Pneumocystis carinii Pneumonia: A Comparison Between Patients with the

Acquired Immunodeficiency Syndrome and Patients with Other

Immunodeficiencies
JOSEPH A. KOVACS, M.D.; JOHN W. HIEMENZ, M.D.; ABE M. MACHER, M.D.; DIANE STOVER, M.D.;
HENRY W. MURRAY, M.D.; JAMES SHELHAMER, M.D.; H. CLIFFORD LANE, M.D.; CARLOS
URMACHER, M.D.; CHRISTINE HONIG, M.D.; DAN L LONGO, M.D.; MARGARET M. PARKER, M.D.;
CHARLES NATANSON, M.D.; JOSEPH E. PARRILLO, M.D.; ANTHONY S. FAUCI, M.D.; PHILIP A.
PIZZO, M.D.; and HENRY MASUR, M.D.; Bethesda, Maryland; and New York, New York

Clinical features of 4 9 episodes of Pneumocystis carinii other centers using similar immunosuppressive regimens,
pneumonia in patients with the acquired attack rates of 22% to 43% were reported in certain chil-
immunodeficiency syndrome were compared with those of
dren receiving several antineoplastic agents ( 4 ) .
39 episodes in patients with other immunosuppressive
diseases At presentation patients with the syndrome The P. carinii pneumonia that has occurred in patients
were found to have a longer median duration of symptoms with the acquired immunodeficiency syndrome seems to
( 2 8 days versus 5 days,/? = 0 . 0 0 0 1 ) , lower mean have clinical features that differ from the disease seen in
respiratory rate (23.4 versus 30, p = 0 . 0 0 5 ) , and higher patients with previously known immunosuppressive ill-
median room air arterial oxygen tension ( 6 9 mm Hg
versus 52 mm Hg,p = 0 . 0 0 0 2 ) . The survival rate from nesses. Anecdotal reports have suggested that P. carinii
1979 to 1983 was similar for the two groups ( 5 7 % and pneumonia in the syndrome presents with more subtle
5 0 % respectively). Patients with the syndrome had a initial symptoms, that the survival rate is unusually poor,
higher incidence of adverse reactions to trimethoprim- and that adverse drug reactions are a common problem.
sulfamethoxazole (22 of 34 versus 2 of 17, p = 0 . 0 0 0 7 ) .
Survivors with the syndrome at initial presentation had a
This study was done to compare clinical features of P.
significantly lower respiratory rate, and higher room air carinii pneumonia in patients with the acquired immuno-
arterial oxygen tension, lymphocyte count, and serum deficiency syndrome with features in patients with other
albumin level compared to nonsurvivors. Pneumocystis immunosuppressive diseases. We report that P. carinii
carinii pneumonia presents as a more insidious disease pneumonia presents as a more insidious disease process
process in patients with the syndrome, and drug therapy
in these patients is complicated by frequent adverse in patients with the acquired immunodeficiency syn-
reactions. drome compared to patients with other immunosuppres-
sive diseases, and that drug therapy in patients with the
PNEUMOCYSTIS CARINII PNEUMONIA is one of the com- syndrome is complicated by an unusually high frequency
monest life-threatening opportunistic infections in pa- of adverse reactions.
tients with the acquired immunodeficiency syndrome.
Materials and Methods
The frequency of P. carinii pneumonia in this patient
Charts of all patients 10 years or older with P. carinii pneu-
population has been high: the Centers for Disease Con- monia diagnosed at the National Institutes of Health (NIH)
trol (CDC) has reported that 59% of patients with the from January 1970 through June 1983 were reviewed retro-
syndrome have had P. carinii pneumonia at the time of spectively. In order to supplement the number of cases of P.
CDC notification (1). Undoubtedly a higher percentage carinii pneumonia in patients with the acquired immunodefici-
of patients have had one or more episodes of P. carinii ency syndrome, available charts of patients with the syndrome
and Pneumocystis pneumonia at The New York Hospital and
pneumonia before death. If epidemiologic data from the Memorial-Sloan Kettering Cancer Center from 1979 through
CDC are used to calculate the frequency of P. carinii March 1983 were also reviewed retrospectively. A diagnosis of
pneumonia in this population, an attack rate of at least P. carinii pneumonia was based on demonstration of three or
30% per year is estimated ( 2 ) . Such an annual attack more typical organisms in specimens of pulmonary secretions
rate is much higher than the 0.01% to 1.1% rate that the or lung tissue obtained by bronchoscopy, transthoracic biopsy,
open lung biopsy, or autopsy.
CDC estimated for patients with leukemias, lymphomas, With regard to underlying disease, diagnosis of the syndrome
and renal transplants between 1967 and 1970 ( 3 ) . The was based on CDC criteria (5). Patients were designated as
attack rate in patients with the acquired immunodeficien- having other immunosuppressive diseases if they had any other
cy syndrome is comparable only to the unique population underlying disease process such as a malignant neoplastic dis-
ease (other than Kaposi's sarcoma) or a collagen vascular dis-
at St. Jude Children's Research Hospital where, unlike ease.
Symptoms, signs, and duration of symptoms were those rec-
• From the Critical Care Medicine Department, Clinical Center, Pediatric
Branch, Medicine Branch, and Laboratory of Pathology, National Cancer Insti-
orded at the time of initial recognition of a pulmonary process
tute, Laboratory of Immunoregulation, National Institute of Allergy and Infec- subsequently confirmed to be P. carinii pneumonia. The time of
tious Diseases, National Institutes of Health, Bethesda, Maryland; the Pulmonary initial recognition of a pulmonary process was defined as the
Division, Department of Medicine, and Department of Pathology, Memorial- day when a physician's evaluation led to documentation in the
Sloan Kettering Cancer Center, and the Infectious Disease Division, Department
of Medicine, and Department of Pathology, The New York Hospital-Cornell patient's record of a suspicion of pneumonia, based on an ab-
Medical Center, N e w York, New York. normal chest radiograph or suggestive symptoms—specifically,
Annals of Internal Medicine. 1984;100:663-671. 663

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 Table 1 . Comparison of Epidemiologic Features in Patients with Pneumocystis Pneumonia

Patients with Patients with Other

the Acquired Immunosuppressive Diseases
Immunodeficiency Syndrome (n = 39)
(77 = 46)

Median age (range), yrs 35 (25-57) 33 (10-69)

Male/female, 77 45/1 26/13
Risk factors, 77 (%)
Homosexual or bisexual 42 (91)
Intravenous drug user 1 (2)
Haitian 1 (2)
Hematologic malignancy ... 30 (77)
Vasculitis ... 5 (13)
Other ... 4(10)
N o known risk factor 2 (4)
Immunosuppressive therapy, 77 (%) *
Steroids only 1 (2) 3 (8)
Cytotoxic therapy only 2 (4) 3 (8)
Both 1 (2) 30 (76)
Neither 42(91) 3 (8)

* Within 1 m o n t h of onset of symptoms.

cough, shortness of breath, dyspnea, or chest pain. This point, fall of at least 1000 cells/mm 3 below pretreatment levels.
which in most cases coincided with the day of admission to the Thrombocytopenia was defined as a complication of drug thera-
hospital, was chosen rather than the day of diagnosis or initia- py if the platelet count fell below 50 000 cells/mm 3 with a fall
tion of therapy because physical examination and laboratory of at least 50 000 cells/mm 3 below pretreatment levels.
data were most complete on admission. The delay between sus- To compare whether the proportions in two groups for differ-
picion of pneumonia and diagnosis or treatment was usually ent variables were statistically equivalent, the Fisher exact test
less than 4 days. Laboratory values reported were those shown was used ( 7 ) . To evaluate whether two independent groups of
at initial presentation or within 96 hours of presentation. N o values were statistically equivalent, the median was determined
laboratory values recorded more than 48 hours after the initia- and the nonparametric Wilcoxon rank sum test was used ( 8 ) .
tion of anti-pneumocystis treatment were used for this analysis. In all cases two-sided p values were used. Because not all data
A patient was considered afebrile at the time of presentation if, were available for all patients in this retrospective review, re-
on the day of presentation, his maximum temperature was less sults are expressed explicitly to show the total number of evalu-
than 38.0 °C. A patient was considered to have defervesced if, able cases for each result.
on two consecutive days, his maximum temperature was less
than 38.0 °C. Room air alveolar-arterial oxygen gradient was
calculated by conventional methods ( 6 ) . Results
Chest radiographs were reviewed by two specialists in pulmo- C o m p l e t e clinical records were available for analysis
nary medicine when the films were available. The official radiol- for all 39 episodes of P. carinii p n e u m o n i a in 39 patients
ogy department reports were reviewed when films were unavail-
with other immunosuppressive diseases, and for 49 epi-
able.
Lung histopathologic findings (including specimens stained sodes in 4 6 patients with the acquired immunodeficiency
both by hematoxylin and eosin and by either Gomori's methe- s y n d r o m e out of a total of 57 episodes in 53 patients w i t h
namine silver or toluidine blue-O) were reviewed by a patholo- the syndrome. Eleven patients with the s y n d r o m e were
gist for presence of cellular infiltrate, alveolar exudate, and P. from the N a t i o n a l Institutes of Health, 16 were from T h e
carinii cysts. The quantity of organisms was considered "few" if
three or fewer alveoli per specimen contained cysts; the quanti- N e w Y o r k Hospital, and 19 were from M e m o r i a l - S l o a n
ty was considered "many" if more than three alveoli contained Kettering Cancer Center. N o differences were seen
cysts. Cytomegalovirus was diagnosed histopathologically by a m o n g the patients from the three centers. R a d i o g r a p h s
the presence of one or more cells with typical nuclear or cyto- were reviewed by o n e of the authors in 32 of 39 episodes
plasmic inclusions.
in patients w i t h other i m m u n o s u p p r e s s i v e diseases and 27
In evaluating episodes of P. carinii pneumonia, outcome was
classified as "death associated with P. carinii pneumonia" if of 4 9 episodes in patients w i t h the syndrome; the remain-
either of the following criteria was met: death during anti-pneu- ing episodes were reviewed radiographically by m e a n s of
mocystis treatment; or, death within 8 weeks of completion of the official radiology reports. H i s t o p a t h o l o g i c findings
anti-pneumocystis treatment, unless death occurred after clini- from biopsies were available for review in 30 episodes in
cal improvement and was clearly secondary to a process unre-
lated to Pneumocystis pneumonia, anti-pneumocystis treat- patients w i t h the s y n d r o m e and 32 episodes in patients
ment, or pulmonary dysfunction. w i t h other i m m u n o s u p p r e s s i v e diseases. Episodes o f P.
Adverse effects of anti-pneumocystis treatment were assessed carinii p n e u m o n i a for w h i c h n o records were available
by looking specifically for previously reported complications, as did not differ from episodes with available records in
well as any other factors that appeared to be treatment related,
based on the clinical record. Adverse effects were evaluated in terms of patient risk factor for the s y n d r o m e or in terms
patients for whom data were available while the patients were of survival from P. carinii p n e u m o n i a .
on a single anti-pneumocystis drug. Specific parameters for
some patients could not be analyzed because of severe abnor- CLINICAL F E A T U R E S
malities at the time drug therapy was initiated, concurrent ad- Table 1 s h o w s a c o m p a r i s o n of the e p i d e m i o l o g i c fea-
ministration of toxic agents, or because appropriate tests were
not ordered. Leukopenia was defined as a complication of drug tures of these patients. O f the episodes of P. carinii pneu-
therapy if the leukocyte count fell below 3000 cells/mm 3 with a m o n i a in patients with the s y n d r o m e , 14 occurred in

664 May 1984 • Annals of Internal Medicine • Volume 100 • Number 5

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 Table 2. Clinical and Laboratory Features at Presentation in Patients with Pneumocystis Pneumonia

Patients with the Acquired Patients with Other

Immunodeficiency Syndrome Immunosuppress >ive Diseases
Evaluable Number of Patients Evaluable Number of Patients
Patients and Values Patients and Values
Symptoms, n (%)
Fever 48 38 (81) 38 33 (87)
Chills 48 12 (26) 38 10 (26)
Shortness of breath 48 32 (68) 38 25 (66)
Cough 48 38 (81) 38 27 (71)
Sputum 48 11 (23) 38 8 (21)
Chest pain 48 11 (23) 38 9 (24)
Median duration of symptoms (range)*, d 40 28 (1-270) 37 5 (1-42)
Temperature > 38.0 °C, n (%) 45 34 (76) 38 35 (92)
Median respiratory rate per minute (range)f 42 24 (14-40) 38 26 (16-60)
Rales, n (%) 46 14 (30) 38 13 (34)
Chest radiograph, n (%)
Bilateral infiltrate 49 46 (94) 39 34 (87)
Unilateral infiltrate 49 1 (2) 39 2 (5)
N o infiltrate 49 2 (4) 39 3 (8)
Median blood gas values (range), mm Hg
Room air arterial oxygen tension* 45 69 (35-116) 33 52 (29-91)
Alveolar-arterial oxygen gradient! 45 41 (1-99) 33 59 (23-91)
Median leukocyte count (range), /mm3
Total leukocytes 47 4900 (1500-14 000) 37 4600 (200-156 000)
Neutrophils 42 3315 (1008-12 600) 32 3562 (80-11 590)
Lymphocytes 42 643 (37-2320) 32 396 (0-63 700)
Median albumin level (range), g/dL 34 , 3.25 (1.3-4.0) 27 3.1 (2.0-4.0)
i
* p < 0.0002.
t p = 0.015.

patients with a history of Kaposi's sarcoma only, 13 in versus 5 days, p = 0.0001, and 28 days versus 8 days,
patients with a history of opportunistic infection only, 5 p = 0.0001, respectively).
in patients with both, and 17 in patients with neither.
Four patients with the syndrome had received corticos- LABORATORY F E A T U R E S
teroids or cytotoxic treatment in the month before pre- At the time the patients were recognized by a physi-
sentation for the treatment of cerebral edema, thrombo- cian to have a pulmonary disorder, patients with the ac-
cytopenia, or Kaposi's sarcoma. Eight patients with the quired immunodeficiency syndrome were less likely to be
syndrome had been treated with alpha interferon (recom- febrile (p = 0.08), had a lower respiratory rate (24
binant or lymphoblastoid), thymosin, or both. Only 1 breaths per minute versus 26 breaths per minute,
patient with the syndrome had received trimethoprim- p = 0.015), and were less hypoxemic on room air (69
sulfamethoxazole within 2 months of presentation. (That mm Hg versus 52 mm Hg, p = 0.0002), with a lower
patient had received trimethoprim-sulfamethoxazole pro- room air alveolar-arterial gradient (41 mm Hg versus 59
phylaxis, one double-strength tablet twice daily, for 2 mm Hg, p = 0.0001). Patients with the syndrome were
days before the onset of symptoms.) The patients with more likely to have a room air arterial oxygen tension of
other immunosuppressive diseases included 17 patients 80 mm Hg or greater (10 of 45 episodes) than were pa-
with lymphoma, 13 with leukemia, 2 with systemic lupus tients with other immunosuppressive diseases (3 of 33
erythematosus, 3 with other vasculitides, 2 with carcino- episodes) (p = 0.22). Most of the patients that were afe-
ma, 1 with Ewing's sarcoma, and 1 with aplastic anemia. brile at initial presentation developed a fever within the
Thirty-six of these thirty-nine patients with other immu- subsequent 48 hours. Although the median respiratory
nosuppressive diseases had received corticosteroids or cy- rates appear similar, they are significantly different. The
totoxic treatment in the month before presentation. N o median understates the difference in respiratory rate be-
patient had received trimethoprim-sulfamethoxazole tween the two groups; it does not adequately show the
within 2 months of presentation. moderate number of patients with other immunosuppres-
The clinical features of the two groups at the time of sive diseases with high respiratory rates. The mean
presentation are shown in Table 2. Specific symptoms ( ± SD) respiratory rate for patients with the syndrome
ocurred with similar frequencies in both groups. The me- is 23.4 ± 6 . 8 breaths per minute and that for patients
dian duration of symptoms before recognition of a pul- with other immunosuppressive diseases is 30 + 12.4
monary disorder, and the median duration of symptoms (p = 0.005, Student's r-test).
before documentation of P. carinii pneumonia or initia- Evaluation of other laboratory parameters showed that
tion of anti-pneumocystis therapy were significantly long- the leukocyte count was similar in both patient groups,
er in patients with the syndrome compared to that in and that both groups were lymphopenic but with no sig-
patients with other immunosuppressive diseases (28 days nificant difference between them (p = 0.18). Both pa-
Kovacs et al. • Pneumocystis carinii Pneumonia 665

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 Table 3. Survival and Alteration of Therapeutic Regimen in Patients with Pneumocystis Pneumonia

Patients with the Acquired Patients with Other

Immunodeficiency Immunosuppressive
Syndrome Diseases

< n (% ; •

Proportion of episodes survived

Overall survival 28/49 (57) 16/39 ( 4 1 ) *
Survival, 1979-1983 28/49 (57) 8/16 (50)

As initial treatment 24/40 (60) 9/17 (53)

As only treatment 14/40 (35) 6/17 (35)

As initial treatment 4 / 9 (44) 5/15 (33)

As only treatment 4 / 9 (44) 5/15 (33)
Proportion of anti-pneumocystis treatment episodes altered J
Trimethoprim-sulfamethoxazole as initial treatment
Alteration due to disease progression 16/40 (40) 10/17 (59)
Alteration due to adverse effects 7/40 (18) 0/17 (0)

Alteration due to disease progression 3/9 (33) 0/15 (0)

Alteration due to adverse effects 1/9 (11) 0/15 (0)

* Seven patients with other immunosuppressive diseases received either no treatment (1 patient) or pyrimethamine and sulfadiazine (six patients) as initial treatment.
t Denominator represents total number of patients initially receiving each anti-pneumocystis drug.
X Addition or substitution of anti-pneumocystis drug not used initially, for all patients (survivors and nonsurvivors).

tient groups had a mild decrease in serum albumin. represented blood contamination, because hematogenous
There was no difference between the two patient dissemination of these organisms is common in the ac-
groups in chest radiographic patterns at presentation. quired immunodeficiency syndrome ( 9 ) .
Most patients had diffuse bilateral interstitial or alveolar
pulmonary infiltrates. However, infiltrates were occasion- OUTCOME
ally patchy or asymmetrical. A few patients in each The outcome of P. carinii pneumonia was similar be-
group had either unilateral infiltrates or normal findings tween the two groups, when assessed by overall survival
on chest radiograph. In the latter patients respiratory (p = 0.20) (Table 3), and particularly when the same
symptoms drew attention to the need to consider the pos- chronologic period (1979 to 1983) was analyzed
sibility of a pulmonary disorder. (p = 0.84). In terms of response to treatment, there was
On review of histopathologic findings from biopsies, no difference in median duration of fever between survi-
considerable variation was found in the number and qual- vors with the syndrome and survivors with the other im-
ity of lung samples available for each patient. In addition, munosuppressive diseases (5.5 days for 18 patients versus
9 of 30 patients with the syndrome and 12 of 32 patients 5.0 days for 13 patients, respectively). However, more
with other immunosuppressive diseases had received 1 to survivors with the syndrome remained febrile for at least
4 days of anti-pneumocystis treatment before biopsy. 10 days (7 of 18 versus 1 of 13, p = 0.12). This group
Evaluation under these circumstances showed no differ- also needed a slightly longer median period to show ini-
ence between the two groups in terms of quantity or qual- tial radiographic improvement (7.5 days for 24 patients
ity of cellular infiltrate or alveolar exudate. However, versus 5 days for 15 patients, p = 0.24). Respiratory rate
"few" organisms were seen in more patients with other and arterial blood gases were measured at such variable
immunosuppressive diseases (12 of 32) than patients intervals with so much variation in fractional inspired
with the syndrome (3 of 30) (p = 0.024). oxygen that meaningful comparison was not possible.
On the basis of histopathologic findings, other patho- Ventilatory assistance for reasons other than a diagnostic
gens did not appear to be important contributors to pul- procedure was required in 19 patients with other immu-
monary dysfunction. Cytomegalovirus was identified his- nosuppresive diseases, 6 of whom survived, and 16 pa-
topathologically in 4 of 30 biopsy samples from patients tients with the acquired immunodeficiency syndrome,
with the syndrome and 3 of 32 biopsy samples in patients none of whom survived (p = 0.034).
with other immunosuppressive diseases, but inclusion Patients who survived received the same median num-
cells were usually rare in number. One patient with the ber of days of treatment in both groups (14 days for each
syndrome had histopathologic evidence of cryptococcus group). For nonsurvivors who received at least 4 days of
infection. N o other pathogens were seen histopathologi- treatment, patients with the syndrome received longer
cally. Cytomegalovirus, Cryptococcus neoformans, and median treatment (21 days) than patients with other im-
Mycobacterium avium-intracellulare were cultured from munosuppressive diseases (11 days) (p = 0.007), pri-
some specimens but their importance was difficult to in- marily because patients were usually treated until they
terpret because some cytomegalovirus may have repre- died, and patients with the syndrome lingered longer be-
sented oropharangeal contamination during the trans- fore death. One patient with the syndrome and three pa-
bronchial approach and some of the cytomegalovirus, tients with other immunosuppressive diseases received
cryptococcus, and M. avium-intracellulare undoubtedly less than 4 days of treatment.
666 May 1984 • Annals of Internal Medicine • Volume 100 • Number 5

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 Table 4. Adverse Effects Due to Anti-pneumocystis Therapy in Patients with Pneumocystis Pneumonia4*

Treatment and Adverse Effects Patients with the Acquired Patients with Other
Immunodeficiency Immunosuppressive Diseases
Syndrome

< — n(%)- •

Trimethoprim-sulfamethoxazole
Rashf 10/34 (29) 0/15 (0)
Leukopeniaft 14/30 (47) 1/12 (8)
Thrombocytopenia§ 2/32 (6) 0/11 (0)
Creatinine increase > 0.3 m g / d L 3/23 (13) 1/14 (7)
Any adverse effectf 22/34 (65) 2/17 (12)
Pentamidine
Rash 1/17 (6) 0/21 (0)
Leukopenia:): 1/16 (6) 0/15 (0)
Thrombocytopenia§ 1/13 (8) 1/12 (8)
Creatinine increase > 0.3 m g / d L 5/12 (42) 8/16 (50)
Glucose < 50 mg/dL 2/17 (12) 2/21 (10)
Sterile abscess 2/17 (12) 1/21 (5)
Hypotension 0/17 (0) 2/21 (10)
Any adverse effect 8/17 (47) 12/21 (57)
* Specific adverse effects were evaluated while on a single anti-pneumocystis drug if follow-up data were available and toxic agents were not administered concurrently.
t p <0.05.
% Less than 3 0 0 0 / m m 3 , and a decrease of at least 1 0 0 0 / m m 3 below baseline.
§ Less than 50 0 0 0 / m m 3 , and a decrease of at least 50 0 0 0 / m m 3 below baseline.

Respiratory dysfunction appeared to play a major role spectively). A total of 20 of 34 evaluable patients with
in the death of at least 17 patients with the syndrome and the syndrome had either or both of these complications; 7
20 patients with other immunosuppressive diseases. At required an alteration in treatment as a result, compared
autopsy, Pneumocystis was usually found both in pa- to no patients with other immunosuppressive diseases.
tients with the syndrome (10 of 15) and patients with Adverse effects due to pentamidine were common in both
other immunosuppressive diseases (15 of 20). Autopsies groups. Creatinine rises greater than 1.0 mg/dL occurred
were done a median of 27 days (range, 6 to 74 days) after in only 3 patients with the syndrome and 4 patients with
initiation of anti-pneumocystis treatment in patients with other immunosuppressive diseases.
the syndrome, and 9.5 days (range, 2 to 29 days) after To determine if there were any clinical or laboratory
initiation of treatment in patients with other immunosup- features at presentation that had prognostic importance
pressive diseases. Other pathogens were more commonly for patients with the syndrome, the features shown in
seen at autopsy than on lung biopsy samples. Cytomega- Table 2 were compared for patients with the syndrome
lovirus infection was shown histopathologically at autop- who survived and those who did not (Table 5). The fol-
sy in 10 of 15 patients with the syndrome but only in 4 of lowing features correlated significantly with survival in
20 patients with other immunosuppressive diseases patients with the acquired immunodeficiency syndrome:
(p = 0.014). lower respiratory rate (p = 0.03), higher room air arteri-
As shown in Table 3, trimethoprim-sulfamethoxazole al oxygen tension (p = 0.04), lower alveolar-arterial gra-
and pentamidine appeared to be equally efficacious when dient (p = 0.03), higher lymphocyte count (p = 0.025),
survival was used as the criterion for efficacy. (Six pa- and higher albumin levels (p = 0.007). Ten of twenty-
tients with other immunosuppressive diseases were treat- three patients with the syndrome who survived had lym-
ed initially with pyrimethamine and sulfadiazine before phocyte counts greater than 1000 cells/mm 3 , and 21 of
1975, and are not included in assessment of treatment.) 26 had alveolar-arterial gradients less than 50 mm Hg, as
Conventional doses (pentamidine, 4 mg/kg body compared to 1 of 17 and 7 of 17 respectively for nonsur-
weight d, and trimethoprim, 20 mg/kg body weight d, vivors (p = 0.018 a n d p = 0.02 respectively).
with sulfamethoxazole, 100 mg/kg body weight-d) were
almost always used. Valid, unbiased comparisons of drug Discussion
efficacy were not possible in this retrospective study, Pneumocystis carinii pneumonia has been one of the
however, because initiation and modification of therapeu- most frequent life-threatening processes in patients with
tic regimens were determined by clinicians based on their the acquired immunodeficiency syndrome. The Pneumo-
own experiences and prejudices. In addition, pentamidine cystis pneumonia seen in these patients has striking clini-
was the only effective treatment available in the early cal differences from the disease seen in patients with
1970s, a time when intensive care support was inferior to other immunosuppressive diseases in terms of clinical
that in the later years. presentation and frequency of adverse reactions to anti-
Adverse effects to trimethoprim-sulfamethoxazole pneumocystis treatment.
were surprisingly common for patients with the acquired Analysis of 49 episodes of P. carinii pneumonia in pa-
immunodeficiency syndrome as shown in Table 4. Rash tients with the syndrome and 39 episodes in patients with
and leukopenia were significantly more common in pa- other immunosuppressive diseases shows that in the ac-
tients with the syndrome (p = 0.032 a n d p = 0.038, re- quired immunodeficiency syndrome this pneumonia has a
Kovacsetal. • Pneumocystis carinii Pneumonia 667

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 Table 5. Clinical and Laboratory Features at Presentation of Survivors and Nonsurvivors with Pneumocystis Pneumonia and the Ac-
quired Immunodeficiency Syndrome

Survivo]rs Nonsurvi ivors

Evaluable Numt>er of Patients Evaluable Num ber of Patients
Patients aiid Values Patients aind Values

Symptoms, n (%)
Fever 28 22 (79) 20 17 (85)
Shortness of breath 28 18 (64) 20 15 (75)
Cough 28 24 (86) 20 15 (75)
Median duration of symptoms (range), d 25 28 (1-270) 13 14 (2-150)
Temperature > 38.0 °C, n (%) 26 17 (65) 19 17 (89)
Median respiratory rate per minute (range)* 23 20 (14-36) 17 24 (16-40)
IVXCUldll UIDLIU g d a VdlUCa ^ I d l l g C J , JIlIIl ±lg
Room air arterial oxygen tension 26 71 (47-108) 17 60 (35-116)
Alveolar-arterial gradient* 26 39 (9-61) 17 54 (1-99)

Total leukocytes 26 4900 (1500-9000) 19 4900 (1800-14 000)

Neutrophils 23 3111 (1260-7740) 17 3920 (1008-12 600)
Lymphocytes* 23 912 (66-2230) 17 504 (37-1014)
Median albumin level (range)*, g/dL 23 3.4 (1.4-4.0) 11 2.8 (1.3-3.6)
*p < 0.04.

much more subtle and prolonged clinical presentation. this study by the higher median initial RA02 seen in
Although the frequency of specific complaints such as these patients versus patients with other immunosuppres-
fever, cough, dyspnea, and chest pain were similar in sive diseases (p = 0.0002). Previous studies have tended
both groups, patients with the acquired immunodeficien- to confirm this finding. Winston and colleagues (16) re-
cy syndrome had symptoms for a median of 28 days, ported an initial RA02 of less than 60 mm Hg in 11 of 11
versus 5 days for patients with other immunosuppressive patients with other immunosuppressive diseases; Sattler
diseases (p = 0.0001) (Table 2). This difference in dura- and Remington (17) reported an RA02 of 39 to 88 mm
tion of symptoms has also been found in other studies. Hg (mean, 53 mm Hg) in 15 of 15 patients with other
For patients with other immunosuppressive diseases Wal- immunosuppressive diseases; and Walzer and associates
zer and associates (3) reported that 52% of 194 patients (3) reported an RA02 of 21 to 70 mm Hg (mean, 46 mm
had symptoms for less than 2 weeks, and 83% for less Hg) in all 42 patients with other immunosuppressive dis-
than 4 weeks. Lau and Young (10) reported symptoms eases for whom RA02 were available. Documentation of
for 1 to 10 days (median, 3 days) in 8 patients with other initial RA02 in patients with the acquired immunodefici-
immunosuppressive diseases. In contrast, recent reports ency syndrome has been less thorough: wide ranges are
of P. carinii pneumonia in patients with the acquired im- reported without individual values or statistical analysis
munodeficiency syndrome have shown symptoms for 10 (12, 18).
days to 4 months in 11 patients (11), 3 days to 6 months In this series the diagnosis of P. carinii pneumonia was
in 11 patients (12), 3 months in 2 patients (13), and 3 to established by finding Pneumocystis cysts in lung biopsy
14 days in 4 patients (14). samples for most cases. Analysis of available tissue sug-
If the natural history of Pneumocystis pneumonia were gested that there was no consistent difference in histopa-
the same in both patient populations, patients with the thologic response between patients with the acquired
acquired immunodeficiency syndrome should have had immunodeficiency syndrome and patients with other im-
more florid disease at the time of initial clinical and labo- munosuppressive diseases when the quantity of cellular
ratory assessment. However, patients with the acquired infiltrate, the cell type involved, or the extent of alveolar
immunodeficiency syndrome were less febrile, less exudate were analyzed. Patients with the syndrome were
tachypneic, and less hypoxemic than patients with other more likely to have many P. carinii cysts than patients
immunosuppressive diseases when P. carinii pneumonia with other immunosuppressive diseases, but the small
was recognized. Analysis of numerous individual cases sample sizes inherent in transbronchial biopsies and the
suggests that in the acquired immunodeficiency syn- marked variation from alveolus to alveolus made quanti-
drome, P. carinii pneumonia was an unusually subacute tative comparison of organism burden unreliable. Clear-
and subtle illness that often presented with minimal ly, however, the number of organisms did not correlate
cough, almost imperceptible shortness of breath, or low with severity of disease in individual patients. As suggest-
grade fever. The room air arterial oxygen pressure ed by previous studies, some patients with minimal symp-
(RA02) was often above 90 mm Hg, and the chest radio- toms showed large clusters of organisms whereas other
graph frequently showed only faint infiltrates. Lung biop- patients with fulminant life-threatening disease had small
sies ultimately were done either because of symptomatic clusters that were found only after extensive searching of
progression or an aggressive diagnostic approach (15). the histopathologic specimen (19). Other pathogens did
The unusually subtle nature of P. carinii pneumonia in not seem to be major contributors to pulmonary dysfunc-
the acquired immunodeficiency syndrome is supported in tion at the time of lung biopsy. Histopathologic evidence

668 May 1984 • Annals of Internal Medicine • Volume 100 • Number 5

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 for cytomegalovirus infection was seen in about 10% of drome than patients with other immunosuppressive dis-
the lung biopsy samples in both groups, but cytomegalo- eases. Eight of the forty-six patients with the syndrome
virus inclusion cells were sparse in number and difficult had more than one documented episode of Pneumocystis
to find. Cytomegalovirus inclusion cells were relatively pneumonia, versus only 1 of 39 patients with other im-
commoner at autopsy, however, particularly in patients munosuppressive diseases. Because of the frequent use of
with the acquired immunodeficiency syndrome (10 of 15 empiric treatment in the former group, as well as the lack
autopsies) compared to patients with other immunosup- of consistent and long-term follow-up, the true incidence
pressive diseases (4 of 20 autopsies). Rosen and cowork- of recurrent episodes cannot be determined from this
ers (20, 21) also found that cytomegalovirus inclusions study. It is uncertain whether the subsequent episodes
were more readily found in autopsy specimens from pa- represented relapse or reinfection, or whether they could
tients dying of Pneumocystis pneumonia (8 of 20 pa- have been prevented by more prolonged or intensive
tients) than in biopsy specimens (none of 18 patients). treatment.
Even after allowing for differences in quality of tissue A striking finding of this study was the high frequency
examined, this finding suggests that cytomegalovirus may of adverse effects to trimethoprim-sulfamethoxazole in
not be an important factor in contributing to pulmonary patients with the acquired immunodeficiency syndrome.
dysfunction when patients initially present with P. carinii Adverse effects to trimethoprim-sulfamethoxazole oc-
pneumonia, but progressive cytomegalovirus disease may curred in 65% of patients with the syndrome compared
contribute to death regardless of the efficacy of anti- to 12% of patients with other immunosuppressive diseas-
pneumocystis treatment. es (p = 0.0007). Rashes, often quite severe, occurred in
In this retrospective study the choice of anti-pneumo- 29% of patients with the syndrome but not in patients
cystis treatment was not random. For some patients in with other immunosuppressive diseases. Previous reports
each group, trimethoprim-sulfamethoxazole and pentam- show that hypersensitivity reactions to trimethoprim-
idine were effective single agents, in some cases when sulfamethoxazole in patients with other immunosuppres-
given for 10 to 14 days, in some cases when given for sive diseases rarely occur: none of 8 (10), 1 of 11 (16),
prolonged periods. Previous studies have shown that both and 1 of 15 (17) patients had such reactions in three
trimethoprim-sulfamethoxazole and pentamidine are ef- recent series. The reason for the unusually high frequen-
fective drugs for the treatment of P. carinii pneumonia in cy of hypersensitivity reaction to trimethoprim-sulfa-
patients with other immunosuppressive diseases (3, 10, methoxazole in the acquired immunodeficiency syn-
16, 17). In children, Hughes and associates (22) showed drome is unknown; however, this reaction has been seen
that trimethoprim-sulfamethoxazole and pentamidine at other centers (25, 26). Our figures probably underesti-
were equally efficacious for P. carinii pneumonia, though mate the true incidence, because some patients with the
trimethoprim-sulfamethoxazole was less toxic; there has syndrome were treated with pentamidine alone because
been no similar comparative trial in adults. Most patients they had previously shown hypersensitivity to
treated in this study since 1979 received both trimetho- trimethoprim-sulfamethoxazole and thus were purposely
prim-sulfamethoxazole and pentamidine either concur- prevented from receiving trimethoprim-sulfamethoxazole
rently or sequentially, because the initial drug chosen during this study.
caused toxicity or was perceived as ineffective. It could Leukopenia due to trimethoprim-sulfamethoxazole
not be determined if altering treatment did in fact benefit also occurred more frequently in patients with the ac-
patients who responded poorly to the first few days of quired immunodeficiency syndrome (14 of 30) than pa-
their initially chosen therapeutic regimen. tients with other immunosuppressive diseases (1 of 12)
Survival regardless of treatment was compared for the (p = 0.038). In patients with other immunosuppressive
two groups. There was no significant difference in rapidi- diseases the absence of leukopenia secondary to trimetho-
ty of improvement, though about a third of survivors prim-sulfamethoxazole during treatment of P. carinii
with the acquired immunodeficiency syndrome had pro- pneumonia is supported by other reports (10, 16, 17), in
longed low grade fever. Time to defervesence and initial which only 1 of 34 patients in three series developed leu-
radiographic improvement were similar to previous re- kopenia. The reason for the unusually high frequency of
ports for both groups (10-14, 16, 17). leukopenia among patients with the syndrome is unclear.
Overall survival rates were similar for patients with the Bone marrow biopsies in several patients failed to show a
acquired immunodeficiency syndrome and patients with cause, and patients generally did not respond to empiric
other immunosuppressive diseases, being 57% and 50%, folinic acid. Other drugs could not be implicated as caus-
respectively, when comparable years (1979 to 1983) were es of this phenomenon.
assessed. These survival rates are similar to rates of 42% Trimethoprim-sulfamethoxazole therapy had to be dis-
to 72% reported for adults with other immunosuppres- continued because of adverse effects in patients with the
sive diseases (3, 10, 16, 17), 56% reported by the CDC acquired immunodeficiency syndrome more frequently
for patients with the acquired immunodeficiency syn- than in patients with other immunosuppressive diseases
drome (23), and 68% compiled from a series of reports (7 of 34 versus none of 17). In contrast, although pen-
of P. carinii pneumonia in patients with the syndrome tamidine was often associated with adverse reactions, it
(11-14, 24). From limited data in this series and from was rarely discontinued, perhaps because it was consid-
CDC data it appears that subsequent episodes of P. cari- ered the last therapeutic possibility. Thus, for patients
nii pneumonia were commoner in patients with the syn- with the syndrome, trimethoprim-sulfamethoxazole was
Kovacs at al. • Pneumocystis carinii Pneumonia 669

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 not as safe as for patients with other immunosuppressive vent P. carinii pneumonia.
diseases. Toxicity of trimethoprim-sulfamethoxazole ACKNOWLEDGMENTS: The authors thank Dr. Robert Wesley for his
needs to be an important consideration in the choice of assistance in the statistical analysis.
treatment of P. carinii pneumonia in patients with the • Requests for reprints should be addressed to Joseph A. Kovacs, M.D.;
syndrome. Critical Care Medicine Department, Room 10D-48, Clinical Center, Nation-
al Institutes of Health; Bethesda, M D 20205.
Our data suggest that in patients with the syndrome, P.
carinii pneumonia is usually a more subtle, slowly pro- References
gressive pneumonia than the disease commonly recog- 1. CENTERS FOR DISEASE CONTROL. Acquired immunodeficiency syn-
drome (AIDS) update—United States. MMWR. 1983;32:309-11.
nized in patients with other immunosuppressive diseases. 2. D E T E L S R. Epidemiologic perspectives, pp 214-6. In G O T T L I E B MS,
Analysis of individual case histories shows that there is a moderator. The acquired immunodeficiency syndrome. Ann Intern
spectrum of disease presentations among patients with Med. 1983;99:208-20.
3. W A L Z E R P D , P E R L D P , K R O G S T A D DJ, R A W S O N PG, S C H U L T Z M G .
the syndrome. Most commonly, patients have a very mild Pneumocystis carinii pneumonia in the United States: epidemiologic,
cough or dyspnea for many weeks or months. Diagnostic diagnostic, and clinical features. Ann Intern Med. 1974;80:83-93.
4. H U G H E S WT, K U H N S, C H A U D H A R Y S, et al. Successful chemoprophy-
procedures are finally done when patients are only slight- laxis for Pneumocystis carinii pneumonitis. N Engl J Med.
ly hypoxemic and findings on chest radiographs are nor- 1977;297:1419-23.
mal or strikingly abnormal. Some of these patients re- 5. CENTERS FOR DISEASE CONTROL. Update on acquired immunodefici-
ency syndrome (AIDS)—United States. MMWR. 1982;31:507-8, 513-4.
spond rapidly to treatment and are persistently symptom 6. LAPINSKI M. Collection, processing, and measurement of blood gases.
free after a 10 to 14 day course of trimethoprim-sulfa- In: H E N R Y JB, ed. Clinical Diagnosis and Management by Laboratory
Methods. 16th edition. Philadelphia: W.B. Saunders Co.; 1979:119.
methoxazole or pentamidine; some have persistent symp- 7. A R M I T A G E P. Statistical Methods in Medical Research. New York:
toms after 10 to 14 days of treatment, have persistent John Wiley and Sons; 1971:135-8.
Pneumocystis cysts at a second bronchoscopic examina- 8. B R O W N B, HOLLANDER B. Statistics: A Biomedical Introduction. New
York: John Wiley and Sons; 1971:328-32.
tion, but gradually resolve their symptoms after 4 to 6 9. M A C H E R AM, K O V A C S JA, G I L L V, et al. Bacteremia due to Mycobac-
weeks of treatment; some deteriorate progressively and terium avium-intracellulare in the acquired immune deficiency syn-
die after many weeks of treatment, and are frequently drome. Ann Intern Med. 1983;99:782-5.
10. L A U WK, YOUNG LS. Trimethoprim-sulfamethoxazole treatment of
found to have extensive cytomegalovirus pneumonitis at Pneumocystis carinii pneumonia in adults. N Engl J Med. 1976;295:716-
autopsy, with variable intensity of Pneumocystis infec- 8.
11. F O L L A N S B E E SE, BUSCH D F , WOFSKY CB, et al. An outbreak of Pneu-
tion. There are patients with the syndrome that present mocystis carinii pneumonia in homosexual men. Ann Intern Med.
with rapidly progressive fulminant disease. These pa- 1982;96(6Pt 1):705-13.
tients are usually very hypoxemic at time of diagnosis; 12. M A S U R H, M I C H E L I S MA, G R E E N E JB, et al. An outbreak of communi-
ty-acquired Pneumocystis carinii pneumonia: initial manifestation of
response to treatment differs. cellular immune dysfunction. N Engl J Med. 1981;305:1431-8.
The duration of symptoms did not correlate signifi- 13. M I L D V A N D, M A T H U R U, E N L O W RW, et al. Opportunistic infections
cantly with prognosis; however, patients with the syn- and immune deficiency in homosexual men. Ann Intern Med. 1982;96(6
Pt l):700-4.
drome who were in less respiratory distress at the time of 14. G O T T L I E B MS, SCHROFF R, S C H A N K E R HM, et al. Pneumocystis carinii
presentation, as measured by respiratory rate, RA02, and pneumonia and mucosal candidiasis in previously healthy homosexual
men: evidence of a new acquired cellular immunodeficiency. N Engl J
alveolar-arterial gradient (Table 5) were more likely to Med. 1981;305:1425-31.
survive. The lymphocyte count was also significantly 15. L E V I N M, M C L E O D R, Y O U N G Q, et al. Pneumocystis pneumonia: im-
higher in survivors, suggesting that the degree of immu- portance of gallium scan for early diagnosis and description of a new
immunoperoxidase technique to demonstrate Pneumocystis carinii. Am
nosuppression was probably an important factor. Similar Rev Respir Dis. 1983;128:182-5.
correlations of hypoxia, leukopenia, and survival have 16. W I N S T O N DJ, L A U WK, G A L E R P , Y O U N G LS. Trimethoprim-sulfam-
been made previously for patients with other immuno- ethoxazole for the treatment of Pneumocystis carinii pneumonia. Ann
Intern Med. 1980;92:762-9.
suppressive diseases by Walzer and associates ( 3 ) . Serum 17. SATTLER FR, REMINGTON JS. Intravenous trimethoprim-sulfamethox-
albumin was higher in survivors with the acquired immu- azole therapy for Pneumocystis carinii pneumonia. Am J Med.
nodeficiency syndrome in the present study, supporting 1981;70:1215-21.
18. G A M S U G, H E C H T ST, BIRNBERG FA, C O L E M A N D L , G O L D E N JA.
the observations linking nutritional status and P. carinii Pneumocystic carinii pneumonia in homosexual men. AJR.
pneumonia that have previously been made in both mu- 1982;139:647-51.
rine and human P. carinii pneumonia (27). 19. BURKE BA, G O O D RA. Pneumocystis carinii infection. Medicine.
1973;52:23-51.
Why P. carinii pneumonia usually presents in patients 20. ROSEN P, ARMSTRONG D, R A M O S C. Pneumocystis carinii pneumonia:
with the acquired immunodeficiency syndrome as a more a clinicopathologic study of twenty patients with neoplastic diseases.
Am J Med. 1972;53:428-36.
subtle, slowly progressive pneumonia, and why adverse 21. ROSEN PP, M A R T I N I N, ARMSTRONG D. Pneumocystis carinii pneumo-
reactions to trimethoprim-sulfamethoxazole therapy are nia: diagnosis by lung biopsy. Am J Med. 1975;58:794-802.
22. HUGHES WT, F E L D M A N S, C H A U D H A R Y SC, OSSI M G , Cox F,
so common remain to be explained. It seems unlikely that SANYAL SK. Comparison of pentamidine isethionate with trimetho-
the behavior of the pneumonia is related to a different prim-sulfamethoxazole in the treatment of Pneumocystis carinii pneu-
strain of Pneumocystis because both populations at risk monia. / Pediatr. 1978;92:285-92.
23. HAVERKOS HW. Assessment of Pneumocystis carinii pneumonia
interact in the same environment. More likely, some facet (PCP) therapy: a cooperative study. In: Twenty-Second Interscience
of the immunoregulatory defect of the acquired immuno- Conference on Antimicrobial Agents. Washington, D.C.: American So-
deficiency syndrome permits a different host-parasite re- ciety for Microbiology; 1982: abstract 852.
24. M A S U R H, M I C H E L I S MA, W O R M S E R G, et al. Opportunistic infection
lationship from that usually seen in patients treated with in previously healthy women: initial manifestation of a community-ac-
corticosteroids or cytotoxic agents. Further examination quired cellular immunodeficiency. Ann Intern Med. 1982;97:533-8.
25. M I T S U Y A S U R, G R O O P M A N J, V O L B E R D I N G P. Cutaneous reaction to
of differences in immune response between the two pa-
trimethoprim-sulfamethoxazole in patients with AIDS and Kaposi's sar-
tient groups should help to define host factors that pre- coma [Letter]. N Engl J Med. 1983;308:1535-6.

670 May 1984 • Annals of Internal Medicine • Volume 100 • Number 5

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017

 26. J A F F E MS, A B R A M S DI, A M M A N N AJ, L E W I S BJ, G O L D E N JA. Com- 27. H U G H E S WT, P R I C E RA, SISKO F, et al. Protein-calorie malnutrition: a
plications of co-trimoxazole in treatment of AIDS-associated Pneumo- host determinant for Pneumocystis carinii infection. Am J Dis Child.
cystis carinii pneumonia in homosexual men. Lancet. 1983;2:1109-11. 1974;128:44-52.

Annals of Internal Medicine. 1984;100:671-676. © 1 9 8 4 American College of Physicians 671

Downloaded From: http://annals.org/ by a University of California San Diego User on 01/02/2017