International Journal of Infectious Diseases (2009) 13, 5966

http://intl.elsevierhealth.com/journals/ijid

Acute respiratory failure due to Pneumocystis

pneumonia: outcome and prognostic factors
Viboon Boonsarngsuk *, Supinda Sirilak, Sumalee Kiatboonsri

Division of Pulmonary and Critical Care Medicine, Department of Medicine, Faculty of Medicine,
Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand

Received 3 August 2007; received in revised form 17 December 2007; accepted 26 March 2008
Corresponding Editor: N. Kumarasamy, Chennai, India

KEYWORDS Summary
Acute respiratory failure; Objectives: To examine the outcome and prognostic factors of in-hospital mortality in patients
Pneumocystis pneumonia; with acute respiratory failure (ARF) caused by Pneumocystis pneumonia (PCP) admitted to a
Prognostic factors; medical intensive care unit.
Immunosuppression; Methods: A retrospective review was conducted of all patients with ARF from PCP in Ramathibodi
Positive end-expiratory Hospital between 2000 and 2006. Patient characteristics, clinical presentation, and laboratory,
pressure; radiological and microbiological findings, as well as therapy and clinical course were included in
Mortality the analysis of prognostic factors of death.
Results: A total of 14 HIV-infected and 30 otherwise immunosuppressed patients were identified.
The overall mortality rate was 63.6%. Logistic regression analysis demonstrated that APACHE II
score on day 1 and level of PEEP used on day 3 of respiratory failure were associated with higher
hospital mortality. In a comparison between the HIV group and the non-HIV group, the early
mortality rate was significantly higher in the HIV group, but late hospital mortality was not
different between the two groups. Using a univariate logistic regression model, four parameters
were found to be significantly associated with death in the HIV group: sex, APACHE II score on day
1, CMV co-infection, and level of PEEP on day 3 of ARF. In the non-HIV group, corticosteroid use
prior to diagnosis of PCP and level of PEEP on day 3 of ARF were found to be the significant
parameters.
Conclusion: The mortality rate in patients with ARF caused by PCP was high. Various variable
factors were related to a poor prognosis. For improved survival, multimodality treatments are
needed to reduce these risk factors.
# 2008 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Introduction

Pneumocystis pneumonia (PCP), the disease caused by Pneu-

mocystis jirovecii, emerged as an important cause of mor-
* Corresponding author. Tel.: +66 2 201 1619. bidity and mortality in HIV-infected patients from early in the
E-mail address: bss-vb@hotmail.com (V. Boonsarngsuk). AIDS epidemic.1,2 Currently, with the use of highly active

1201-9712/$36.00 # 2008 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijid.2008.03.027

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
60 V. Boonsarngsuk et al.

antiretroviral therapy (HAART), the prescription of prophy- predicted body weight and optimal positive end-expiratory
lactic agents to persons at high clinical risk,3 and the avail- pressure (PEEP) by lung mechanics as described by Suter
ability of more sensitive immunofluorescence methods of P. et al.13 With this technique, the PEEP was increased sequen-
jirovecii detection,4 the overall incidence of PCP cases and tially with a consistent tidal volume, and the static compli-
survival following PCP in AIDS patients has generally ance was measured at each interval. Optimal PEEP, defined as
improved.5 In spite of this, the high mortality of patients the level of PEEP corresponding to maximal compliance, was
requiring mechanical ventilation (MV) has remained chosen. The level of optimal PEEP was titrated once daily and
unchanged, ranging from 50% to 60%.6 recorded for three consecutive days. A plateau pressure of
In contrast to AIDS-related cases, cases of PCP in patients 3035 cmH2O was allowed.
with other predisposing immunodeficiency states, such as
organ transplant recipients, patients with hematologic and Statistical analysis
solid tumors receiving chemotherapeutic agents, and persons
with chronic inflammatory diseases requiring prolonged use All values were expressed as the mean  standard deviation
of corticosteroids, may be increasing,7 and the associated (SD) for continuous variables and percent for categorical
mortality may be >50%.7,8 However, the previous literature variables. To determine the association of independent vari-
on both HIV- and non-HIV-related PCP has indicated that the ables with hospital mortality, continuous variables were
mortality of patients with acute respiratory failure (ARF) compared using the Students two-tailed t-test or nonpara-
requiring MV does not differ widely and ranges between metric MannWhitney U-test, in case of distribution not
40% and 60%, despite the use of new aggressive supportive being normal. The Chi-square test or the Fishers exact test,
interventions and monitoring.6,912 in case of low expected frequencies, was used for compar-
In order to examine the outcome and prognostic factors of isons of categorical variables. Variables identified as signifi-
in-hospital mortality in patients with ARF requiring MV cant in the univariate analysis were assessed as predictors of
caused by PCP at our institution, we retrospectively collected mortality using logistic regression analysis. Then, subgroup
data for PCP patients requiring MV and admitted to a medical analysis in both the HIV group and non-HIV-related PCP group
intensive care unit (ICU) between 2000 and 2006. Clinical, was performed to find the prognostic factors associated with
laboratory, and radiologic features, as well as mechanical hospital mortality in the same manner. In-hospital survival
ventilation parameters were examined as prognostic factors was assessed by KaplanMeier methods, and differences
of patient outcome. between the HIV group and the non-HIV-related PCP group
were assessed by the log-rank test. All statistical tests were
Materials and methods two-sided, and p < 0.05 was considered statistically signifi-
cant. All data were analyzed with a statistical software
Subjects package (SPSS, version 11.5 for Windows; SPSS Inc., Chicago,
IL, USA).
We performed a retrospective analysis on all consecutive
patients 15 years of age with a microbiologically confirmed Results
diagnosis of PCP, who were admitted to the medical ICU of
Ramathibodi Hospital, a tertiary university referral hospital Demographic features
in Bangkok, Thailand, and required treatment for ARF with
MV between January 1, 2000 and November 30, 2006. All A total of 44 confirmed cases of PCP in adult patients who
cases of PCP included in the study had cytologic documenta- developed ARF and required treatment with MV were identi-
tion of the organisms by immunofluorescence or Giemsa fied during the period between January 1, 2000 and November
staining in specimens of bronchoalveolar lavage (BAL) fluid 30, 2006. PCP was diagnosed by BAL in 41 out of 44 patients,
or transbronchial biopsy (TBBX) specimens. Cases of pre- while 13 cases were diagnosed by TBBX. Fourteen cases were
sumptive diagnosis of PCP were not included. The study HIV-seropositive and 30 cases had other conditions associated
protocol was approved by the Ethics Committee on Human with immunosuppression. Of these 44 patients, 27 were
Experimentation of Ramathibodi Hospital, Faculty of Medi- female. The mean age of the patients was 46.3 years. The
cine, Mahidol University. mean APACHE II score on day 1 was 22.3 (Table 1). The mean ICU
and hospital LOS were 18 days and 25 days, respectively.
Data collection
Immunosuppressive conditions
Clinical data abstracted included the following: general
demographic information, HIV status, underlying immuno- In the HIV group, seven (50%) were female and the mean age
suppressive condition, including medications and PCP pro- was 35.2 years. Of the 14 patients, 10 had been tested for
phylaxis, laboratory analysis, radiology, microbiology, CD4 count and the mean result was 53.1 cells/ml. Only one
APACHE (acute physiology and chronic health evaluation) II case had received trimethoprim/sulfamethoxazole (TMP/
score on the day of ARF, mechanical ventilation parameters, SMZ) prophylaxis.
antibiotic and steroid therapy, and hospital and ICU length of In the non-HIV group, 20 (66.7%) were female and the
stay (LOS), as well as overall hospital mortality. mean age was 51.5 years. The underlying diseases of the
In our ICU, to ventilate any patient who developed acute conditions associated with immunosuppression are presented
respiratory failure with diffuse bilateral lung diseases, we in Table 1. Eight patients had received systemic corticoster-
routinely used a tidal volume of 810 ml per kilogram of oids only. Two patients were treated with cytotoxic che-

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
Acute respiratory failure due to Pneumocystis pneumonia 61

Table 1 Clinical characteristics of 44 patients with Pneumocystis pneumonia-related acute respiratory failure

Variables Total (N = 44) Patients who Patients who p-Value

survived (N = 16) died (N = 28)
Age, years 46.3 (14.6) 47.4 (13.4) 45.7 (15.4) 0.717
Sex, female, n (%) 27 (61.4) 7 (25.9) 20 (74.1) 0.017
Underlying immune defect, n (%)
HIV 14 (31.8) 6 (42.9) 8 (57.1) 0.541
Non-HIV 30 (68.2)
Hematologic malignancy 10 (22.7) 4 (40.0) 6 (60.0)
Connective tissue disease 16 (36.4) 4 (25.0) 12 (75.0)
Miscellaneous 4 (9.1) 2 (50.0) 2 (50.0) 0.549
a
Immunosuppressive drug, n (%) (N = 30)
Corticosteroids alone 8 (26.7) 3 (37.5) 5 (62.5)
Chemotherapy alone 2 (6.7) 1 (50.0) 1 (50.0)
Combined steroids + chemotherapy 20 (66.7) 6 (30.0) 14 (70.0) 0.814
CD4, cells/ml (n = 10) b 53.1 (47.2) 54.6 (56.3) 50.7 (37.3) 0.907
PED before ARF, mg (n = 24) c 41.5 (16.5) 34.1 (20.1) 44.0 (15.0) 0.014
Duration PED before ARF, months (n = 24) c 4.1 (3.0) 5.6 (4.9) 3.6 (2.0) 0.361
Duration before treatment, days 6.9 (5.3) 5.8 (4.1) 7.5 (5.9) 0.329
Duration before ARF, days 9.7 (6.4) 9.3 (5.9) 10.0 (6.7) 0.737
Duration from treatment to ARF, days (n = 29) d 5.0 (4.1) 3.8 (3.6) 5.8 (4.4) 0.197
LDH, U/l 632.7 (239.2) 606.9 (247.6) 646.3 (240.4) 0.681
APACHE II score 22.3 (4.9) 19.8 (4.8) 23.6 (4.4) 0.016

Anti-PCP drug, n (%)

TMP/SMZ 36 (81.8) 13 (36.1) 23 (63.9)
Clindamycin + primaquine 4 (9.1) 1 (25.0) 3 (75.0)
TMP/SMZ ! clindamycin + primaquine 4 (9.1) 2 (50.0) 2 (50.0) 0.761
PED treatment, mg 66.8 (31.5) 60.6 (31.7) 70.3 (31.4) 0.331
Lung mechanics
Compliance, ml/cmH2O 24.7 (6.6) 27.3 (7.5) 23.5 (6.3) 0.372
PaO2:FiO2, mmHg 173.4 (91.5) 187.0 (72.5) 166.9 (100.0) 0.523
PEEP day 1, cmH2O 5.9 (1.8) 4.9 (1.4) 6.5 (1.8) 0.004
PEEP day 2, cmH2O 6.1 (1.8) 5.3 (0.8) 6.6 (2.1) 0.008
PEEP day 3, cmH2O 6.3 (1.8) 5.3 (1.0) 6.9 (1.9) 0.001
Respiratory co-pathogen, n (%)
Bacterial 21 (47.7) 8 (38.1) 13 (61.9) 0.820
Tuberculosis 7 (15.9) 2 (28.6) 5 (71.4) 0.640
Cytomegalovirus 15 (34.1) 4 (26.7) 11 (73.3) 0.336
Fungus 4 (9.1) 0 (0.0) 4 (100.0) 0.113
Strongyloides 3 (6.8) 0 (0.0) 3 (100.0) 0.175
Other 8 (18.2) 4 (50.0) 4 (50.0) 0.375
Pneumothorax, n (%) 16 (36.4) 4 (25.0) 12 (75.0) 0.236
Data are presented as mean (SD) or n (%).
PCP, Pneumocystis pneumonia; PED, prednisolone-equivalent dose; ARF, acute respiratory failure; LDH, lactate dehydrogenase; TMP/SMZ,
trimethoprim/sulfamethoxazole; PEEP, positive end-expiratory pressure.
a
All 30 patients who received immunosuppressive drugs were in the non-HIV group.
b
Ten out of 14 in the HIV group had CD4 cell count results.
c
All 24 patients who had received prednisolone prior to the diagnosis of PCP were in the non-HIV group.
d
The 29 patients who had received anti-PCP medication prior to acute respiratory failure.

motherapy not containing corticosteroids, and the remainder Clinical symptoms

had received a combination of both corticosteroids and
chemotherapy. The mean prednisolone-equivalent dose The mean duration of symptoms prior to ARF was about 9.7
(PED) and duration of corticosteroid use prior to diagnosis days and was found to be significantly longer in the HIV group
of PCP was 41.5 mg and 4.1 months, respectively. No one was (13.1  5.7 vs. 8.1  6.1 days for the HIV group and the non-
given prophylactic medications for PCP prior to onset of HIV group, respectively; p = 0.015). Twenty-nine patients had
pneumonia. received anti-PCP medication prior to ARF with a mean

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
62 V. Boonsarngsuk et al.

duration of 5 days. Despite no statistical significance, the

duration from anti-PCP treatment to ARF in those who sur-
vived was less than in those who died (3.8  3.6 vs. 5.8  4.4
days; p = 0.197). All patients in the HIV group had received
anti-PCP medication prior to ARF, while only 15 of the 30
patients in the non-HIV group had received it.

Laboratory and radiographic features

The mean lactate dehydrogenase (LDH) was 632.7 U/l. No

one had an LDH value within the normal range. No significant
difference was detected in the value of LDH between the HIV
and non-HIV groups.
As expected, chest radiographic examinations generally
revealed diffuse interstitial and ground-glass appearance on
the day of ARF. However, perihilar and basal interstitial
infiltration were present in 9.1% of all cases when they
developed ARF, and all of them were in the non-HIV group.
Subsequent chest radiographs in all patients identified 16 Figure 1 KaplanMeier estimate of overall survival for
patients with pneumothorax over the course of their ICU stay. patients with acute respiratory failure caused by Pneumocystis
pneumonia.
Co-pathogens
Outcomes
Co-pathogens identified from BAL or TBBX are presented in
Table 1. Cytomegalovirus (CMV) infection was diagnosed The overall in-hospital mortality rate for the 44 identified
in 15 patients, nine by demonstration of cytomegalic cases was 63.6% (Figure 1). In the first 7 days after ARF, the
intranuclear inclusions in TBBX specimens and six by posi- mortality rate was significantly higher in the HIV group
tive PCR results for CMV-DNA (the AMPLICOR CMV test; (mortality 14.3% vs. 0%; log rank p = 0.025) (Figure 2). How-
Roche Diagnostics, Branchburg, NJ, USA) in BAL fluid ever, the in-hospital mortality rate was higher in the non-HIV
obtained at the time of the diagnosis of PCP. Tuberculosis group, although without statistical significance (66.7% vs.
was found as co-pathogen in seven patients. Furthermore, 57.1%; log rank p = 0.606).
Strongyloides infection was found in three patients and all
of them died. Logistic regression analysis
Fungus was present in BAL fluid in four patients: Candida
albicans in two patients (one in the HIV group and the other in Statistical analyses were performed to identify clinical
the non-HIV group), Aspergillus spp in one in the non-HIV features associated with mortality. Using univariate ana-
group, and Cryptococcus neoformans in one in the HIV group. lyses, clinical features associated with mortality are pre-
Positive respiratory specimen results of Candida were sented in Table 1. Three parameters were proved to be
deemed to be secondary to colonization rather than active significantly associated with death from PCP: female gen-
disease. der ( p = 0.017), APACHE II score on day 1 ( p = 0.016), and
level of PEEP on all of the first three days of ARF ( p = 0.004,
Treatment and lung mechanics

Forty of the 44 patients were treated with TMP/SMZ.

Because the physician suspected TMP/SMZ-resistant PCP
infection, four of these 40 patients had their initial anti-
microbial therapy changed to clindamycinprimaquine. As
there was no pentamidine available at our hospital during
this period, four patients who had a history of suspected
allergy to TMP/SMZ received clindamycinprimaquine
instead. All patients were additionally treated with adjunc-
tive corticosteroids. The mean PED of corticosteroids used in
the treatment for ARF related to PCP in these patients was
66.8 mg.
The mean duration of MV requirement was 14.9 days. The
levels of PEEP applied on the first three days of ARF are shown
in Table 1. Although it did not reach statistical significance,
the level of PEEP applied on day 3 was higher in non-HIV
patients (6.6  2.0 vs. 5.7  1.0 cmH2O; p = 0.087). The
mean lung compliance was 24.7 ml/cmH2O. The mean ratio Figure 2 KaplanMeier plot of survival for HIV patients and
of PaO2 to fraction of inspired oxygen was 173.4, with 12 non-HIV patients with acute respiratory failure caused by Pneu-
(27.3%) patients having a ratio >200. mocystis pneumonia.

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
Acute respiratory failure due to Pneumocystis pneumonia 63

Table 2 Odds ratios for variables independently associated Discussion

with an increased risk of death; multivariate analysis
PCP is an opportunistic infection that is associated with
Variables OR (95% CI) p-Value substantial morbidity and mortality even when it is properly
APACHE II scores 1.247 (1.0121.537) 0.039 treated.8,9 Recently, the frequency of ARF due to PCP has
PEEP day 3, cmH2O 2.741 (1.1366.613) 0.025 been reported to be decreasing. This may be as a result of
early diagnosis and treatment, including the early use of
OR, odds ratio; CI, confidence interval; PEEP, positive end-expira-
adjunctive corticosteroids prior to ICU admission.14 Unfortu-
tory pressure.
nately, the casefatality rate for mechanically ventilated
patients has increased, suggesting that the development of
0.008, and 0.001, respectively). In the multivariate respiratory failure despite maximal therapy, including corti-
approach, only APACHE II score on day 1 and level of PEEP costeroids, carries an extremely poor prognosis. In our series,
on day 3 remained independently associated with lethal the mortality rate in patients with ARF requiring MV caused
outcome (Table 2). by PCP was 63.6%, which is comparable to previous stu-
Although age was significantly higher in the non-HIV group dies.6,15,16 Although early mortality was higher in the HIV
( p < 0.001), it was not an influence on mortality ( p = 0.131). group, hospital mortality between the HIV and non-HIV
In the HIV group, using a univariate logistic regression model, groups was not statistically different. This finding is consis-
sex, APACHE II score on day 1, level of PEEP on day 3, and CMV tent with the previous reports of Mansharamani et al.6 and
co-infection were independent predictors of survival, Ewig et al.17
whereas history of prior use of corticosteroid, duration of PCP was diagnosed by BAL alone in 93% of our patients.
symptoms before treatment, level of PEEP on day 3, and the Broaddus et al.18 found BAL alone had high diagnostic yields.
subsequent development of pneumothorax were found to be Because of this, some authors recommend it be done without
independent predictors of survival in the non-HIV group biopsy in HIV patients with a high suspicion for PCP.19 The
(Table 3). Although it did not attain statistical significance, yield of BAL in patients on prophylaxis with aerosolized
the non-survivors had higher APACHE II scores on day 1 than pentamidine has been reported to be decreased;20 in these
survivors in the non-HIV group (24.0  5.1 vs. 21.4  5.2; cases, BAL or TBBX, or both, performed in areas of radio-
p = 0.228). In a multivariate logistic regression model, in the graphic findings should be considered. In our series, only one
non-HIV group, the results indicated only two clinical para- case had had anti-PCP prophylaxis with TMP/SMZ and BAL was
meters that were independently associated with hospital able to diagnose PCP. None received aerosolized pentami-
mortality. They were history of prior use of corticosteroid dine.
and level of PEEP on day 3 (Table 4). Because of small In our study, we found that elevated APACHE II score on
numbers in the HIV group, multivariate analysis was not day 1 and level of PEEP on day 3 were independently asso-
performed. ciated with lethal outcome in patients with ARF caused by

Table 3 Univariate analysis for independent factors associated with hospital mortality; subgroup analysis for HIV and non-HIV
patients

Variables Patients who survived Patients who died p-Value

HIV group (n = 14) 6 (42.9) 8 (57.1)
Sex, female, n (%) 1 (14.3) 6 (85.7) 0.031
APACHE II score 17.0 (2.4) 22.8 (2.2) 0.001
Level of PEEP day 3, cmH2O 5.1 (0.4) 6.2 (1.1) 0.037
CMV co-pathogen, n (%) 0 (0.0) 4 (100) 0.040
Non-HIV group (n = 30) 10 (33.3) 20 (66.6)
Prior used corticosteroid, n (%) 6 (25.0) 18 (75.0) 0.020
Duration before treatment, days 4.2 (3.0) 8.4 (6.6) 0.026
Level of PEEP day 3, cmH2O 5.4 (1.2) 7.2 (2.0) 0.019
Pneumothorax, n (%) 1 (10) 9 (90) 0.037
Data are presented as mean (SD) or n (%).
PEEP, positive end-expiratory pressure; CMV, cytomegalovirus.

Table 4 Multivariate analysis of independent factors associated with hospital mortality in non-HIV patients

Variables OR (95% CI) p-Value

Prior use of corticosteroid 15.487 (1.018235.488) 0.048
Level of PEEP day 3, cmH2O 2.667 (0.9997.119) 0.050
Pneumothorax 14.219 (0.769263.034) 0.075
OR, odds ratio; CI, confidence interval; PEEP, positive end-expiratory pressure.

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
64 V. Boonsarngsuk et al.

PCP. In agreement with our series, Forrest et al.11,16 and development of P. jirovecii including CD4+ lymphocyte deple-
Benson et al.21 also found that APACHE II could predict the tion and immune dysfunction.9
mortality in patients with HIV-related PCP and respiratory The subsequent development of pneumothorax was asso-
failure. In contrast, in our patients, APACHE II score on day 1 ciated with mortality in univariate analysis; however, it was
was not found to be a significant variable associated with not found to be a significant factor when using multivariate
hospital outcome in non-HIV PCP. To the best of our knowl- analysis. Development of pneumothorax complicating PCP is
edge, there has been no report demonstrating APACHE II thought to represent a poor prognosis.15,24,34 In a study by
score to predict the mortality in non-HIV PCP. In a study by Festic et al.,15 all of their non-HIV-related PCP and ARF
Torres et al.,22 by using univariate analysis, high APACHE II patients who developed pneumothorax died, compared to
score was found to be one of the predictors of death. How- a 90% mortality rate in our patients.
ever, using multivariate analysis, it could not be identified as Because of the high mortality in patients with ARF requir-
an independent predictor of death. The updated APACHE III ing MV caused by PCP and the fact that it is not possible to
was designed to estimate the probability of in-hospital mor- distinguish which of these patients will or will not survive to
tality for adult ICU patients,23 and the findings of Festic hospital discharge based on information routinely available
et al.15 showed that APACHE III scores were predictive of before ICU admission, we agree with the use of anti-PCP
mortality among non-HIV patients with PCP. prophylaxis in both the HIV- and non-HIV patients who are at
The level of PEEP applied as a predictor of hospital out- high risk of developing PCP. Although the clinical significance
come has been reported by others.2426 Furthermore, Peruzzi of prophylaxis for PCP remains controversial in non-HIV
et al.25 found that the level of PEEP required begins to patients, some authors have suggested that immunosuppres-
decrease after approximately 72 hours of ICU care in the sion induced by chemotherapy or radiotherapy, or patients
survivor group, whereas the non-survivor group demon- with inflammatory diseases receiving glucocorticosteroids
strated the need for continued escalation of support. In 20 mg/day or more for 4 weeks or more should receive
the earliest stage of disease course, the histology in the lung prophylaxis.32,35,36 Furthermore, strongyloidiasis and CMV
shows only intra-alveolar exudates and minimal inflamma- disease should be considered as co-infections in these
tion. Progression to the chronic, organizing phase of diffuse patients, and aggressive work-up may be required in cases
alveolar damage is common, and evidence of interstitial and who are not improving despite maximal therapy. Protective
intraluminal fibrosis are present. In the late stage, extensive lung ventilation strategies in mechanical ventilated patients
septal thickening with fibrosis is found with much less alveo- should be used to prevent pneumothorax and other ventila-
lar exudates.19,27 This results in recruitable and non-recrui- tor-associated lung injury.37
table units under PEEP application. With the technique to There are several limitations to this study. The number of
determine the optimal PEEP as described by Suter,13 the patients studied is relatively small. Presently, with the use of
higher the level of PEEP achieved, the more the alveolar HAART, the prescription of prophylactic agents to persons at
process is represented and, on the other hand, the lower the high clinical risk, and empirical treatment for patients with
level of PEEP achieved, the more the restrictive process is clinically suspected PCP, the overall incidence of PCP cases
represented. So, in our patients, the higher level of PEEP with ARF is reduced, especially in AIDS patients. Further-
applied on day 3 may represent the ongoing disease process more, only microbiologically confirmed cases were eligible.
despite standard treatment, which resulted in fatal out- Although, in our study, we evaluated the role of many prog-
come, whereas the lower level of PEEP applied on day 3 nostic variables, only three prognostic factors were proved to
may represent change to a chronic process. be significantly associated with death in univariate analysis
In HIV-related PCP with ARF, the identifiable different and included in the logistic model. Nevertheless, the main
variables between the survivors and non-survivors other prognostic factors identified in this study are statistically
than APACHE II score on day 1 and level of PEEP applied on significant, correlate with findings of previous investiga-
day 3 were sex and CMV co-infection. Various studies have tions,11,16,21,2426 and are clinically plausible. Pooling of data
confirmed that patients with concomitant CMV infections from several centers could add statistical power to an ana-
have a higher mortality rate as compared with patients lysis of prognostic markers of poor outcome.
with PCP alone;2830 however, these studies were con- Another limitation is related to the retrospective nature
ducted after the introduction of adjunctive corticosteroid of this review. It remains possible that some important
treatment in patients with severe PCP. Jensen et al.30 variables may not have been recorded. However, we believe
showed that CMV-positive patients treated with adjunctive this to be unlikely, since the data set analyzed for each
corticosteroids have a worse vital prognosis than CMV- significant prognostic factor was nearly complete (100% data
positive patients without corticosteroid treatment. Corti- availability for gender, level of PEEP, history of prior use of
costeroid therapy may result in a more rapid development corticosteroid, duration of symptoms before treatment, and
of CMV disease in HIV-infected patients and thus the the subsequent development of pneumothorax and 91% data
clinical importance of concomitant CMV infection in PCP availability for APACHE II scores).
has changed.31 Finally, our ventilator management was perhaps not the
For those who developed ARF in the non-HIV-related PCP best and was open to debate. Even though optimal PEEP was
group, the level of PEEP on day 3 as well as history of prior use applied, the low tidal volume strategy was not used. The
of corticosteroid were associated with hospital mortality. Acute Respiratory Distress Syndrome Network demonstrated
Previous studies have demonstrated prior use of corticoster- that mechanical ventilation with a lower tidal volume (6 ml
oids as a risk of death in these patients,32,33 and a resultant per kilogram of predicted body weight) than is traditionally
high mortality rate.6,14,17 Several mechanisms have been used (12 ml per kilogram of predicted body weight) results in
postulated to explain the role of steroids in promoting the decreased mortality.38 However, when plateau pressure was

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
Acute respiratory failure due to Pneumocystis pneumonia 65

not high, a tidal volume of 7 ml/kg was found not to be with respiratory failure caused by AIDS-related Pneumocystis
associated with lower mortality compared with a tidal carinii pneumonia. Arch Intern Med 1999;159:7417.
volume of 10 ml/kg.39,40 Nevertheless, the widespread belief 12. Franklin C, Friedman Y, Wong T, Hu TC. Improving long-term
that tidal volume reduction is without benefit when plateau prognosis for survivors of mechanical ventilation in patients with
AIDS with PCP and acute respiratory failure: five-year follow-up of
pressure is already lower than 3035 cmH2O has not been
intensive care unit discharges. Arch Intern Med 1995;155:915.
substantiated. Further research is necessary to demonstrate 13. Suter PM, Fairley B, Isenberg MD. Optimum end-expiratory air-
the difference in benefit for these patients. way pressure in patients with acute pulmonary failure. N Engl J
In conclusion, we found various risk factors associated Med 1975;292:2849.
with hospital mortality in patients with ARF requiring MV 14. Hawley PH, Ronco JJ, Guillemi SA, Quieffin J, Russell JA, Lawson
caused by PCP include APACHE II score on day 1, level of PEEP LM, et al. Decreasing frequency but worsening mortality of acute
on day 3, sex, and CMV co-infection, as well as history of prior respiratory failure secondary to AIDS-related Pneumocystis car-
use of corticosteroid. Multimodality treatments, such as anti- inii pneumonia. Chest 1994;106:14569.
PCP prophylaxis in those patients at high risk of developing 15. Festic E, Gajic O, Limper AH, Aksamit TR. Acute respiratory
failure due to Pneumocystis pneumonia in patients without
PCP, aggressive work up for co-infection in non-responsive
human immunodeficiency virus infection: outcome and asso-
patients, and protective lung ventilation strategies are
ciated features. Chest 2005;128:5739.
needed to improve survival. 16. Forrest DM, Djurdjev O, Zala C, Singer J, Lawson L, Russell JA,
et al. Validation of the modified multisystem organ failure score as
Acknowledgment a predictor of mortality in patients with AIDS-related Pneumocys-
tis carinii pneumonia and respiratory failure. Chest 1998;114:
199206.
The authors thank Dr Amnuay Thitapandha for constructive 17. Ewig S, Bauer T, Schneider C, Pickenhain A, Pizzulli L, Loos U,
suggestions and English editing. et al. Clinical characteristics and outcome of Pneumocystis
Conflict of interest: We all declare that we do not have a carinii pneumonia in HIV-infected and otherwise immunosup-
conflict of interest and that we do not have a financial pressed patients. Eur Respir J 1995;8:154853.
relationship with a commercial entity that has an interest 18. Broaddus C, Dake MD, Stulbarg MS, Blumenfeld W, Hadley WK,
in the subject of this manuscript. Golden JA, et al. Bronchoalveolar lavage and transbronchial
biopsy for the diagnosis of pulmonary infections in the acquired
immunodeficiency syndrome. Ann Intern Med 1985;102:74752.
References 19. Levine SJ. Pneumocystis carinii. Clin Chest Med 1996;17:665
95.
1. Montgomery AB. Pneumocystis carinii pneumonia in patients 20. Jules-Elysee KM, Stover DE, Zaman MB, Bernard EM, White DA.
with the acquired immunodeficiency syndrome: pathophysiol- Aerosolized pentamidine: effect on diagnosis and presentation
ogy, therapy and prevention. Semin Respir Dis 1989;4:10210. of Pneumocystis carinii pneumonia. Ann Intern Med 1990;112:
2. Rogers PL, Lane HC, Henderson DK, Parrillo J, Masur H. Admission 7507.
of AIDS patients to a medical intensive care unit: causes and 21. Benson CA, Spear J, Hines D, Pottage JC, Kessler HA, Trenholme
outcome. Crit Care Med 1989;17:1137. GM. Combined APACHE II score and serum lactate dehydrogenase
3. Masur H, Ognibene FP, Yarchoan R, Shelhamer JH, Baird BF, Travis as predictors of in-hospital mortality caused by first episode
W, et al. CD4 counts as predictors of opportunistic pneumonias in Pneumocystis carinii pneumonia in patients with acquired immu-
human immunodeficiency virus (HIV) infection. Ann Intern Med nodeficiency syndrome. Am Rev Respir Dis 1991;144:31923.
1989;111:22331. 22. Torres HA, Chemaly RF, Storey R, Aguilera EA, Nogueras GM,
4. Kovacs JA, Ng VL, Masur H, Leoung G, Hadley WK, Evans G, et al. Safdar A, et al. Influence of type of cancer and hematopoietic
Diagnosis of Pneumocystis carinii pneumonia: improved detec- stem cell transplantation on clinical presentation of Pneumo-
tion in sputum with use of monoclonal antibodies. N Engl J Med cystis jiroveci pneumonia in cancer patients. Eur J Clin Microbiol
1988;318:58993. Infect Dis 2006;25:3828.
5. Masur H. Prevention and treatment of Pneumocystis carinii 23. Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M,
pneumonia. N Engl J Med 1993;327:185360. Bastos PG, et al. The APACHE III prognostic system: risk predic-
6. Mansharamani NG, Garland R, Delaney D, Koziel H. Management tion of hospital mortality for critically ill hospitalized adults.
and outcome patterns for adult Pneumocystis carinii pneumonia, Chest 1991;100:161936.
1985 to 1995: comparison of HIV-associated cases to other 24. Bedos JP, Dumoulin JL, Gachot B, Veber B, Wolff M, Regnier B, et al.
immunocompromised states. Chest 2000;118:70411. Pneumocystis carinii pneumonia requiring intensive care manage-
7. Arend SM, Kroon FP, Vant Wout JW. Pneumocystis carinii pneu- ment: survival and prognostic study in 110 patients with human
monia in patients without AIDS, 1980 through 1993: an analysis of immunodeficiency virus. Crit Care Med 1999;27:110915.
78 cases. Arch Intern Med 1995;155:243641. 25. Peruzzi WT, Skoutelis A, Shapiro BA, Murphy RM, Currie DL, Cane
8. Sepkowitz KA. Pneumocystis carinii pneumonia in patients with- RD, et al. Intensive care unit patients with acquired immuno-
out AIDS. Clin Infect Dis 1993;17(Suppl):S41622. deficiency syndrome and Pneumocystis carinii pneumonia: sug-
9. Yale SH, Limper AH. Pneumocystis carinii pneumonia in patients gested predictors of hospital outcome. Crit Care Med 1991;19:
without acquired immunodeficiency syndrome: associated ill- 892900.
ness and prior corticosteroid therapy. Mayo Clin Proc 1996;71: 26. Friedman Y, Franklin C, Rackow EC, Weil MH. Improved survival in
513. patients with AIDS, Pneumocystis carinii pneumonia, and severe
10. Curtis JR, Yarnold PR, Schwartz DN, Weinstein RA, Bennett CL. respiratory failure. Chest 1989;96:8626.
Improvements in outcomes of acute respiratory failure for 27. Maxfield RA, Sorkin IB, Fazzini EP, Rapoport DM, Stenson WM,
patients with human immunodeficiency virus-related Pneumo- Goldring RM. Respiratory failure in patients with acquired immu-
cystis carinii pneumonia. Am J Respir Crit Care Med 2000;162: nodeficiency syndrome and Pneumocystis carinii pneumonia.
3938. Crit Care Med 1986;14:4439.
11. Forrest DM, Zala C, Djurdjev O, Singer J, Craib KJ, Lawson L, 28. Kales CP, Murren JR, Torres RA, Crocco JA. Early predictors of in-
et al. Determinants of short- and long-term outcome in patients hospital mortality for Pneumocystis carinii pneumonia in the

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.
66 V. Boonsarngsuk et al.

acquired immunodeficiency syndrome. Arch Intern Med 1987; 35. Sepkowitz KA, Brown AE, Armstrong D. Pneumocystis carinii
147:14137. pneumonia without acquired immunodeficiency syndrome: more
29. Benfield TL, Helweg-Larsen J, Bang D, Junge J, Lundgren JD. patients, same risk. Arch Intern Med 1995;155:11258.
Prognostic markers of short-term mortality in AIDS-associated 36. Sepkowitz KA. Pneumocystis carinii pneumonia without acquired
Pneumocystis carinii pneumonia. Chest 2001;119:84451. immunodeficiency syndrome: who should receive prophylaxis?
30. Jensen AM, Lundgren JD, Benfield T, Nielsen TL, Vestbo J. Does Mayo Clin Proc 1996;71:1023.
cytomegalovirus predict a poor prognosis in Pneumocystis carinii 37. Fan E, Needham DM, Stewart TE. Ventilatory management of
pneumonia treated with corticosteroids? A note for caution. acute lung injury and acute respiratory distress syndrome. JAMA
Chest 1995;108:4114. 2005;294:288996.
31. Nelson MR, Erskine D, Hawkins DA, Gazzard BG. Treatment with 38. The Acute Respiratory Distress Syndrome Network. Ventilation
corticosteroidsa risk factor for the development of clinical with lower tidal volumes as compared with traditional tidal
cytomegalovirus disease in AIDS. AIDS 1993;7:3758. volumes for acute lung injury and the acute respiratory distress
32. Sowden E, Carmichael AJ. Autoimmune inflammatory disorders, syndrome. N Engl J Med 2000;342:13018.
systemic corticosteroids and Pneumocystis pneumonia: a strat- 39. Stewart TE, Meade MO, Cook DJ, Granton JT, Hodder RV, Lapinsky
egy for prevention. BMC Infect Dis 2004;4:42. SE, et al. Evaluation of a ventilation strategy to prevent baro-
33. Pagano L, Fianchi L, Mele L, Girmenia C, Offidani M, Ricci P, et al. trauma in patients at high risk for acute respiratory distress
Pneumocystis carinii pneumonia in patients with malignant syndrome. Pressure- and Volume-Limited Ventilation Strategy
haematological diseases: 10 years experience of infection in Group. N Engl J Med 1998;338:35561.
GIMEMA centres. Br J Haematol 2002;117:37986. 40. Brochard L, Roudot-Thoraval F, Roupie E, Delclaux C, Chastre
34. Beers MF, Sohn M, Swartz M. Recurrent pneumothorax in AIDS J, Fernandez-Mondejar E, et al. Tidal volume reduction for
patients with Pneumocystis pneumonia. A clinicopathologic prevention of ventilator-induced lung injury in acute respira-
report of three cases and review of the literature. Chest tory distress syndrome. Am J Respir Crit Care Med 1998;
1990;98:26670. 158:18318.

Downloaded for kaprodi bedah (kps_bedah@fk.unand.ac.id) at ClinicalKey Global Guest Users from ClinicalKey.com by Elsevier on April 12, 2017.
For personal use only. No other uses without permission. Copyright 2017. Elsevier Inc. All rights reserved.