Prospective Evaluation of a Prognostic Score for Pneumocystis carinii Pneumonia in HIV-Infected Patients· Rudolf Speich, M.D.; Milos Opravil, M.D.; Rainer Webet; M.D.; Thomas Hess, M. D.; Ruedi Luethy, M. D.; and Erich W Russi, M.D., F.C.C.R Serum lactate dehydrogenase levels, alveolar-arterial oxygen gradient, and percentage of neutrophils in bronchoalveolar lavage correlate most strongly with early mortality in Pneumocystis carinii pneumonia (PCP) in HIV-infected patients. However, the individual outcome can not be predicted by these parameters due to a considerable overlap between survivors and nonsurvivors. We prospectively investigated a PCP severity score, which has been developed earlier based on a retrospective analysis. Seven of 94 consecutively examined HIV-infected patients died within 14 days after diagnosis of PCP. A PCP severity score greater than 7 had a positive predictive value for early fatal outcome of 66.7 percent (6/9) and a negative predictive value of 98.8 percent (84/85). The overall diagnostic accuracy was 95.7 percent (90/94). The positive predictive value
for early fatal outcome of a P(A-a)O! >35 mm Hg was 24 percent (6/25); the negative predictive value was 98.6 percent (68/69). However, the overall diagnostic accuracy was only 78.7 percent (74/94). The PCP severity score is a valuable tool for clinical decision making, for the early identification of patients with a prognostic unfavorable course, and for the comparison of patient populations in future studies of HIV-associated PCP. (Chest 1992; 102:1045-48)
Despite advances in prevention, diagnosis, and treatment of Pneurrwcystis carinii pneumonia (PCP), this disease continues to be a major source of morbidity and mortality in patients infected with the human immunodeficiency virus (HIV). Among various prognostic factors for adverse outcome in patients with PCE serum lactate dehydrogenase (LDH), alveolararterial oxygen gradient P(A-a)02' and the percentage of neutrophils in bronchoalveolar lavage (BAL) correlate most strongly with early mortality due to respiratory failure .1-7 Unfortunately, these three parameters do not allow the prediction of individual outcome due to a considerable overlap between survivors and nonsurvivors. Based on a retrospective analysis, combining these three parameters, we have developed a prognostic score (Table 1) that enabled us to predict early fatal outcome quite reliably.6 We present a prospective evaluation ofthis PCP severity score (PSS).
and previous episodes of PCP were noted. Serum LDH (normal S 460 U/L, measured at 37°C) and arterial blood gas samples on room air (with calculation of the P[A-a]02 by the simplified alveolar gas equation) were obtained at the time of hospital admission. Chest roentgenogram changes were not scored. BAL was performed within 4B h of hospital admission by fiberoptic bronchoscopy using four 5O-ml aliquots of normal saline solution immediately aspirated manually with a syringe. Differential cell counts were determined on Wright-stained, cytocentrifuged preparations of BAL 8uid. The PCP severity score (Table 1) was calculated as previously described. 6 Choice of therapy was determined by the individual physicians caring for the patient. Co-trimoxazole was used as the first-line drug, either intravenously in hospitalized patients or orally in patients treated on an out-patient basis. In six patients, the initial treatment consisted of intravenous pentamidine, and in about 25 percent of the patients, co-trimoxazole was replaced by pentamidine because ofadverse reactions. Eight patients received corticosteroids and four were mechanically ventilated for a period of 5 to 10 days. Four patients with severe PCP were not admitted to ICU because of severe chronic wasting syndrome or the patient's refusal. Patients were grouped as follows: (A) survival after bronchoscopy for more than 14 days, and (B) early fatal outcome, defined as death within 14 days after diagnosis of PCE All data are expressed as mean ± so. The significance of differences between the groups was determined with the Mann-Whitney U test for continuous variables and the x2 test for discrete variables. Not significant (NS) indicated a p value ~0.05.
METHODS
Patients
Between January 1989 and December 1990, we prospectively evaluated 94 consecutive HIV-infected patients with PCP diagnosed by BAL. The studied population consisted ofBl men and 13 women; the mean age was 36.2 ± 10.5 years. There were 58 homosexual men, 35 IV drug users, and one patient with transfusion-related disease. The duration of the three most common symptoms consisting of fever, cough, and dyspnea, the current smoking status, *From the Department of Internal Medicine, University Hospital of Zurich, Switzerland. Manuscript received July 8; revision acceptedJanuary 21. Reprint requests: Dr. Speich, Department 0 Internal Medicine, University Hospital, CH-8091 Zurich, Switzerland
BAL = bronchoalveolar lavage; UIV = human immunodeficiency virus; LDH = lactate dehydrogenase; pcp = Pneumocystis carinii pneumonia; PSS Pneumocy.tis carinii pneumonia severity score
=
Table I-PCP Severity Score Points
P(A-a)02' mmHg
1 2 3 4 5
>20 >30 >40 >50 >60
LDH, UIL >460 (> normal) >920 (> 2x normal) > 1,380 (> 3x normal)
BAL Neutrophils, %
>2 >5 >10 >15 >20
CHEST I 102 I 4 I OCTOBER, 1992
1045
Table 2-CompariBon between Survivors (Group A) and Nonsurvivors (Group B)* Group A (n=87) Age, yr Sex Risk category Current smokers Recurrent PCP Duration of symptoms, wk Concomitant bacterial infection Therapy with steroids Mechanical ventilation Serum LDU, U/L P(A-a)02, mm Hg BAL neutrophils, % BAL eosinophils, % BAL lymphocytes, % PCP severity score
Group B (n=7)
tively (p = 0.004). These three parameters overlapped considerably, however, between the two patient groups (Fig 1). In contrast to our retrospective analysis, also, the percentage of BAL eosinophils differed significantly between group A and B (1.8±2.6 percent and 6.7±7.8 percent respectively; p=0.02). No concomitant bacterial or viral respiratory infections could be documented in the nonsurvivors. Autopsy was performed in six of seven nonsurvivors. It revealed severe PCP with moderate to severe interstitial fibrosis in all of them. One patient showed small peripheral pulmonary emboli and a small cavitation containing noninvasive Aspergillus fumigatus. No other additional pulmonary abnormalities or infections could be found. The calculated mean PCP severity score (PSS) was 3.0±2.4 and 9.9±3.1 in survivors and nonsurvivors, respectively (p = 0.00(1). Six ofthe seven nonsurvivors had a score >7 (Table 3), whereas only three of the 87 survivors did (p = 0.(01). A PSS >7 had a positive predictive value for early fatal outcome of 6/9 (66.7 percent), ie, mortality of patients with a PSS >7 was 66.7 percent. The negative predictive value was 84/85 (98.8 percent), ie, mortality of patients with a score $:7 was 1.2 percent. The overall diagnostic accuracy was 90/94 (95.7 percent). Mortality of patients with a P(A-a)02 >35 mm Hg was 6/25 (24 percent) compared with 1/69 (1.4 percent) in patients with a value $:35 mm Hg. Thus, the positive predictive value for early fatal outcome of a P(A-a)02 was 24 percent; the negative predictive value was 98.6 percent. The overall diagnostic accuracy of a P(A-a)02 >35 mm Hg predicting early fatal outcome, however, was only 74/94 (78.7 percent).
p
36.1 ± 10.7 13f,74m 52H, 34D, IT 38 14 3.1 ±2.3 5
38.1 ± 7.4 7m 6H,ID 3 0 3.4±2.4 0
NS NS NS NS NS NS NS
4 1 635±597 26.4± 12.6 8.9± 17.9 1.8±2.6 25.9±20.8 3.0±2.4
4 3 1267 ± 302 60.1 ± 16.3 37.1±33.2 6.7±7.8 17.0±20.5 9.9±3.1
0.0002 0.0001 0.004 0.02 NS 0.0001
*H = homosexual; D = IV drug use; T = transfusion-related; NS = not significant. RESULTS
Seven (7.4 percent) of the 94 patients died within 14 days after diagnosis of PCP (group B). Death was caused by respiratory failure in all of them. Age, sex, risk factors for acquired immunodeficiency syndrome (AIDS), current smoking status, duration of symptoms, the number of patients with recurrent PC~ and the incidence of concomitant bacterial infections did not significantly differ between group A and B (Table 2). The mean serum LDH was 635 ± 597 U/L and 1,267 ± 302 U/L in survivors and nonsurvivors, respectively (p=O.OOO2). The mean P(A-a)02 was 26.4± 12.6 mm Hg and 60.1 ± 16.3 mm Hg in survivors and nonsurvivors, respectively (p = 0.00(1). The percentage of BAL neutrophils was 8.9 ± 17.9 percent and 37.1 ± 33.2 percent in survivors and nonsurvivors, respec-
DISCUSSION
Pneumocystis carinii pneumonia (PCP) is still the
o 5500 2000
1500
til'
80
0
70
8 0 0
8 0
1000
-'
60
g 0
E E
40
ti
30
..
% 500
C 20
0:-
-' p=O.OOO2 A
B FIGURE
1048
50
0
Q
0
Q
:I:
0
3
0 0
10 0
80
0 0
60
0
'if. 50
0 0
I
I 8
§ A
0 0
70
0
9
0
!ff.
:c
~
0
0
;
~
40 30
e c -' 20 ct
m
p=O.OOO1
B
10 0
0 0 0 0
8
0
0 0
I
A
B
p:O.OO4
B
1. Overlap of the three parameters bem'een the m'o patient ~roups. Prognostic Score for PCP in HIV-infected Patients (Speich et 8/)
Table 3-Distribution a/PCP Severity Score (PSS) in Patients a/Group A and B
o Group A Group B
8
21
2
3
4
5
6
7
17
10
10 1
9
6
3
most common life-threatening opportunistic infection in patients with AIDS. Due to early diagnosis, aggressive treatment, and possibly the use of adjunctive corticosteroids, the mortality ofPCP has been reduced over the last years, but it remains considerable. In our experience, mortality was 50 percent until the end of 1985, 20 percent during the period of 1986 to 1988,6 and 7.4 percent in the current study group. This is in accordance with the findings of Brenner et al. l The approach toward intensive care for patients with AIDS and respiratory failure caused by PCP has changed over time. During the first phase ofthe AIDS epidemic (1981 to 1984), all potentially life-extending treatments were pursued because of lack of data. In response to the survival rates of 0 to 16 percent in these series, patients and clinicians increasingly declined leu care during the second phase from 1985 to 1987/' However, in the past three years, improved survival rates of 36 to 50 percent for AIDS patients with PCP and respiratory failure have been reported.9-11 The outcome could not be predicted by clinical factors. 9.10 Moreover, the APACHE II classification as a measure of severity of illness grossly underestimates mortality in PCP patients requiring intensive care (44.3 percent predicted vs 86.6 percent observed).12 Thus, a new score system for severity of PCP is urgently needed. As shown by the present study, the most important prognostic factors for PCP are the serum LDH level, the P(A-a)02' and the percentage of neutrophils in BAL.I-7 However, a considerable overlap between the values in survivors and nonsurvivors does not allow the prediction of the individual outcome. This may be due to several concomitant factors. In our study group, for instance, four survivors with serum LDH levels greater than two times normal suffered from extrapulmonary mycobacterioses, cerebral toxoplasmosis, or non-Hodgkin's lymphoma. In five ofour survivors with more than 10 percent neutrophils in the BAL fluid, concomitant bacterial infections could be documented (data not shown). To overcome the influence of these concomitant factors, we created the PCP severity score (PSS) combining these three parameters. 6 The present prospective study confirms the prognostic value of this score. A PSS >7 had a positive predictive value for early fatal outcome of66. 7 percent and a negative predictive value of 98.8 percent. The overall diagnostic accuracy was 95.7 percent. Therefore, a PSS >7 reliably predicts adverse outcome in an individual patient. A drawback of the present study may be the fact
8
9
10
11
12
13
4
that not all of the severe PCP cases received mechanical ventilation either because of advanced AIDS or the patient's refusal. These circumstances, however, represent daily clinical practice. Early aggressive therapy with corticosteroids and mechanical ventilation may well improve the prognosis in patients with a high PSS as illustrated by our patient with a PSS of 13 who survived with this approach. This study was conducted before the publication of trials suggesting that short-term mortality from moderate to severe PCP is reduced by adjunctive therapy with corticosteroids. 13-15 Therefore, only eight patients in our study population received corticosteroids. However, the fact that only four patients in the surviving group were treated with steroids reduces the bias of a PSS $.7 predicting a too favorable outcome. Since mortality was only 1.2 percent (1/85) in the group of patients with a PSS s:7, this may represent a subset of patients that does not benefit from corticosteroid therapy. The Consensus Conference of the National Institute of Health-University of California Expert Panel states that patients with a P(A-a)02 >35 mm Hg should receive corticosteroid therapy.16 Applying this cut-off to our study group, 25 patients should have received corticosteroids. However, using a PSS >7 as indication for corticosteroid therap~ only nine patients would have been treated. The only patient who died and had a PSS s:7 would not have been treated using the criteria of the Consensus Conference either because his P(A-a)02 was 32 mm Hg. The benefit of corticosteroid treatment in mild to moderate PCP remains to be proved. 13 Moreover, steroids may have deleterious side effects such as an increased incidence of herpes simplex infection,13 disseminated tuberculosis,17 and accelerated clinical progression of Kaposi's sarcoma. 18 Therefore, a PSS >7 predicting unfavorable outcome of PCP far better (mortality 66.7 percent) than a P(A-a)02 >35 mm Hg (mortality 24 percent) might serve as a cut-off for the indication of corticosteroid therapy. However, this has to be confirmed in a prospective study, because three of our surviving patients had a PSS s:7 but received steroids because of a P(A-a)02 >35 mm Hg. In conclusion, this prospective study confirms that the severity score for PCP is a valuable tool for clinical decision making and the selection of patient subpopulations with a prognostic unfavorable course that would most probably benefit from innovative therapeutic strategies. 19.20 Moreover, the PSS might serve as a method for comparing patient populations and CHEST I 102 I 4 I OCTOBER, 1992
1047
assuring similar severity of illness in future studies on PCP in "IV-infected patients. REFERENCES
2 3
4
5
6 7
8
9
10
Brenner M, Ognibene F, Lack E, Simmons T, Suffredini A, Lane H, et ale Prognostic factors and life expectancy of patients with acquired immunodeficiency syndrome and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1987; 136:1199-206 Garay S, Greene J. Prognostic indicators in the initial presentation of Pneumocystis carinii pneumonia. Chest 1989; 95:769-72 Kales C, Murren J, Torres R, Crocco J. Early predictors of inhospital mortality for Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. Arch Intern Med 1987; 147:1413-17 Mason G, Hashimoto C, Dickmann ~ Foutty L, Cobb C. Prognostic implications of bronchoalveolar lavage neutrophilia in patients with Pneumocystis carinii pneumonia and AIDS. Am Rev Respir Dis 1989; 139:1336-42 Smith R, EI-Sadr ~ Lewis M. Correlation of bronchoalveolar lavage cell population with clinical severity of Pneumocystis carinii pneumonia. Chest 1988; 93:60-4 Speich R, Weber R, Kronauer C, Opravil M, Russi E. Prognostic score for Pneumocystis carinii pneumonia. Respiration 1990; 57:259-63 Zaman M, White D. Serum lactate dehydrogenase levels and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1988; 137:796-800 Wachter R, Luce J, Turner J, Volberding ~ Hopewell E Intensive care of patients with the acquired immunodeficiency syndrome: outcome and changing patterns of utilization. Am Rev Respir Dis 1986; 134:891-96 Friedman Y, Franklin C, Rackow E, Weil M. Improved survival in patients with AIDS, Pneumocystis carinii pneumonia, and severe respiratory failure. Chest 1989; 96:862-66 Wachter R, Russi M, Bloch D, Hopewell ~ Luce J. Pneumocystis carinii pneumonia and respiratory failure in AI DS. Am Rev Respir Dis 1991; 143:251-56
1048
11 Montaner J, Ruedy J, Lawson L. Acute respiratory failure secondary to Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome: a potential role for systemic corticosteroids [abstract]. Chest 1987; 92:133S 12 Smith R, Levine S, Lewis M. Prognosis of patients with AIDS requiring intensiye care. Chest 1989; 96:857-61 13 Bozette S, Sattler F, Chiu J, Wu A, Gluckstein D, Kemper C, et al. A controlled trial of early adjunctive treatment with corticosteroids for Pneumocystis carinii pneumonia in the acqUired immunodeficiency syndrome. N Engl J Med 1990; 323:1451-57 14 Gagnon S, Boota A, Fischl M, Baier H, Kirskey 0, La Voie L. Corticosteroids as adjunctive therapy for severe Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. N Engl J Med 1990; 323:1444-50 15 Montaner JSG, Lawson LM, Levitt N, Belzberg A, Schechter MT, Ruedy J. Corticosteroids prevent early deterioration in patients with moderately severe Pneumocystis carinii pneumonia and the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 1990; 113:14-20 16 National T. Consensus statement on the use of corticosteroids as adjunctive therapy for Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. N Engl J Med 1990; 323:1500-04 17 Clumeck N, Hermans E Corticosteroids as adjunctive therapy for Pneumocystis carinii pneumonia in patients with AIDS. N Engl J Med 1991; 324:1666-67 18 Gill PS, Loureiro C, Bernstein M, Rarick MU, Sattler F, Levine AM. Clinical effect of glucocorticoids on Kaposi sarcoma related to the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 1989; 110:937-40 19 Miller R, Mitchell D. Management of respiratory failure in patients with the acquired immune deficiency syndrome and Pneumocystis carinii pneumonia. Thorax 1990; 45:140-46 20 Siegel R, Yoo O. Therapeutic lavage for severe Pneumocystis carinii pneumonia in patients with AI DS. Am Rev Respir Dis 1988; 137:A496
Prognostic Score for PCP in HIV-infected Patients (Speich et al)
for early fatal outcome of a P(A-a)O! >35 mm Hg was 24 percent (6/25); the negative predictive value was 98.6 percent (68/69). However, the overall diagnostic accuracy was only 78.7 percent (74/94). The PCP severity score is a valuable tool for clinical decision making, for the early identification of patients with a prognostic unfavorable course, and for the comparison of patient populations in future studies of HIV-associated PCP. (Chest 1992; 102:1045-48)
Despite advances in prevention, diagnosis, and treatment of Pneurrwcystis carinii pneumonia (PCP), this disease continues to be a major source of morbidity and mortality in patients infected with the human immunodeficiency virus (HIV). Among various prognostic factors for adverse outcome in patients with PCE serum lactate dehydrogenase (LDH), alveolararterial oxygen gradient P(A-a)02' and the percentage of neutrophils in bronchoalveolar lavage (BAL) correlate most strongly with early mortality due to respiratory failure .1-7 Unfortunately, these three parameters do not allow the prediction of individual outcome due to a considerable overlap between survivors and nonsurvivors. Based on a retrospective analysis, combining these three parameters, we have developed a prognostic score (Table 1) that enabled us to predict early fatal outcome quite reliably.6 We present a prospective evaluation ofthis PCP severity score (PSS).
and previous episodes of PCP were noted. Serum LDH (normal S 460 U/L, measured at 37°C) and arterial blood gas samples on room air (with calculation of the P[A-a]02 by the simplified alveolar gas equation) were obtained at the time of hospital admission. Chest roentgenogram changes were not scored. BAL was performed within 4B h of hospital admission by fiberoptic bronchoscopy using four 5O-ml aliquots of normal saline solution immediately aspirated manually with a syringe. Differential cell counts were determined on Wright-stained, cytocentrifuged preparations of BAL 8uid. The PCP severity score (Table 1) was calculated as previously described. 6 Choice of therapy was determined by the individual physicians caring for the patient. Co-trimoxazole was used as the first-line drug, either intravenously in hospitalized patients or orally in patients treated on an out-patient basis. In six patients, the initial treatment consisted of intravenous pentamidine, and in about 25 percent of the patients, co-trimoxazole was replaced by pentamidine because ofadverse reactions. Eight patients received corticosteroids and four were mechanically ventilated for a period of 5 to 10 days. Four patients with severe PCP were not admitted to ICU because of severe chronic wasting syndrome or the patient's refusal. Patients were grouped as follows: (A) survival after bronchoscopy for more than 14 days, and (B) early fatal outcome, defined as death within 14 days after diagnosis of PCE All data are expressed as mean ± so. The significance of differences between the groups was determined with the Mann-Whitney U test for continuous variables and the x2 test for discrete variables. Not significant (NS) indicated a p value ~0.05.
METHODS
Patients
Between January 1989 and December 1990, we prospectively evaluated 94 consecutive HIV-infected patients with PCP diagnosed by BAL. The studied population consisted ofBl men and 13 women; the mean age was 36.2 ± 10.5 years. There were 58 homosexual men, 35 IV drug users, and one patient with transfusion-related disease. The duration of the three most common symptoms consisting of fever, cough, and dyspnea, the current smoking status, *From the Department of Internal Medicine, University Hospital of Zurich, Switzerland. Manuscript received July 8; revision acceptedJanuary 21. Reprint requests: Dr. Speich, Department 0 Internal Medicine, University Hospital, CH-8091 Zurich, Switzerland
BAL = bronchoalveolar lavage; UIV = human immunodeficiency virus; LDH = lactate dehydrogenase; pcp = Pneumocystis carinii pneumonia; PSS Pneumocy.tis carinii pneumonia severity score
=
Table I-PCP Severity Score Points
P(A-a)02' mmHg
1 2 3 4 5
>20 >30 >40 >50 >60
LDH, UIL >460 (> normal) >920 (> 2x normal) > 1,380 (> 3x normal)
BAL Neutrophils, %
>2 >5 >10 >15 >20
CHEST I 102 I 4 I OCTOBER, 1992
1045
Table 2-CompariBon between Survivors (Group A) and Nonsurvivors (Group B)* Group A (n=87) Age, yr Sex Risk category Current smokers Recurrent PCP Duration of symptoms, wk Concomitant bacterial infection Therapy with steroids Mechanical ventilation Serum LDU, U/L P(A-a)02, mm Hg BAL neutrophils, % BAL eosinophils, % BAL lymphocytes, % PCP severity score
Group B (n=7)
tively (p = 0.004). These three parameters overlapped considerably, however, between the two patient groups (Fig 1). In contrast to our retrospective analysis, also, the percentage of BAL eosinophils differed significantly between group A and B (1.8±2.6 percent and 6.7±7.8 percent respectively; p=0.02). No concomitant bacterial or viral respiratory infections could be documented in the nonsurvivors. Autopsy was performed in six of seven nonsurvivors. It revealed severe PCP with moderate to severe interstitial fibrosis in all of them. One patient showed small peripheral pulmonary emboli and a small cavitation containing noninvasive Aspergillus fumigatus. No other additional pulmonary abnormalities or infections could be found. The calculated mean PCP severity score (PSS) was 3.0±2.4 and 9.9±3.1 in survivors and nonsurvivors, respectively (p = 0.00(1). Six ofthe seven nonsurvivors had a score >7 (Table 3), whereas only three of the 87 survivors did (p = 0.(01). A PSS >7 had a positive predictive value for early fatal outcome of 6/9 (66.7 percent), ie, mortality of patients with a PSS >7 was 66.7 percent. The negative predictive value was 84/85 (98.8 percent), ie, mortality of patients with a score $:7 was 1.2 percent. The overall diagnostic accuracy was 90/94 (95.7 percent). Mortality of patients with a P(A-a)02 >35 mm Hg was 6/25 (24 percent) compared with 1/69 (1.4 percent) in patients with a value $:35 mm Hg. Thus, the positive predictive value for early fatal outcome of a P(A-a)02 was 24 percent; the negative predictive value was 98.6 percent. The overall diagnostic accuracy of a P(A-a)02 >35 mm Hg predicting early fatal outcome, however, was only 74/94 (78.7 percent).
p
36.1 ± 10.7 13f,74m 52H, 34D, IT 38 14 3.1 ±2.3 5
38.1 ± 7.4 7m 6H,ID 3 0 3.4±2.4 0
NS NS NS NS NS NS NS
4 1 635±597 26.4± 12.6 8.9± 17.9 1.8±2.6 25.9±20.8 3.0±2.4
4 3 1267 ± 302 60.1 ± 16.3 37.1±33.2 6.7±7.8 17.0±20.5 9.9±3.1
0.0002 0.0001 0.004 0.02 NS 0.0001
*H = homosexual; D = IV drug use; T = transfusion-related; NS = not significant. RESULTS
Seven (7.4 percent) of the 94 patients died within 14 days after diagnosis of PCP (group B). Death was caused by respiratory failure in all of them. Age, sex, risk factors for acquired immunodeficiency syndrome (AIDS), current smoking status, duration of symptoms, the number of patients with recurrent PC~ and the incidence of concomitant bacterial infections did not significantly differ between group A and B (Table 2). The mean serum LDH was 635 ± 597 U/L and 1,267 ± 302 U/L in survivors and nonsurvivors, respectively (p=O.OOO2). The mean P(A-a)02 was 26.4± 12.6 mm Hg and 60.1 ± 16.3 mm Hg in survivors and nonsurvivors, respectively (p = 0.00(1). The percentage of BAL neutrophils was 8.9 ± 17.9 percent and 37.1 ± 33.2 percent in survivors and nonsurvivors, respec-
DISCUSSION
Pneumocystis carinii pneumonia (PCP) is still the
o 5500 2000
1500
til'
80
0
70
8 0 0
8 0
1000
-'
60
g 0
E E
40
ti
30
..
% 500
C 20
0:-
-' p=O.OOO2 A
B FIGURE
1048
50
0
Q
0
Q
:I:
0
3
0 0
10 0
80
0 0
60
0
'if. 50
0 0
I
I 8
§ A
0 0
70
0
9
0
!ff.
:c
~
0
0
;
~
40 30
e c -' 20 ct
m
p=O.OOO1
B
10 0
0 0 0 0
8
0
0 0
I
A
B
p:O.OO4
B
1. Overlap of the three parameters bem'een the m'o patient ~roups. Prognostic Score for PCP in HIV-infected Patients (Speich et 8/)
Table 3-Distribution a/PCP Severity Score (PSS) in Patients a/Group A and B
o Group A Group B
8
21
2
3
4
5
6
7
17
10
10 1
9
6
3
most common life-threatening opportunistic infection in patients with AIDS. Due to early diagnosis, aggressive treatment, and possibly the use of adjunctive corticosteroids, the mortality ofPCP has been reduced over the last years, but it remains considerable. In our experience, mortality was 50 percent until the end of 1985, 20 percent during the period of 1986 to 1988,6 and 7.4 percent in the current study group. This is in accordance with the findings of Brenner et al. l The approach toward intensive care for patients with AIDS and respiratory failure caused by PCP has changed over time. During the first phase ofthe AIDS epidemic (1981 to 1984), all potentially life-extending treatments were pursued because of lack of data. In response to the survival rates of 0 to 16 percent in these series, patients and clinicians increasingly declined leu care during the second phase from 1985 to 1987/' However, in the past three years, improved survival rates of 36 to 50 percent for AIDS patients with PCP and respiratory failure have been reported.9-11 The outcome could not be predicted by clinical factors. 9.10 Moreover, the APACHE II classification as a measure of severity of illness grossly underestimates mortality in PCP patients requiring intensive care (44.3 percent predicted vs 86.6 percent observed).12 Thus, a new score system for severity of PCP is urgently needed. As shown by the present study, the most important prognostic factors for PCP are the serum LDH level, the P(A-a)02' and the percentage of neutrophils in BAL.I-7 However, a considerable overlap between the values in survivors and nonsurvivors does not allow the prediction of the individual outcome. This may be due to several concomitant factors. In our study group, for instance, four survivors with serum LDH levels greater than two times normal suffered from extrapulmonary mycobacterioses, cerebral toxoplasmosis, or non-Hodgkin's lymphoma. In five ofour survivors with more than 10 percent neutrophils in the BAL fluid, concomitant bacterial infections could be documented (data not shown). To overcome the influence of these concomitant factors, we created the PCP severity score (PSS) combining these three parameters. 6 The present prospective study confirms the prognostic value of this score. A PSS >7 had a positive predictive value for early fatal outcome of66. 7 percent and a negative predictive value of 98.8 percent. The overall diagnostic accuracy was 95.7 percent. Therefore, a PSS >7 reliably predicts adverse outcome in an individual patient. A drawback of the present study may be the fact
8
9
10
11
12
13
4
that not all of the severe PCP cases received mechanical ventilation either because of advanced AIDS or the patient's refusal. These circumstances, however, represent daily clinical practice. Early aggressive therapy with corticosteroids and mechanical ventilation may well improve the prognosis in patients with a high PSS as illustrated by our patient with a PSS of 13 who survived with this approach. This study was conducted before the publication of trials suggesting that short-term mortality from moderate to severe PCP is reduced by adjunctive therapy with corticosteroids. 13-15 Therefore, only eight patients in our study population received corticosteroids. However, the fact that only four patients in the surviving group were treated with steroids reduces the bias of a PSS $.7 predicting a too favorable outcome. Since mortality was only 1.2 percent (1/85) in the group of patients with a PSS s:7, this may represent a subset of patients that does not benefit from corticosteroid therapy. The Consensus Conference of the National Institute of Health-University of California Expert Panel states that patients with a P(A-a)02 >35 mm Hg should receive corticosteroid therapy.16 Applying this cut-off to our study group, 25 patients should have received corticosteroids. However, using a PSS >7 as indication for corticosteroid therap~ only nine patients would have been treated. The only patient who died and had a PSS s:7 would not have been treated using the criteria of the Consensus Conference either because his P(A-a)02 was 32 mm Hg. The benefit of corticosteroid treatment in mild to moderate PCP remains to be proved. 13 Moreover, steroids may have deleterious side effects such as an increased incidence of herpes simplex infection,13 disseminated tuberculosis,17 and accelerated clinical progression of Kaposi's sarcoma. 18 Therefore, a PSS >7 predicting unfavorable outcome of PCP far better (mortality 66.7 percent) than a P(A-a)02 >35 mm Hg (mortality 24 percent) might serve as a cut-off for the indication of corticosteroid therapy. However, this has to be confirmed in a prospective study, because three of our surviving patients had a PSS s:7 but received steroids because of a P(A-a)02 >35 mm Hg. In conclusion, this prospective study confirms that the severity score for PCP is a valuable tool for clinical decision making and the selection of patient subpopulations with a prognostic unfavorable course that would most probably benefit from innovative therapeutic strategies. 19.20 Moreover, the PSS might serve as a method for comparing patient populations and CHEST I 102 I 4 I OCTOBER, 1992
1047
assuring similar severity of illness in future studies on PCP in "IV-infected patients. REFERENCES
2 3
4
5
6 7
8
9
10
Brenner M, Ognibene F, Lack E, Simmons T, Suffredini A, Lane H, et ale Prognostic factors and life expectancy of patients with acquired immunodeficiency syndrome and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1987; 136:1199-206 Garay S, Greene J. Prognostic indicators in the initial presentation of Pneumocystis carinii pneumonia. Chest 1989; 95:769-72 Kales C, Murren J, Torres R, Crocco J. Early predictors of inhospital mortality for Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. Arch Intern Med 1987; 147:1413-17 Mason G, Hashimoto C, Dickmann ~ Foutty L, Cobb C. Prognostic implications of bronchoalveolar lavage neutrophilia in patients with Pneumocystis carinii pneumonia and AIDS. Am Rev Respir Dis 1989; 139:1336-42 Smith R, EI-Sadr ~ Lewis M. Correlation of bronchoalveolar lavage cell population with clinical severity of Pneumocystis carinii pneumonia. Chest 1988; 93:60-4 Speich R, Weber R, Kronauer C, Opravil M, Russi E. Prognostic score for Pneumocystis carinii pneumonia. Respiration 1990; 57:259-63 Zaman M, White D. Serum lactate dehydrogenase levels and Pneumocystis carinii pneumonia. Am Rev Respir Dis 1988; 137:796-800 Wachter R, Luce J, Turner J, Volberding ~ Hopewell E Intensive care of patients with the acquired immunodeficiency syndrome: outcome and changing patterns of utilization. Am Rev Respir Dis 1986; 134:891-96 Friedman Y, Franklin C, Rackow E, Weil M. Improved survival in patients with AIDS, Pneumocystis carinii pneumonia, and severe respiratory failure. Chest 1989; 96:862-66 Wachter R, Russi M, Bloch D, Hopewell ~ Luce J. Pneumocystis carinii pneumonia and respiratory failure in AI DS. Am Rev Respir Dis 1991; 143:251-56
1048
11 Montaner J, Ruedy J, Lawson L. Acute respiratory failure secondary to Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome: a potential role for systemic corticosteroids [abstract]. Chest 1987; 92:133S 12 Smith R, Levine S, Lewis M. Prognosis of patients with AIDS requiring intensiye care. Chest 1989; 96:857-61 13 Bozette S, Sattler F, Chiu J, Wu A, Gluckstein D, Kemper C, et al. A controlled trial of early adjunctive treatment with corticosteroids for Pneumocystis carinii pneumonia in the acqUired immunodeficiency syndrome. N Engl J Med 1990; 323:1451-57 14 Gagnon S, Boota A, Fischl M, Baier H, Kirskey 0, La Voie L. Corticosteroids as adjunctive therapy for severe Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. N Engl J Med 1990; 323:1444-50 15 Montaner JSG, Lawson LM, Levitt N, Belzberg A, Schechter MT, Ruedy J. Corticosteroids prevent early deterioration in patients with moderately severe Pneumocystis carinii pneumonia and the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 1990; 113:14-20 16 National T. Consensus statement on the use of corticosteroids as adjunctive therapy for Pneumocystis carinii pneumonia in the acquired immunodeficiency syndrome. N Engl J Med 1990; 323:1500-04 17 Clumeck N, Hermans E Corticosteroids as adjunctive therapy for Pneumocystis carinii pneumonia in patients with AIDS. N Engl J Med 1991; 324:1666-67 18 Gill PS, Loureiro C, Bernstein M, Rarick MU, Sattler F, Levine AM. Clinical effect of glucocorticoids on Kaposi sarcoma related to the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 1989; 110:937-40 19 Miller R, Mitchell D. Management of respiratory failure in patients with the acquired immune deficiency syndrome and Pneumocystis carinii pneumonia. Thorax 1990; 45:140-46 20 Siegel R, Yoo O. Therapeutic lavage for severe Pneumocystis carinii pneumonia in patients with AI DS. Am Rev Respir Dis 1988; 137:A496
Prognostic Score for PCP in HIV-infected Patients (Speich et al)
