20)
.....
••
10
* Median (range).
less than initial BAL, p < 0.01. P. carin;; seen on all slides when entire slide reviewed. § Significantly different from other groups (ANOVA), p < 0.001.
t Significantly
III.
*
patients). The initial and follow-up P. cariniicluster counts for both groups are shown in figure 1. Patients were considered nonresponders based on the clinical situation, including persistent fever, hypoxemia, or chest roentgenogram pattern. The nonresponders were divided into two groups based on their lavage findings. A total of 14patients had a new pathogen or condition (congestive heart failure) found at repeat bronchoscopy. These included fungal pneumonia (2 patients), routine bacterial infection (7 patients), nocardiosis (1 patient), CMV (3 patients), and congestive heart failure (1 patient). The fungal, routine bacterial, and nocardial infections were all determined by culture of the bronchoalveolar fluid, with only 1 case of fungal pneumonia detected by culture from the first lavage. The 3 cases of CMV were patients with cytologic evi-
Without Relapse
dence of CMV in the BAL fluid. In 10 nonresponders P. carinii alone was found in follow-up BAL fluid. The initial and follow-up cluster counts of the two nonresponder groups are shown in figure 2. The percentage of P. carinii in the initial BAL found in the follow-up BAL is shown in figure 3. Of the 18patients who had more than 50070 of the initial P. carinii number in the follow-up lavage, 17 were either nonresponders with P. carinii alone (10 patients) or responders with relapse (7 patients). Shown in table 3 is the cellular differential count of the lavage cells. In the current study, an increased percentage of lymphocytes was seen in all groups, with the upper limit of normal in our laboratory of 10070 (16). The percentage of neutrophils of the initial BAL was elevated in all groups but not significantly differ-
~
~
~
~
I
'2 U :::J
U
:::J
Z
o
30
~:::J
20
~
(3
40
j
40
Z
30
Is
20
(3
~ 10
~110 0:1
0:1 =II:
=II:
Initial Initial
Follow-up
Without Relapse
Initial
Follow-up
With Relapse
Fig. 1. The 32 patients who were clinical responders were divided into two categories based on whether they relapsed or did not relapse in the next 6 months. The number of P. carinii clusters seen with 500 nucleated cells initially and after 3 wk of antipneumocystis therapy are shown. The number of P. carinil clusters fell for the without relapse group (p < 0.01), but it did not fall to less than 50% of the initial counts for any of the with relapse group.
f...CI.!lJlIl Alone New Pathogen
Responders
50 50
With Relapse
Follow-up
.f. kIr.1nI.I Alone
Initial
Follow-up
New Pathogen
Fig. 2. The 24 patients who underwent repeat SAL because they were not responding to their antipneumocystis medication were divided into two groups, those with a new pathogen identified and those in whom only P. carinii was found. The number of P. carinii clusters associated with 500 nucleated cells for the initial and followup SAL are shown. Although the number of P. carinil clusters significantly fell in the new pathogen group (p < 0.01), it did not significantly change for the group with P. carinii alone.
Nonresponders
Fig. 3. The percentages of the initial cluster count of P. carinii found in the follow-up SAL are shown. Of the 18 patients who did not have a 50% or more reduction in their SAL cluster count, 17 were either responders with relapse or non responders with P. carinii alone.
ent between the four groups despite their different outcomes. There was no difference in the follow-up lavages of the two groups of responders. There was normalization of the percentage of neutrophils in those patients who were responders, with only 2 of 32patients stillhaving more than 5070 neutrophils, but 17 of 24 nonresponders still had more than 5070 neutrophils (chi-square = 25.5, p < 0.0001). Discussion
This study was designed to evaluate the utility of follow-up BAL in HIV-infected patients with cytologically proven P. carinii pneumonia. A semiquantitative estimate of P. carinii burden was used to compare the response to therapy between the groups. Our technique for P. carinii cluster counting has been reported elsewhere(14) and has been found to be simple and reproducible. Other investigators have used different techniques to relate P. carinii burden to outcome in patients with P. carinii pneumonia. Greco and coworkers have related P. carinii cluster counts to A-a gradient in patients with P. carinii pneumonia and found a significant correlation between A-a gradient and P. carinii counts (12). On the other hand, Shivaram and colleagues (20) studied 14 patients with AIDS and P. cariniipneumonia and found no relationship between P. carinii casts in the BAL and hypoxemia. No follow-up data were presented for either of these studies. Limper and coworkers report the number of cysts per milliliter lavage fluid (13). They did not find a correlation between cyst number and survival. There was a
1070
COLANGELO, BAUGHMAN, DOHN, AND FRAME
ly to have persistent P. carinii on transbronchial biopsy and cleared P. carinii more slowly than patients with lower P. % Lymphocytes % Neutrophils· Number of cariniiscores. However, the relationship Initial Follow-up Initial Follow-up Patients of P. carinii burden to survival failed to reach statistical significance. Responders 1 (0-10)t 9 (1-23) 13 (0-46) 12 (1-5) 25 Without relapse It is known that patients with P. carinii 23 (3-45) 16 (8-4) 2 (1-42) 2 (1-45) 7 With relapse pneumonia are at risk for repeated epiNonresponders sodes. In this study, all patients who 25 (0-45) 14 (1-35) 11 (1-4) 12 (8-36) P. carinii alone 8 relapsed were on aerosol prophylaxis at 14 (1-93) 6 (1-23) 7 (1-3) 3 (1-78) 16 Other pathogens the time of relapse. However, there have • Mean (range). been no previous reports of P. cariniioust Significantly differs from initial, p < 0.01. den after clinically successful therapy as an indicator of risk for early relapse. Our weak correlation between increased cyst separation of the groups difficult (10). data suggest that patients at risk for earnumber and a higher arterial oxygen ten- Our cohort study used a prospective de- ly relapse with P. carinii pneumonia can sion in a group that included both AIDS sign and evaluated only cases of cytolog- be identified by comparing P. carinii clusand non-AIDS patients with P. carinii ically proved P. carinii pneumonia, which ter counts at initial and follow-up lavage. Patients who did not respond to therpneumonia. They also found that pa- allowed a well-defined population of patients with AIDS had significantly more tients and subgroups. The study by Al- apy clinically for P. carinii pneumonia P. cariniicystsand fewerneutrophils than legra and coworkers was designed to had reduced P. carinii cluster counts by non-AIDS patients with P. carinii evaluate the efficacy of trimetrexate, and more than 50% in 14 of 24 cases. In the the authors did not comment on repeat 14 patients who had reduced P. carinii pneumonia. In our study there was no relationship lavage or P. carinii burden in prognosis clusters by more than 50070, an additionbetween P. carinii burden and A-a gra- (21). There was no attempt by the Allegra al, treatable pathogen was identified in dient at entry into the study. Neither group to identify alternative pathogens 10 cases. Only one of these pathogens previous study related the P. carinii bur- or relate P. carinii burden to the risk of was found on the initial lavage. Of the remaining 4 cases 1 had congestive heart den to the risk of relapse or evaluated relapse. The first part of this study was to con- failure responding to diuretics and 3 had the response to therapy with follow-up firm that P. carinii cluster counts could cytologicevidence of CMV.Although the BAL. Repeat BAL also has been used by oth- be used when examining a nonselected presence of CMV in the culture of BAL er investigators. Barrio and colleagues group of patients. Based on this cohort specimens is common in AIDS patients retrospectively examined HIV-infected study, in which the outcome of all pa- (22, 23), cytologic changes caused by patients who underwent serial bronchos- tients with P. cariniipneumonia was fol- CMV can be associated with hypoxemia copy in the evaluation of persistent or lowed, we used the 50% decrease in P. and fever (23), criteria for considering worsening pulmonary infiltrates or wor- carinii cluster counts as a cutoff to patients in this study as nonresponders. sening clinical status (10). In the group differentiate between the various groups There is no clear-cut effective treatment reexamined between 1 and 30 days, P. of P. carinii pneumonia patients under- for CMV pneumonitis; therefore 11 of carinii pneumonia was present in only 9 going repeat lavage. 14(79%) patients with more than a 50070 Patients who clinically responded to decrease in P. carinii were found to have of 21 (43%) of the initial lavages. In follow-up bronchoscopy only one patient therapy for P. carinii pneumonia fell in- treatable cause of nonresponse. (5070) had a new, treatable pathogen. In to two groups: those with and without Previous data have demonstrated a this group, Barrio could not demonstrate relapse with P. carinii pneumonia dur- high mortality in patients with P. carinii a benefit to early, repeat bronchoscopy. ing the first 6 months from the time of pneumonia not responsive to initial therAllegra and coworkers used serial lavage therapy. The group of patients who de- apy (24). All 10 nonresponders with onin the assessment of the safety and effi- veloped early relapse were unable to sig- ly P. carinii identified in the follow-up cacy of trimetrexate in the treatment of nificantly reduce their P. carinii burden lavage were treated with a new antipneuP. carinii pneumonia (21). A gross esti- as determined by P. carinii cluster count- mocystis therapy (usually intravenously mate of P. carinii burden was made by ing. The group of patients who did not administered pentamidine). However, 6 a blinded observer and recorded on a relapse had reduced P. carinii cluster of these patients died from this episode scale of 1 to 4. There were a significant counts by more than 50% in 24 of 25 of P. carinii pneumonia. number of relapses. However, the pa- cases. Shelhamer and colleagues have In the original cohort of 40 patients tients were not specifically characterized demonstrated persistent P. carinii in pa- there were5 patients who died of P. carinii tients with a history of P. carinii pneu- pneumonia within 3 wk of presentation with respect to P. carinii burden. Both studies differ significantly from monia but did not relate this to the risk and did undergo repeat BAL. These 5 pathe current investigation. In only nine of of relapse (8). Brenner and coworkers tients had significantly higher initial P. Barrios' patients was P. carinii pneumo- demonstrated worse long-term survival carinii cluster counts than patients who nia found on the initial lavage. They did in the group of patients with persistent survived their episode of P. cariniipneunot use BAL in either initial or follow-up P. carinii on transbronchial biopsy 3 wk monia. Although the number of patients examinations. Rather, bronchial wash- after therapy for P. carinii pneumonia (9). is small, this trend suggests that P. carinii ings, brushings, and trans bronchial biop- The cause of death in this group was not burden may be an indicator of prognosies wereused. No bacterial cultures were reported, however. They also noted that sis in patients with P. carinii pneumonia. performed. Also, the retrospective design patients with higher estimates of P. carinii It has been found that patients with eliminatespatients who improved, making burden (graded 0 to 3 +) were more like- HIV infection with P. carinii pneumoTABLE 3
SUMMARY OF BAL FINDINGS IN PATIENTS UNDERGOING FOLLOW-UP LAVAGE
1071
FOLLOW·UP SAL IN P. CARIN" PNEUMONIA
nia routinely have an increased percentage of lymphocytes in their BAL fluid (25, 26). It has also been shown that an increased percentage of neutrophils in the BAL fluid, as was seen in the nonresponders, is associated with an increased mortality (15, 16),although some believe this can be reversed by aggressive therapy (27). Information on the followup lavages has not been previously reported. However, it is known that an increased percentage of lymphocytes can be found in the BAL fluid of HIVinfected patients without R carinii (28). We found that the percentage of neutrophils fell significantly in those patients treated successfully, but the nonresponders still had a high percentage of neutrophils at follow-up lavage. This suggests that persistent neutrophilia in a follow-up BAL supports nonresponse. For the responders, neither the neutrophil nor lymphocyte percentage could identify those patients at high risk for early relapse. It should be noted that only three patients in this study weretreated with aerosol pentamidine for acute infection. It has recently been reported that lavage is less sensitive for R carinii pneumonia in patients on aerosol pentamidine (29). The conclusions of this study in assessing the response to therapy may not be accurate for all patients with R carinii pneumonia, including those patients treated with aerosol pentamidine for acute infection. In summary, we have used follow-up BAL and a semiquantitative clustercounting technique in HIV-infected patients with R cariniipneumonia to identify patients at high risk for developing early relapse. This may be useful in evaluating new antipneumocystis agents or deciding about various prophylactic regimens. It should be pointed out that all our early relapsers were on aerosol pentamidine at the time of relapse. We also found follow-up lavage useful to identify previously unsuspected pathogens in patients not responding to treatment for R carinii pneumonia. In a small number of patients who died early in the course of R carinii pneumonia, a significantly higher R carinii cluster count was found. Acknowledgment
The writers thank Dr. Robert Loudon for his
review and Mackie Price for her assistance in preparing this manuscript. References 1. Masur H, Lane C, Kovacs J, Allegra C, Edman J. Pneumocystis pneumonia: from bench to clinic: NIH conference. Ann Intern Med 1989; 111:813-26. 2. Murray JF, Garay SM, Hopewell PC, Mills J, Snider GL, Stover DE. Pulmonary complications of the acquired immunodeficiency syndrome: an update. Report of the Second National Heart, Lung and Blood Institute Workshop. Am Rev Respir Dis 1987; 135:504-9. 3. Murray JF, Felton CP, Garay SM, et al. Pulmonary complications of the acquired immunodeficiency syndrome. N Engl J Med 1984; 310:1682-8. 4. Broaddus C, Dake MD, Stulbarg MS, et al. Bronchoalveolar lavage and transbronchial biopsy for the diagnosis of pulmonary infections in the acquired immunodeficiency syndrome. Ann Intern Med 1985; 102:747-52. 5. Golden JA, Hollander H, Stulbarg MS, Gamsu G. Bronchoalveolar lavage as the exclusive diagnostic modality for Pneumocystis carinii pneumonia: a prospective study among patients with acquired immunodeficiency syndrome. Chest 1986; 90:18-22. 6. Stover DE, White DA, Romano PA, Gellene RA. Diagnosis of pulmonary disease in acquired immune deficiency syndrome (AIDS): role of bronchoscopy and bronchoalveolar lavage. Am Rev Respir Dis 1984; 130:659-62. 7. Stover DE, Zaman MB, Hajdi SI, Lange M, Gold J, Armstrong D. Bronchoalveolar lavage in the diagnosis of diffuse pulmonary infiltrates in the immunosuppressed host. Ann Intern Med 1984; 101:1-7. 8. Shelhammer JH, Ognibene FP, Macher AM, et al. Persistence of Pneumocystis carinii in lung tissue of acquired immunodeficiency syndrome patients treated for pneumocystis pneumonia. Am Rev Respir Dis 1984; 130:1161-5. 9. Brenner M, Ognibene FP, Lack EE, et al. Prognostic factors and life expectancy of patients with acquired immunodeficiency syndrome and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1987; 136:1199-206. 10. Barrio JL, Harcup C, Baier JH, Pitchenik AB. Value of repeat fiberoptic bronchoscopies and significance of non diagnostic bronchoscopic results in patients with the acquired immunodeficiency syndrome. Am Rev Respir Dis 1987; 135:422-5. 11. Kim CK, Foy JM, Cushion MT, et al. Comparison of histologic and quantitative techniques in evaluation of therapy for experimental Pneumocystis carinii pneumonia. Antimicrob Agents Chemother 1987; 31:197-201. 12. Greco MJ, Heimann A, Steibigel RT, Smaldone GC. Severity of Pneumocystis carinii pneumonia (PCP) correlates with number of organisms recovered from lavage (abstract). Am Rev Respir Dis 1989; 139:AI49. 13. Limper AH, Offord KD, Smith TF, Martin WJ. Pneumocystis cariniipneumonia: differences in lung parasite number and inflammation in patients with and without AIDS. Am Rev Respir Dis 1989; 140:1204-9. 14. Baughman RP, Strohofer S, Colangelo G,
Frame PT. A semi-quantitative technique for estimating Pneumocystis cariniiburden in the lung. J Clin Microbiol 1990; 28:1425-7. 15. Mason GR, Hashimoto CH, Dickman PS, Foutty LF, Cobb CJ. Prognostic implications of bronchoalveolar lavage neutrophilia in patients with Pneumocystis carinii pneumonia and AIDS. Am Rev Respir Dis 1989; 139:1336-42. 16. Smith RL, El-Sadr WM, Lewis ML. Correlation of bronchoalveolar lavage cell populations with clinical severity of Pneumocystis carinii pneumonia. Chest 1988; 92:60-4. 17. Thorpe JE, Baughman RP, Frame PT, WesseIer TA, Staneck JL. Bronchoalveolar lavage for diagnosing acute bacterial pneumonia. J Infect Dis 1987; 155:855-61. 18. Baughman RP, Strohofer S, Kim CK. Variation of differential cell counts of bronchoalveolar lavage fluid. Arch Pathol Lab Med 1986; 110:341-3. 19. Tollerud OJ, Kim CK, Weseler TA, Baughman RP. Use of a rapid differential stain for identifying Pneumocystis carinii in bronchoalveolar lavage fluid. Chest 1988; 95:493-7. 20. Shivaram I, Wooten 0, Klapper P, Demetis S, Korlipara 0, Zaman M. Relationship of inflammatory cellsand organism load in lung lavage (BAL) to severity of Pneumocystis carinii pneumonia (PCP) (abstract). Am Rev Respir Dis 1990; 141:A270. 21. Allegra CJ, Chabner BA, Tuazon CA, et al. Trimetrexate for the treatment of Pneumocystis cariniipneumonia in patients with the acquired immunodeficiency syndrome. N Engl J Med 1987; 317:978-85. 22. Millar AB, Patou G, Miller RF, et al. Cytomegalovirus in the lungs of patients with AIDS. Respiratory pathogen or passenger? Am Rev Respir Dis 1990; 141:1474-7. 23. Miles PR, Baughman RP, Linnemamm CC. Cytomegalovirus in the bronchoalveolar lavage fluid of patients with AIDS. Chest 1990; 97:1072-6. 24. Haverkos H. PCP therapy project group. Assessment of therapy for Pneumocystiscarinii pneumonia. Am J Med 1984; 76:501-8. 25. Wallace JM, Barbers RG, Oishi JS, Prince H. Cellular and T-Iymphocyte subpopulation profiles in bronchoalveolar lavage fluid from patients with acquired immunodeficiency syndrome and pneumonitis. Am Rev Respir Dis 1984; 130:786-90. 26. White DA, Gellene RA, Gupta S, Cunningham-Rundles C, Stover DE. Pulmonary cell populations in the immunosuppressed patient: bronchoalveolar lavage findings during episodes of pneumonitis. Chest 1985; 88:352-9. 27. Savleda J, Gea J, Aran X, et al. Usefulness of neutrophilia in the bronchoalveolar lavage, effect of an aggressive therapeutic approach in patients with AIDS and Pneumocystis carinii pneumonia (abstract). Am Rev Respir Dis 1990; 141:A270. 28. Venet A, Clavel F, Israel-Biet D, et al. Lung in acquired immune deficiency syndrome: infectious and immunologic status assessed by bronchoalveolar lavage. Bull Eur Physiopathol Respir 1985; 21:535-43. 29. Jules-Elusee KM, Stover DE, Zaman MB, Bernard EM, White DA. Aerosolized pentamidine: effect on diagnosis and presentation of Pneumocystis carinii pneumonia. Ann Intern Med 1990; 112:750-7.
