Clinical Infectious Diseases

MAJOR ARTICLE

Identification of Predictive Markers and Outcomes of

Late-onset Pneumocystis jirovecii Pneumonia in Kidney
Transplant Recipients

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

Hannah Kaminski, 1,2 Julie Belliere,3,4 Laure Burguet,1 Arnaud Del Bello,3 Benjamin Taton,1,5 Stéphane Poirot-Mazères,1 Isabelle Accoceberry,6
Laurence Delhaes,6 Jonathan Visentin,2,7 Marco Gregori,1 Xavier Iriart,8,9 Elena Charpentier,8,9 Lionel Couzi,1,2 Nassim Kamar,3,4,9 and Pierre Merville1,2
1
Department of Nephrology, Transplantation, Dialysis and Apheresis, Pellegrin University Hospital, Bordeaux, France, 2Centre National de Recherche Scientifique- Unité Mixte de Recherche
5164 ImmunoConcEpT, Bordeaux University, Bordeaux, France, 3Department of Nephrology and Organ Transplantation, Centre Hospitalier Universitaire Toulouse, Toulouse, France, 4Paul Sabatier
University, Toulouse, France, 5Mathematics Modeling for Oncology, Institute of Bordeaux Mathematics, Institut National de Recherche en Informatique et en automatique-Unité Mixte de Recherche
5251, Talence, France, 6Laboratory of Parasitology-Mycology, Pellegrin University Hospital, Bordeaux, France, 7Laboratory of Immunology and Immunogenetics, Pellegrin University Hospital,
Bordeaux, France, 8Department of Parasitology-Mycology, Toulouse University Hospital Toulouse, France, and 9Institut national de la santé et de la recherche médicale U1043, Institut Fédératif de
Recherche Bio-Médicale de Toulouse, Toulouse, France

(See the Editorial Commentary by Cervera et al on pages e1464–6.)

Background.  In the era of prophylaxis, Pneumocystis pneumonia (PCP) has become a late-onset opportunistic infection re-
quiring indications for prolonged prophylaxis to be defined. The primary objective of our study was therefore to evaluate risk factors
associated with late-onset PCP. The secondary objective was to assess the impact of this infection on graft and patient survival.
Methods.  We conducted a French case-control study in Bordeaux and Toulouse center by matching 1 case to 1–2 controls from
the same center based on the transplant date and the type of induction treatment.
Results.  Seventy cases and 134 controls were included. PCP occurred at a median of 3 years after transplantation. The total
lymphocyte count and CD4+ and CD8+ T-lymphocyte values were lower in the cases than in their matched controls on the day of
infection and annually up to 4 years earlier. The covariables independently associated with PCP were the total lymphocyte count
1 year before Pneumocystis, mTOR inhibitors used as maintenance immunosuppressive drugs, and the administration of cortico-
steroid boluses used in acute rejection. A total lymphocyte count threshold <1000/µL offered the best predictive value for infection
occurrence. PCP was associated with high incidence of graft loss and patient death (30% and 17% respectively, 3 years after PCP).
Conclusions.  Pneumocystis pneumonia has dramatic consequences in kidney transplant recipients; a targeted prophylaxis based
on simple criteria, such as chronic lymphopenia and/or history of corticosteroid boluses, could be useful to avoid life-threatening
complications.
Keywords.  Pneumocystis pneumonia; lymphopenia; corticosteroid boluses; kidney transplantation.

With the widespread use of anti–Pneumocystis jirovecii prophy- strategies for handling immunocompromised inpatients or out-
laxis for all solid organ transplant (SOT) recipients for at least patients with suspected or proven PCP have been elaborated
6–12 months posttransplantation following international guide- [5]. PCP incidence in SOT recipients ranges now from 0.3%
lines [1], the epidemiology of Pneumocystis pneumonia (PCP) to 2.5%. Moreover, PCP still leads to a high mortality rate of
has changed over the last decades. Nowadays this fungal infec- 14% in kidney transplant recipients as reported in the studies
tion occurs later, with an increased incidence during the second published after 2010 [2]. Trimethoprim-sulfamethoxazole
year posttransplantation [2], often during nosocomial outbreaks (TMP-SMX) is the drug of choice for PCP initial prophylaxis
in the hospital environment [3]. Indeed, interindividual trans- and is very efficient. However, it has become critical to define
missions have been well documented [4], and some practical conditions requiring an extension or a reinitiation of prophy-
laxis in order to avoid this life-threatening infection since re-
commendations are currently lacking on this issue. The overall
low incidence of PCP discourages a lifelong prophylaxis in all
kidney transplant recipients, yet a personalized approach based
 

Received 28 April 2020; editorial decision 23 September 2020; published online 23 October
2020. on individual risks could be an acceptable alternative.
Correspondence: P.  Merville, Department of Nephrology, Transplantation, Dialysis and
Apheresis, Pellegrin University Hospital, Place Amélie Raba Léon, 33076 Bordeaux, France
In this retrospective case-control study, we described a co-
([email protected]). hort of 70 PCP and 134 matched-control kidney transplant
Clinical Infectious Diseases®  2021;73(7):e1456–63 recipients. We analyzed the risk factors and the outcomes asso-
© The Author(s) 2020. Published by Oxford University Press for the Infectious Diseases Society
of America. All rights reserved. For permissions, e-mail: [email protected].
ciated with this opportunistic infection to individualize param-
DOI: 10.1093/cid/ciaa1611 eters that could guide long-term prophylaxis.

e1456 • cid 2021:73 (1 October) • Kaminski et al

PATIENTS AND METHODS (Beckman Coulter) and were acquired on a FC500 cytometer
Study Design and Patients
(Beckman Coulter).
We performed a case-control study of posttransplant PCP in At Toulouse University hospital, whole blood lymphocytes
kidney transplant recipients between 2004 and 2015 in Toulouse were acquired on an automatic AQUIOS cytometer (Beckman
University Hospital and between 2007 and 2018 in Bordeaux Coulter). CD3/CD4/CD8 immunophenotyping was performed
University Hospital. Patients with pneumonia, compatible ra- as previously described [9]. All antibodies and other reagents
diographic findings on chest computed tomographic scan, and used in this study were from BD Biosciences (BD, Le Pont De
positive microbiological assessment from bronchoalveolar la- Claix, France).
vage (BAL) or sputum were considered as “cases” of PCP ac-

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

Immunosuppressive Regimen
cording to the latest international guidelines [1].
Immunosuppressive regimens were similar in both centers.
In Toulouse University hospital, diagnosis of PCP was per-
Induction treatment used rATG (Thymoglobulin, Sanofi-
formed on BAL by direct microscopic examination after
Aventis) in human leukocyte antigen (HLA)–sensitized pa-
staining to detect P.  jirovecii cysts and/or trophozoites (indi-
tients: 1.5  mg/kg of body weight at the day of the graft and
rect immunofluorescence assay [Monofluokit Pneumocystis,
then 1 mg/kg at days 1–4 posttransplant. Non-HLA-sensitized
Bio-Rad, Marnes-la-Coquette, France] and/or May-Grünwald-
patients received anti-IL2RA (SIMULECT, Novartis Pharma
Giemsa staining) and/or by quantitative polymerase chain re-
SAS), 20 mg at day 0 and day 4 posttransplant. Maintenance im-
action (qPCR) for P. jirovecii DNA detection [6]. In Bordeaux
munosuppressive regimen was mainly based on calcineurin in-
University hospital, the respiratory specimens (sputum,
hibitors, either tacrolimus (trough level 8–12 ng/mL for the first
tracheobronchial aspiration, and/or BAL) were submitted to
3  months, then 5–10  ng/mL) or cyclosporine A  (trough level
direct examination (Gomori-Grocott and May-Grünwald-
150–200  ng/mL for the first 3  months, then 75–125  ng/mL),
Giemsa staining) and to qPCR (LightMix kit P.  jirovecii TIB
mainly associated with mycophenolic acid (720 mg twice a day).
MOLBIOL, Berlin, Germany) [7].
Some patients received mammalian target of rapamycin inhibi-
The reference day of PCP for the cases was defined as the day
tors (mTORi) with trough level targets between 5 and 8 ng/mL.
when microbiological evidence of PCP was obtained (day 0) and
Five hundred milligrams to 1 g of intravenous corticosteroids
for the control patients as the matched day from transplantation
for borderline acute rejection or T-cell and antibody-mediated
date. Early-onset PCP and late-onset PCP were defined as an
rejection, respectively, was used both during 3 days.
infection diagnosed ≤365 and >365 days posttransplant, respec-
tively [8]. Each case was matched with 1 or 2 control patients.
Control patients were identified at the same medical center as Clinical Definitions

the respective case and were defined as such for showing no Cytomegalovirus (CMV) infection was defined as a positive
clinical, radiographic, or microbiological signs of PCP during whole blood CMV quantitative nucleic acid testing [10], con-
the entire follow-up. For each case, controls were selected sistent with the American Society of Transplantation and the
among consecutive patients who had a functional graft on day CMV Drug Development Forum recommendations [11]. “All
0 of PCP and paired by (1) the type of induction treatment, that CMV events” were defined as occurring either before or after
is, rabbit antithymocyte globulin (rATG) or anti–interleukin 2 day 0.  Allograft rejection was biopsy-proven. We defined al-
receptor antibody (anti-IL2RA); (2) and by transplantation date lograft failure as requirement to return to permanent dial-
(± 3  months). No patient presented a concomitant infection ysis. Glomerular filtration rate (GFR) was calculated with the
with human immunodeficiency virus (HIV). Cases and con- Modification of Diet in Renal Disease (MDRD) formula.
trols were followed up from the time of PCP (day 0)  up to a All patients were similarly treated in each center with a
maximum of 3 years postinfection or death/graft failure. 6-month posttransplantation prophylaxis, either with oral
Data were collected for both cases and controls thanks to TMP-SMX or aerosolized pentamidine.
the local medical software for which all patients had signed a
Statistical Analyses
written informed consent. Both centers obtained local institu- 2
tional ethics board approval. Mann-Whitney and χ  tests were used when appropriate.
Alternatively, Fisher exact test was used for a low number of pa-
Immunophenotyping tients. A  P value <.05 was considered statistically significant.
Each kidney transplant recipient had yearly peripheral blood Comparisons of lymphocyte counts were performed at day 0
lymphocyte immunophenotyping at annual systematic visits. and every year, up to 4 years. For each date, comparison was per-
At Bordeaux University hospital, whole blood lympho- formed only with available values for both cases and their paired
cytes were stained with CD45, CD3, CD4 and CD8 antibodies controls. Assessment of PCP risk with receiver operating charac-
(Beckman Coulter, Marseille, France) and mixed with Flow- teristic (ROC) curve analysis as well as multivariable analysis was
Count Fluorospheres for the absolute count determination performed with total lymphocyte counts and CD4+ and CD8+

Risks/Outcomes of Pneumocystis Pneumonia  •  cid 2021:73 (1 October) • e1457

T-lymphocytes 1 year before day 0. Comparison of GFR was per- after transplantation (1st-d quartiles: 1.5–8.6). After acute rejec-
formed 6 months before day 0. Univariate logistic regression anal- tion episodes, 3 periods could be identified (Figure 1B): no cases
ysis of PCP risk factors was first performed, then covariables with of PCP in the first 6 months, then an incidence of 1.25/month
a P value <.25 were included in a multivariable logistic regression from 6 to 18 months and finally a low rate of 0.14/month from
analysis. Results were expressed as odds ratios (ORs) with 95% 1.5 to 6.9 years. Baseline characteristics of patients are described
confidence intervals (95% CIs). To take into account the competing in Table 1. Cases were older than controls, and were comparable
risks of death and graft loss, we conducted cumulative incidence in terms of sex ratio, transplantation rank, initial renal disease,
analysis using Nelson-Aalen estimator [12]. GraphPad Prism soft- and CMV events before day 0 of PCP. Cases underwent more
ware (version 6.0, GraphPad Software, San Diego, California) and acute rejection episodes. Postrejection secondary prophylaxis

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

RStudio statistical software (version 1.1.423) were used. was given for 3–6 months in 62.5% of cases and 100% of controls.
The maintenance immunosuppressive regimen was not different
RESULTS except for mTORi (22.8% of cases vs 8% of controls) and cortico-
steroids (82.8 vs 66.4%). We therefore included these covariables
Epidemiological and Clinical Characteristics of Patients
in the analysis of PCP risk factors (see below).
Seventy cases of PCP were identified in both centers and matched
with 134 controls. The incidence rate of PCP was 1.6 per 1000
patient-years in Toulouse center and 3.5 in Bordeaux center (2.34 Kinetics of Lymphocytes Differ Between Cases and Controls
per 1000 patient-years pooling both centers). After transplanta- We analyzed the evolution of total lymphocyte counts and CD4+
tion, 5 different periods of PCP onset were described (Figure 1A): and CD8+ T-lymphocyte counts (Figure 2A–C). Cases displayed
no PCP during the prophylaxis period, 3 PCP cases shortly after lower total lymphocyte counts for the 4 years preceding PCP ex-
prophylaxis discontinuation, a high incidence of 1.66/month be- cept year 3 before PCP (Supplementary Table 1). Strikingly, the
tween 1 and 2 years, then a moderate incidence of 0.59/month be- difference was higher 1 year before PCP with a mean ± standard
tween 2 and 5.5 years, and finally a low incidence of 0.13/month deviation (SD) of 702 ± 239 in cases vs 1472 ± 661 for controls
between 5.5 and 19.8 years. PCP occurred at a median of 3 years (P < .001). From day 0 up to 4 years before PCP, CD4+ and CD8+

A
100
Patients without PCP (%)

80

60

40

20

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Time after transplantation (years)

B
100
Patients without PCP (%)

80

60

40

20

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Time after rejection (years)

Figure 1.  Development of Pneumocystis pneumonia (PCP) in transplant recipients after transplantation and acute rejection, over time. A, Proportion of transplant recipients
without PCP after transplantation over time (years). B, Proportion of transplant recipients without PCP after acute rejection over time (years). Dotted lines represent the limits
between 2 periods with different incidences of PCP.

e1458 • cid 2021:73 (1 October) • Kaminski et al

Table 1.  Characteristics of Cases and Controls also higher 1 year before with a mean ± SD of PCP 377 ± 258 vs
589 ± 337 (P < .001) and 286 ± 184 vs 496 ± 306 (P < .001) for
Controls
Characteristic Cases (n = 70) (n = 134)
CD4+ and CD8+ T-lymphocyte counts, respectively.
Age, y, mean ± SD 54.8 ± 13.9 50.9 ± 14.3
Male sex 45 (64.3) 88 (65.7)
Risk Factors Associated With Late-onset PCP
Tx rank In the univariate analysis, recipient age, mTORi, azathioprine,
 1 56 (80) 117 (87.3) and corticosteroids were associated with a higher risk of PCP,
 >1 14 (20) 17 (12.7) whereas mycophenolic acid use was associated with a lower
Best GFR before PCP, mL/min/m2, 70.1 ± 27.8 74.1 ± 27.6
risk of PCP (Table 2). Following acute rejection episodes, both
mean ± SD

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

GFR 6 mo before day 0 of PCP, 39.83 ± 13.44 45.66 ± 17.04 rituximab, plasma exchanges, and corticosteroid boluses were
mL/min/m2, mean ± SD associated with a higher risk of PCP. Total lymphocyte counts
Origin of renal disease and CD4+ and CD8+ T-lymphocyte counts from day 0 of PCP
 Vascular 3 (4.3) 7 (5.2)
and up to 4 years before were associated with PCP. Finally, GFR
 Tubulo-interstitial 5 (7.1) 19 (14.2)
 Glomerular 24 (34.3) 58 (43.3)
6 months before day 0 was also associated with PCP. We per-
 Diabetes 5 (7.1) 8 (5.9) formed a first multivariable analysis including acute rejection
 Malformative 6 (8.5) 7 (5.2) treatment, in which the only covariables independently associ-
 ADPKD 13 (18.6) 20 (14.9) ated with PCP were total lymphocyte count, corticosteroid bo-
 Other 2 (2.8) 3 (2.2) luses, age, and mTORi (Table  2). We next performed another
 Undetermined 12 (17.1) 12 (8.9)
multivariable analysis including acute rejection (Supplementary
CMV infection before day 0 25 (35.7) 41 (30.6)
TMP-SMX prophylaxis post-Tx 68 (97) 134 (100)
Methods) in which lymphocyte counts, age, mTORi, and
Pentamidine for prophylaxis 5 (7.3) 14 (10.4) acute rejection were the independently associated covariables
post-Tx (Supplementary Table 2). Altogether, this suggests that corti-
TMP-SMX/pentamidine prophy- 15 (62.5) 12 (100)
costeroid boluses, given either during antibody-mediated or
laxis after acute rejection
Maintenance immunosuppressive T-cell–mediated acute rejection, increased PCP risk.
regimen
  Cyclosporine A 14 (20) 32/125a (25.6)
 Tacrolimus 39 (55.7) 82/125a (61.2) Lymphocyte Count Predictive Values of PCP Onset
  mTOR inhibitor 16 (22.8) 10/125a (8) We sought to define lymphocyte counts thresholds that would
 Azathioprine 9 (12.8) 6/125a (4.5) allow us to identify patients at risk of late-onset PCP; therefore
  Mycophenolic acid 45 (64.3) 100/125a (80) the 3 patients who presented early-onset PCP were excluded
 Belatacept 3 (4.3) 4/125a (3.2)
from this analysis. We first performed a ROC curve anal-
 Corticosteroids 58 (82.8) 83/125a (66.4)
Acute rejection before day 0 of 24 (34.3) 13 (9.7)
ysis of total lymphocytes and CD4+ and CD8+ T lymphocytes
PCP 1  year before PCP. The best area under the curve (AUC) was
ABMR 11 (15.7) 3 (2.2) obtained for total lymphocyte counts (0.75 [95% CI, .67–.82];
TCMR 13 (18.5) 10 (7.5) Supplementary Figure 1A). For CD4+ T lymphocytes, the AUC
Treatment for acute rejection
was 0.68 (95% CI, .60–.76]; Supplementary Figure 1B) and
 rATG 0 4 (3)
 Rituximab 9 (12.8) 3 (2.2)
for CD8+ T lymphocytes, the AUC was 0.72 (95% CI, .65–.80;
  Plasma exchange 8 (11.4) 3 (2.2) Supplementary Figure 1C).
  Corticosteroid boluses 22 (31.4) 12 (8.9) The optimal predictive threshold of total lymphocyte counts
Time from Tx to PCP, y, median 3 (1.5–8.6) NA for PCP occurrence was 1000/µL, with a negative predictive
(1st-d quartiles)
value of 81.6%, a positive predictive value of 60.1%, sensi-
Time from rejection to PCP, mo, 16.5 (11.5–27.2) NA
median (1st-d quartiles) tivity of 71.4%, and specificity of 73.4%. Consequently, we
Data are presented as no. (%) unless otherwise indicated. also performed the multivariable analysis using lymphocyte
Abbreviations: ABMR, antibody-mediated rejection; ADPKD, autosomal dominant poly- counts as a discrete covariable (<1000 or >1000/µL). mTORi,
cystic kidney disease; CMV, cytomegalovirus; GFR, glomerular filtration rate (Modification
of Diet in Renal Disease formula); NA, not applicable; PCP, Pneumocystis pneumonia; rATG, age, lymphocyte counts, and corticosteroid boluses were
rabbit antithymocyte globulins; mTOR, mammalian target of rapamycin; SD, standard de-
viation; TCMR, T-cell mediated rejection; TMP-SMX, trimethoprim-sulfamethoxazole; Tx,
still the covariables associated with an increased risk of PCP
transplantation.
a
(Supplementary Table 3). Using the total lymphocyte count
Data for maintenance immunosuppressive regimen were available only for 125 controls.
<1000/µL and/or corticosteroid boluses (the most associated
criteria), we calculated that restarting prophylaxis on those
criteria would lead to a 88.9% of relative risk reduction and
T-lymphocyte counts were lower in PCP patients than in their a number needed to treat of 142 (Supplementary Table 4). As
paired controls (Supplementary Table 1). The difference was opposed to lymphocyte count, no threshold was found with

Risks/Outcomes of Pneumocystis Pneumonia  •  cid 2021:73 (1 October) • e1459

 A. Total lymphocytes B. CD4+ T lymphoctes C. CD8+ T lymphocytes Cases
Controls
2000 ns 900 700 ***
** *** *** ** ** *** * ***
absolute count (/µL) * * * *

absolute count (/µL)

absolute count (/µL)

750 600
Total lymphocytes

1500 *** 500

CD4 T cells

CD8 T cells
600
400
1000 450
300
300
500 200
150 100
0 0 0

-4

-3

-2

-1

-4

-3

-2

-1

0
-4

-3

-2

-1

0
Years before pneumocystis Years before pneumocystis Years before pneumocystis

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

Cases 24 34 46 63 65 Cases 24 34 46 63 65 Cases 24 34 46 63 65

Controls 39 52 66 111 111 Controls 39 52 66 111 111 Controls 39 52 66 111 111

Figure 2.  Comparison of lymphocyte evolution over time from 4 years before to day 0 of Pneumocystis jirovecii pneumonia between cases and controls. A, Total lymphocyte
counts were compared between cases and controls every year from 4 years before to day 0 of Pneumocystis pneumonia (PCP). Comparisons were performed with Mann-
Whitney test for each time-point and only when data were available for both cases and their paired controls. B, CD4+ T-lymphocyte counts were compared between cases
and controls every year from 4 years before to day 0 of PCP. Comparisons were performed with Mann-Whitney test for each time-point and only when data were available for
both cases and their paired controls. C, CD8+ T-lymphocyte counts were compared between cases and controls every year from 4 years before to day 0 of PCP. Comparisons
were performed with Mann-Whitney test for each time-point and only when data were available for both cases and their paired controls. All panels: *P < .05 to < .01; **P <
.01 to < .001; ***P < .001. Abbreviation: ns, not significant.

the covariable “age,” as indicated by the poor value of the ROC or until 7  years following acute rejection. This reinforced the
curve analysis (AUC, 0.63 [95% CI, .65–.71]) (Supplementary need to identify easy-to-use parameters that can identify the pa-
Figure 2). tients at risk of late-onset PCP needing targeted prophylaxis. We
identified age, mTORi use, corticosteroid boluses, and chronic
Outcome of PCP on Patient and Graft Survival lymphopenia as independent risk factors associated with PCP.
PCP was associated with high incidence of graft loss and patient Age is often identified as a risk factor, with various thresholds
death (Figure  3A and 3B) in the 3-year cumulative incidence [13, 15], which probably reflects the mean age of the considered
curves. Indeed, 3-year graft loss was about 30% and 3-year cohort. However, in our cohort, no pertinent age threshold was
death was about 17%. isolated. As previously described, we also identified mTORi use
as an independent risk factor of PCP [16, 17]. This significant
DISCUSSION
positive association with mTORi could be either linked to their
In this study conducted in 2 French centers and including 70 immunosuppressive effect [18] or to the rapamycin-lung syn-
consecutive kidney transplant recipients with PCP, we analyzed drome provoked by PCP [19]. We did not isolate CMV infection
risk factors associated with PCP occurrence and its prognosis as a risk factor. Indeed, it has been recently described as a marker
with the aim to assess a strategy to target patients with high risk of severity of PCP rather than a risk factor [20, 21]. In our study,
of developing late-onset PCP and thereby to guide the prescrip- corticosteroid boluses were associated with a higher risk of PCP,
tion of PCP chemoprophylaxis toward a more individualized as previously described in non-HIV patients, transplanted or not
approach. The main finding of this study was that the 2 param- [2]. In our programs, we followed current recommendations to
eters that were the most associated with an increased risk of resume prophylaxis for 3–6 months [1], but we have shown that
PCP were a total lymphocyte count <1000/µL at annual system- the at-risk period can be longer.
atic visit and/or corticosteroid boluses, and these were found in Another important finding of this study was the independent
88.9% of our cases. Those criteria could be retained to extend association between chronic lymphopenia, for at least 4  years
or resume long-term prophylaxis. In addition, PCP in kidney before infection, and PCP onset. Lymphopenia has been de-
transplant recipients had a very poor prognosis and was associ- scribed in other studies in the days preceding infection only,
ated with high incidence of graft loss and patient death. and therefore this parameter cannot be used as a predictive
A 3- to 6-month routine prophylaxis for P. jirovecii has been marker to indicate prophylaxis after the first 6 months [13, 22].
shown to very efficiently prevent PCP either following trans- In the absence of CD4+ T cells, as in HIV infection, CD8
plantation or after the treatment of an acute rejection, as re- T cells’ influx into the lung plays an important role to resolve
commended in the current guidelines [1]. However, the time of PCP, notably through their interferon-γ secretion. Indeed, their
higher-risk period for infection has shifted within the 1- to 2-year depletion in the CD4-depleted mouse model of PCP exacer-
postprophylaxis period [13, 14]. We observed a peak occurring (1) bates infection [23], while generating a global lymphopenia,
between 12 and 24 months after transplantation, and (2) between as observed in our patients. Interestingly, in a model of
6 and 18 months after acute rejection treatment, but some occur- dexamethasone-treated Wistar rats, it was demonstrated that
rences remained even later, until 20  years after transplantation recovery from PCP was associated with the withdrawal of

e1460 • cid 2021:73 (1 October) • Kaminski et al

Table 2.  Univariable and Multivariable Analyses of Risk Factors for the Development of Pneumocystis Pneumonia in Transplant Recipients

Univariate Analysis Multivariable Analysis

Covariables OR 95% CI P Value OR 95% CI P Value

Age, y 1.031 1.009–1.055 .006 1.033 1.006–1.064 .019

Sex, female vs male 0.94 .515–1.735 .843 … …
Tx rank >1 vs 1 1.720 .782–3.737 .170 … …
CMV before day 0 1.260 .679–2.317 .459 … …
Immunosuppressive regimen
  Cyclosporine A 0.726 .348–1.455 .378 … …

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

 Tacrolimus 0.675 .371–1.228 .197 … …
 mTORi 3.437 1.483–8.319 .004 3.230 1.111–9.938 .033
 Azathioprine 2.950 1.017–9.162 .049 … …
  Mycophenolic acid 0.468 .242–.898 .022 … …
 Corticosteroids 2.50 1.246–5.341 .012 … …
Acute rejection episodes 4.816 2.296–10.512 <.001 … …
Antibody-mediated rejection 3.456 2.381–9.435 .002 … …
T-cell–mediated rejection 2.547 1.039–6.372 .041 … …
Acute rejection treatment
 Rituximab 6.295 1.805–29.131 .0072 … …
  Plasma exchange 5.505 1.533–25.795 .014 … …
  Corticosteroid boluses 4.545 2.118–10.180 <.001 4.131 1.711–10.503 .002
CD4 count at day 0 0.995 .994–.997 <0001
  1 y before day 0 0.997 .996–.998 <.001 … …
  2 y before day 0 0.998 .997–.999 .030 … …
  3 y before day 0 0.997 .995–.999 .009 … …
  4 y before day 0 0.996 .993–.998 .009 … …
CD8 count at day 0 0.996 .994–.997 <.001
  1 y before day 0 0.996 .994–.997 <.001 … …
  2 y before day 0 0.998 .997–.999 .048 … …
  3 y before day 0 0.997 .995–.999 .038 … …
  4 y before day 0 0.997 .995–.999 .051 … …
Total lymphocyte count at day 0 0.997 .996–.998 <.001
  1 y before day 0 0.998 .997–.999 <.001 0.997 .996–.999 <.001
  2 y before day 0 0.998 .997–.999 .013 … …
  3 y before day 0 0.998 .997–.999 .038 … …
  4 y before day 0 0.998 .997–.999 .015 … …
GFR 6 mo before day 0 0.968 .947–.988 .002 … …
Abbreviations: CI, confidence interval; CMV, cytomegalovirus; F, female; GFR, glomerular filtration rate (Modification of Diet in Renal Disease formula); M, male; mTORi, mammalian target
of rapamycin inhibitor; OR, odds ratio; Tx, transplantation.

steroid therapy and the recruitment into the lung of both CD4 implement, precisely because of this low incidence, and that is
and CD8 [24]. In transplanted patients, in comparison with why most studies on PCP in organ transplantation adopted a
HIV patients, therapeutic immunosuppression has a pleio- case-control design. Even though this study helped us to define
tropic inhibitory effect on CD4 and CD8 compartments and the criteria for restarting prophylaxis, further inquiry into the
also on other lymphocytic subsets [25], perhaps the reason why criteria of termination of prophylaxis is needed. Both centers
we identified lymphopenia as the most significantly associated performed 3-month universal prophylaxis for CMV prevention
lymphocytic marker to PCP. Finally, we observed a very poor after the diagnosis of rejection but no additional monitoring
prognosis of PCP, with 30% of graft failure and 17% of death following the complete course of CMV prophylaxis, which
in the 3  years following PCP. These results confirm a recent would decrease the number of detected CMV DNAemias [27]
multicenter case-control study reporting 38% of graft loss after but not the number of CMV diseases. Cases had significantly
PCP [26]. In the absence of systematic kidney graft biopsies, we less acute postrejection prophylaxis, but all PCP occurred after
could not analyze exhaustively the causes of graft failure after this prophylaxis period (median, 16.5 [quartiles, 11.5–27.2]).
PCP (see Supplementary Results for details). Consequently, one can suppose that PCP would still have oc-
The limitations of the present study are its retrospective and curred if those patients had received prophylaxis. The asso-
case-control design, inherent to the scarcity of this infection. ciation between rituximab and PCP risk has been difficult to
However, randomized prospective trials would be complex to analyze because no patients received rituximab as induction

Risks/Outcomes of Pneumocystis Pneumonia  •  cid 2021:73 (1 October) • e1461

 A Notes
40 Acknowledgments. The authors thank Professor Jay Fishman for his
helpful advice; Catherine Rio as a nurse coordinator of Bordeaux’s center;
and Guillaume Rebillon for his invaluable English reading and corrections.
Cumulative incidence
of graft failure (%)
30 Financial support. H.  K.  is the recipient of a “Bourse doctorale de la
Fondation pour la Recherche Médicale.”
Potential conflicts of interest. The authors: No reported conflicts of
20
interest. All authors have submitted the ICMJE Form for Disclosure of
Potential Conflicts of Interest. 
10
References

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

0 1. Fishman  JA, Gans  H; AST Infectious Diseases Community of Practice.
Pneumocystis jiroveci in solid organ transplantation: guidelines from the
0 6 12 18 24 30 36
American Society of Transplantation Infectious Diseases community of practice.
B Clin Transplant 2019; 33:e13587.
40 2. Iriart X, Bouar ML, Kamar N, Berry A. Pneumocystis pneumonia in solid-organ
transplant recipients. J Fungi (Basel) 2015; 1:293–331.
3. Sassi M, Ripamonti C, Mueller NJ, et al. Outbreaks of Pneumocystis pneumonia
Cumulative incidence

30 in 2 renal transplant centers linked to a single strain of Pneumocystis: implications

of death (%)

for transmission and virulence. Clin Infect Dis 2012; 54:1437–44.

4. Le Gal S, Damiani C, Rouillé A, et al. A cluster of Pneumocystis infections among
20 renal transplant recipients: molecular evidence of colonized patients as potential
infectious sources of Pneumocystis jirovecii. Clin Infect Dis 2012; 54:e62–71.
5. de Boer MGJ, Walzer PD, Mori S. Healthcare related transmission of Pneumocystis
10
pneumonia: from key insights toward comprehensive prevention. Transpl Infect
Dis 2018; 20:e12942.
6. Fillaux J, Malvy S, Alvarez M, et al. Accuracy of a routine real-time PCR assay for
0
the diagnosis of Pneumocystis jirovecii pneumonia. J Microbiol Methods 2008;
0 6 12 18 24 30 36
75:258–61.
7. Issa N, Gabriel F, Baulier G, et al. Pneumocystosis and quantitative PCR. Med Mal
Time from day 0 of Pneumocystis pneumonia (months) Infect 2018; 48:474–80.
8. Neofytos D, Hirzel C, Boely E, et al; Swiss Transplant Cohort Study. Pneumocystis
Number jirovecii pneumonia in solid organ transplant recipients: a descriptive analysis for
70 58 51 41 38 33 33
at risk the Swiss Transplant Cohort. Transpl Infect Dis 2018; 20:e12984.
9. Apoil PA, Puissant-Lubrano B, Congy-Jolivet N, et al. Influence of age, sex and
HCMV-serostatus on blood lymphocyte subpopulations in healthy adults. Cell
Figure 3.  Three-year cumulative incidence curves of graft loss (A) and patient Immunol 2017; 314:42–53.
death (B) using Nelson-Aalen estimator. 10. Garrigue I, Doussau A, Asselineau J, et al. Prediction of cytomegalovirus (CMV)
plasma load from evaluation of CMV whole-blood load in samples from renal
transplant recipients. J Clin Microbiol 2008; 46:493–8.
11. Ljungman P, Boeckh M, Hirsch HH, et al; Disease Definitions Working Group of
treatment, and all rituximab-treated patients for rejection re-
the Cytomegalovirus Drug Development Forum. Definitions of cytomegalovirus
ceived concomitantly corticosteroid boluses, which led recent infection and disease in transplant patients for use in clinical trials. Clin Infect Dis
guidelines not to recommend prophylaxis for patients exclu- 2017; 64:87–91.
12. Klein JP. Small sample moments of some estimators of the variance of the Kaplan-
sively receiving rituximab [28]. Finally, the small number of pa- Meier and Nelson-Aalen estimators. Scand J Statist 1991; 18:333–40.
tients receiving rATG during acute rejection (n = 4) led us to 13. Iriart X, Challan Belval T, Fillaux J, et al. Risk factors of Pneumocystis pneumonia
in solid organ recipients in the era of the common use of posttransplantation pro-
avoid exploring it as a risk factor. phylaxis. Am J Transplant 2015; 15:190–9.
The strengths of this study include its large number of cases 14. de Boer MG, Kroon FP, le Cessie S, de Fijter JW, van Dissel JT. Risk factors for
Pneumocystis jirovecii pneumonia in kidney transplant recipients and appraisal
compared with other studies [8, 13, 14, 26, 29]. Our cohort was of strategies for selective use of chemoprophylaxis. Transpl Infect Dis 2011;
also very homogeneous, receiving the same 6-month prophy- 13:559–69.
15. de Boer MG, de Fijter JW, Kroon FP. Outbreaks and clustering of Pneumocystis
laxis and similar immunosuppressive regimens, and included
pneumonia in kidney transplant recipients: a systematic review. Med Mycol 2011;
kidney transplant recipients only and no other SOT recipients 49:673–80.
who could display other risk factors of P. jirovecii infection. In 16. De  Castro  N, Xu  F, Porcher  R, Pavie  J, Molina  JM, Peraldi  MN. Pneumocystis
jirovecii pneumonia in renal transplant recipients occurring after discontinuation
summary, PCP is a major graft- and life-threatening compli- of prophylaxis: a case-control study. Clin Microbiol Infect 2010; 16:1375–7.
cation in kidney transplant recipients. A patient-targeted pro- 17. Neff RT, Jindal RM, Yoo DY, Hurst FP, Agodoa LY, Abbott KC. Analysis of USRDS:
incidence and risk factors for Pneumocystis jiroveci pneumonia. Transplantation
phylaxis based on simple criteria, such as chronic lymphopenia 2009; 88:135–41.
<1000/μL and/or corticosteroid bolus treatment, could allow 18. Ali  MF, Dasari  H, Van  Keulen  VP, Carmona  EM. Canonical stimulation of the
NLRP3 inflammasome by fungal antigens links innate and adaptive B-lymphocyte
some selected patients to be spared from this permanent danger. responses by modulating IL-1β and IgM production. Front Immunol 2017; 8:1504.
19. Ghadimi  M, Mohammadpour  Z, Dashti-Khavidaki  S, Milajerdi  A. m-TOR in-
hibitors and risk of Pneumocystis pneumonia after solid organ transplantation: a
systematic review and meta-analysis. Eur J Clin Pharmacol 2019; 75:1471–80.
Supplementary Data
20. Korkmaz Ekren P, Töreyin ZN, Nahid P, et al. The association between cytomeg-
Supplementary materials are available at Clinical Infectious Diseases on- alovirus co-infection with Pneumocystis pneumonia and mortality in immuno-
line. Consisting of data provided by the authors to benefit the reader, compromised non-HIV patients. Clin Respir J 2018; 12:2590–7.
the posted materials are not copyedited and are the sole responsibility 21. Lee S, Park Y, Kim SG, Ko EJ, Chung BH, Yang CW. The impact of cytomegalo-
of the authors, so questions or comments should be addressed to the virus infection on clinical severity and outcomes in kidney transplant recipients
corresponding author. with Pneumocystis jirovecii pneumonia. Microbiol Immunol 2020; 64:356–65.

e1462 • cid 2021:73 (1 October) • Kaminski et al

22. Werbel WA, Ison MG, Angarone MP, Yang A, Stosor V. Lymphopenia is associ- solid organ transplant recipients: a systematic review and meta-analysis. Clin
ated with late onset Pneumocystis jirovecii pneumonia in solid organ transplanta- Transplant 2018; 32:e13339.
tion. Transpl Infect Dis 2018; 20:e12876. 27. Jorgenson  MR, Descourouez  JL, Lyu  B, et  al. The risk of cytomegalovirus in-
23. Kolls JK, Habetz S, Shean MK, et al. IFN-gamma and CD8+ T cells restore host fection after treatment of acute rejection in renal transplant recipients. Clin
defenses against Pneumocystis carinii in mice depleted of CD4+ T cells. J Immunol Transplant 2019; 33:e13636.
1999; 162:2890–4. 28. Mikulska  M, Lanini  S, Gudiol  C, et  al. ESCMID Study Group for Infections in
24. Sukura A, Konttinen YT, Sepper R, Lindberg LA. Recovery from Pneumocystis Compromised Hosts (ESGICH) consensus document on the safety of targeted and
carinii pneumonia in dexamethasone-treated Wistar rats. Eur Respir J 1995; biological therapies: an infectious diseases perspective (agents targeting lymphoid
8:701–8. cells surface antigens [I]: CD19, CD20 and CD52). Clin Microbiol Infect 2018;
25. Garvy BA, Qureshi MH. Delayed inflammatory response to Pneumocystis carinii 24:S71–82.
infection in neonatal mice is due to an inadequate lung environment. J Immunol 29. Hosseini-Moghaddam  SM, Shokoohi  M, Singh  G, et  al. A multicenter case-
2000; 165:6480–6. control study of the effect of acute rejection and cytomegalovirus infection on
26. Hosseini-Moghaddam  SM, Krishnan  RJ, Guo  H, Kumar  D. Cytomegalovirus Pneumocystis pneumonia in solid organ transplant recipients. Clin Infect Dis

Downloaded from https://academic.oup.com/cid/article/73/7/e1456/5937156 by Universidad Pontificia Bolivariana user on 16 March 2023

infection and graft rejection as risk factors for pneumocystis pneumonia in 2019; 68:1320–6.

Risks/Outcomes of Pneumocystis Pneumonia  •  cid 2021:73 (1 October) • e1463