Received: 4 June 2020 | Revised: 29 December 2020 | Accepted: 4 January 2021

DOI: 10.1111/tid.13569

ORIGINAL ARTICLE

Association between cytomegalovirus infection and allograft

rejection in a large contemporary cohort of heart transplant
recipients

David Boutolleau1 | Guillaume Coutance2 | Eva Désiré2 | Adrien Bouglé3 |

Nicolas Bréchot4,5 | Pascal Leprince2 | Shaida Varnous2

1
Virology Department, Sorbonne Université,
INSERM UMR U1136, Institut Pierre Louis Abstract
d’Epidémiologie et de Santé Publique Background: Cytomegalovirus (CMV) infection remains a common complication after
(iPLESP), Team 3 THERAVIR, and Assistance
Publique-­Hôpitaux de Paris (AP-­HP), Pitié-­ heart transplantation (HTx). The association between CMV infection and allograft
Salpêtrière Hospital, National Reference rejection is debated in the era of efficient prophylactic antiviral therapies.
Centre for Herpesviruses, Paris, France
2 Methods: This single-­center cohort study utilized a highly phenotyped database of
Department of Cardiac and Thoracic
Surgery, Cardiology Institute, Pitié HTx recipients (2012-­2016). The primary endpoint was the analysis of the associa-
Salpêtrière Hospital, Assistance Publique-­
tion between CMV infection (CMV load ≥ 500 IU/mL whole blood) and the risk of
Hôpitaux de Paris (AP-­HP), Sorbonne
University Medical School, Paris, France allograft rejection (cellular rejection ≥ 1R1B, antibody-­mediated rejection ≥ pAMR1).
3
Department of Anesthesiology and Critical Secondary endpoints included the analysis of a higher CMV load threshold
Care Medicine, Sorbonne Université, UMR
INSERM 1166, IHU ICAN, Assistance
(≥10 000 IU/mL) and different risk periods after PCR positivity. A mixed-­effect logis-
Publique-­Hôpitaux de Paris (AP-­HP), tic regression model with a random intercept was applied. Results were adjusted for
Cardiology Institute, Pitié-­Salpêtrière
Hospital, Paris, France
important risk factors of rejection.
4
Department of Medical Intensive Care Results: Overall, 384 patients were included and 6388 CMV loads and 3,494 endo-
Unit, Cardiology Institute, Pitié Salpêtrière myocardial biopsies were analyzed. CMV infections ≥ 500 IU/mL were diagnosed
Hospital, Assistance Publique-­Hôpitaux de
Paris (AP-­HP), Sorbonne University Medical on 1223 (19.2%) blood samples from 284 (72.1%) patients and allograft rejections
School, Paris, France on 246 biopsies (7%) from 149 patients (38.8%). We did not find any association be-
5
INSERM, UMRS 1166-­ICAN, Institute of
tween CMV infection ≥ 500 IU/mL and rejection (univariable: OR 0.94, 95% CI [0.61,
Cardiometabolism and Nutrition, Paris,
France 1.45], P = .78, multivariable: OR 0.86, 95% CI [0.55, 1.33], P = .85). These results
were consistent when analyzing a higher CMV load threshold and different periods
Correspondence
Dr Shaida Varnous, Service de Chirurgie of risk, reinforced by internal validation procedures and a posteriori calculation of the
Thoracique et Cardio-­Vasculaire, Institut
power (primary endpoint: power = 0.82, 95% CI [0.79-­0.84]) and reproducible across
de Cardiologie -­ Groupe hospitalier Pitié-­
Salpêtrière, 47-­83, boulevard de l'Hôpital different clinical scenarios.
-­ 75651 Paris Cedex 13.
Conclusions: CMV infection was not associated with an increased risk of rejection in
Email: shaida.varnous@aphp.fr
a contemporary cohort of HTx recipients.
Funding information
The study was funded by a grant from
KEYWORDS
Biotest laboratories. Guillaume Coutance
received research grants from the allograft rejection, cytomegalovirus, heart transplantation, risk factor
Assistance Publique des Hôpitaux de Paris
(2019).

David Boutolleau and Guillaume Coutance equal contribution.

Transpl Infect Dis. 2021;00:e13569. wileyonlinelibrary.com/journal/tid © 2021 Wiley Periodicals LLC | 1 of 9

https://doi.org/10.1111/tid.13569
2 of 9 | BOUTOLLEAU et al.

1 | I NTRO D U C TI O N 2 | PATI E NT S A N D M E TH O DS

Heart transplantation (HTx) remains the most valuable thera- The data that support the findings of this study are available from
peutic option for patients with end-­s tage heart failure.1-­3 More the corresponding author upon reasonable request.
than 5000 HTx procedures are performed each year worldwide.4
Despite major advances in the field, cytomegalovirus (CMV) in-
fection remains one of the most common complications affecting 2.1 | Patients, sources of data, and study design
solid organ transplant recipients. 5,6 In addition to their direct ef-
fects, CMV infections also have indirect effects associated with We performed a retrospective single-­center observational study
higher rates of all types of infections, graft loss, allograft rejection, based on the analysis of data collected prospectively at our in-
morbidity, and mortality. 6-­9 Anti-­CMV therapies have been shown stitution (La Pitié-­S alpêtrière Hospital, Paris, France). We con-
not only to decrease the incidence of CMV infection10,11 but also sidered all consecutive heart allograft recipients transplanted
to reduce the incidence of allograft rejection.12-­14 Whether CMV at our institution between January 1, 2012 and December 31,
infections still constitute an independent risk factor for allograft 2016 (n = 453). Patients were excluded from the study if they
rejection after HTx in the era of efficient antiviral strategies re- died ≤ 2 weeks post-­t ransplant, were followed-­up at another
mains a matter of debate. Recent studies have reported conflicting center, did not undergo any CMV load in blood or endomyocardial
results but they were limited by several factors, including limited biopsy during the first year, or refused to participate. A flowchart
sample size, lack of important predictive variables in multivariable is provided in Figure 1. Patients were followed until September
models to assess accurately the independent association between 2019. Our study complies with the 2000 Declaration of Helsinki
CMV infection and rejection, and inability of the statistical mod- and 2008 Declaration of Istanbul. An ethics committee reviewed
els applied to account for the longitudinal characteristics of data the study and waived the need for approval (French law for the
15,16
and the risk of recurrence of allograft rejection. We aimed to retrospective analysis of data collected as part of usual patient
evaluate the association between CMV infection and allograft care). Informed consent concerning permission to use collected
rejection during the first year post-­t ransplant in a contemporary data was obtained from all patients at the time of enrollment on
cohort of HTx recipients. To achieve this, we applied a statisti- the waiting list. Baseline donor and recipient characteristics were
cal model adapted to the analysis of longitudinal data to a highly obtained from the prospective national registry CRISTAL database
phenotyped cohort of HTx recipients and adjusted the association (French National Agency for Organ Procurement, CNIL—­National
between CMV infection and rejection for the main risk factors for French Commission for Bioinformatics Data and Patient Liberty:
rejection. 363505). Virologic, immunological, and histological data were

F I G U R E 1 Flow chart of the study. A

total of 384 patients were included in the
study. CMV, Cytomegalovirus
BOUTOLLEAU et al. | 3 of 9

collected as part of the HEARTS registry (NCT03393793, CNIL: therapy based on either intravenous ganciclovir (5 mg/kg bid) or oral val-
2206319v0). Additional details concerning the databases are pro- ganciclovir (900 mg qd) for 3 months followed by a preemptive strategy
vided in the Supplementary appendix. until the end of the first year. Seropositive recipients (R+) were treated
according to a preemptive strategy during the first year. Full-­dose ganci-
clovir/valganciclovir (ganciclovir 5 mg/kg bid, valganciclovir 900 mg bid)
2.2 | Objectives and outcome measurements was given for at least 1 month in case of a CMV load ≥ 10 000 IU/mL
of blood until 2 consecutive undetectable CMV loads, followed by sec-
The primary aim of our study was to analyze the association between ondary prophylaxis at half-­dose for 2 additional weeks. CMV load mon-
CMV infections and biopsy-­proven allograft rejection using a compre- itoring was performed at least once weekly during initial hospitalization
hensive database that included detailed characteristics of donors, re- and in the first month (additional tests in case of increasing viral load),
cipients, and transplant procedure, as well as virologic, immunological, then every 10 days during the second month, every 3 weeks between 3
and histological data collected prospectively. Our secondary objective and 6 months, and monthly between 6 and 12 months post-­transplant.
was to describe the epidemiology of CMV infections and allograft re- Additional CMV loads were performed in case of clinical indication or to
jection in a contemporary cohort of HTx recipients. monitor the response to antiviral therapies.
The primary endpoint was the analysis of the association be-
tween CMV load ≥ 500 IU/mL in whole blood (virologic threshold
for CMV infection) and allograft rejection. Allograft rejection was 2.4 | Cytomegalovirus serology and viral load
defined as a combined endpoint including both acute cellular rejec-
tions (ACR) greater than or equal to 1R1B and/or antibody-­mediated Prior to HTx, donors and recipients were tested for CMV immune
rejection greater than or equal to pAMR1.17,18 status using the LIAISON® CMV IgG II test (DiaSorin, Saluggia, Italy).
Secondary endpoints included a) the analysis of the association During the post-­HTx period, CMV load was measured in whole blood
between CMV load ≥ 10 000 IU/mL in whole blood (clinical thresh- samples using the artus® CMV QS-­RGQ kit (Qiagen).
old corresponding to the initiation of preemptive antiviral therapies
at our center) and allograft rejection; b) the analysis of different risk
period after PCR positivity (Figure S1): i) during CMV load greater than 2.5 | Immunology
or equal to the threshold value + 1 week after, ii) during CMV load
greater than or equal to the threshold value + 2 weeks after, iii) during Detection of anti-­HLA antibodies pre-­ and post-­HTx was based on
CMV load greater than or equal to the threshold value + 4 weeks Luminex single antigen bead technology (One Lambda), with the
after and, iv) during CMV load greater than or equal to the threshold MFI value recoded after normalization with the baseline formula.
value + 3 months after; and c) the analysis of association between Beads showing a Mean Fluorescence Intensity (MFI) value ≥ 500
CMV infection and cardiac allograft vasculopathy (CAV) ≥ grade 1 ac- were considered significant. This value was based on the valida-
cording to international guidelines.19 tion criteria for negative control sera. Donor-­specific antibodies
Sensitivity analyses were performed for severe rejections (DSA) were defined as anti-­HLA antibodies directed against do-
defined as an ACR ≥ 2R3A and/or AMR ≥ pAMR2 and for CMV nors’ HLA antigens.
diseases. CMV diseases were defined as a CMV infection with attrib-
utable symptoms requiring hospitalization for intravenous antiviral
therapies corresponding mostly to tissue-­invasive CMV diseases as 2.6 | Endomyocardial biopsies
6
defined by international guidelines.
Endomyocardial biopsies (EMBs) were performed, processed, and
examined according to current standards. 20 The routine protocol
2.3 | Immunosuppressive and and details of the preparation and interpretation of EMBs have been
cytomegalovirus protocols described previously. 21 Additional biopsies were performed in case
of clinical indication. The EMBs results were classified as cellular
Initial immunosuppression after HTx was based on induction therapy (ACR 0R to 3R, including detailed diagnosis according to the 1990
with rabbit anti-­thymocyte globulin (rATG; Thymoglobuline; Genzyme, working formulation) or AMR (pAMR 0 to 3) according to the ISHLT
Lyon, France; 1.5 mg/kg/d for 5 days) or basiliximab (Simulect, Novartis, guidelines.17,18 ACR ≥ 1R1B and AMR, even subclinical, were treated
Basel, Switzerland). Prophylactic immunosuppressive therapy included according to our local protocol detailed in Supplementary Appendix.
calcineurin inhibitors (CNI), mycophenolate mofetil, and corticoster-
oids with dosage recommended by ISHLT guidelines.20
Our therapeutic protocol for CMV infection remained unchanged 2.7 | Cardiac allograft vasculopathy
during the study period and depended on the patient's CMV serostatus
at transplant. High-­risk patients, defined as seronegative recipients with Routine coronary angiograms were performed 1 year after HTx and
seropositive donors (D+/R−), received primary prophylaxis antiviral then every 2 years afterwards, or in the case of clinical indication.
4 of 9 | BOUTOLLEAU et al.

Staging of cardiac allograft vasculopathy (CAV) was performed using TA B L E 1 Main characteristics of patients
19
the recommended ISHLT nomenclature.
Overall
Recipient characteristics N population

Age, years—­mean (SD) 384 49.9 ± 12.2

2.8 | Statistical analyses
Female gender—­no. (%) 384 80 (20.8)
Ischemic cardiomyopathy—­no. (%) 384 127 (33.1)
Qualitative variables are presented as frequencies and quan-
Long term MCS (LVAD or TAH)—­no. (%) 384 51 (13.3)
titative variables as the mean and standard deviation (SD).
ECMO at transplant—­no. (%) 384 108 (28.1)
Cumulative survival curves for the time-­t o-­e vent analyses
were constructed according to the Kaplan-­M eier method. The Mechanical ventilation at 384 13 (3.4)
transplant—­no. (%)
curves describing the evolution of the probability of CMV infec-
CMV seropositivity—­no. (%) 384 255 (66.4)
tions and rejection over time were constructed by calculating
the probability of a positive test for each time post-­t ransplant Donor characteristics

(±1 week) and by smoothing the resulting curve with the local Age, years—­mean (SD) 384 46.2±14

polynomial option. The association between CMV infection and Female gender—­no. (%) 384 137 (35.7)
allograft rejection was studied using a mixed-­e ffect logistic re- Cause of death—­no. (%) 384
gression model with a random intercept. Multivariable analyses Traumatic 123 (32.1)
were then performed to adjust the results for important risk fac- Cerebrovascular 181 (47.1)
tors for rejection including the presence of donor-­s pecific an- Other 80 (20.8)
tibodies, recipient's age, and number of HLA mismatches. 21-­23 Transplant characteristics
Internal validation procedures were then performed using 200
Ischemic time, ≥4 h—­no. (%) 384 62 (16.2)
bootstrap samples for the primary endpoint. Descending selec-
Dual organ transplantation—­no. (%) 384 25 (6.5)
tion was used and the final multivariable model selected fac-
Redo transplantation—­no. (%) 384 10 (2.6)
tors that were significant at 5%. A posteriori calculation of the
Gender mismatch (female D, male 384 88 (22.9)
power of the statistical tests was performed using the ipdpower
R)—­no. (%)
package (simulation-­b ased power calculations, 1000 samples
CMV status at transplant—­no. (%) 384
generated). 24 Sensitivity analyses were performed using (a) time
Donor seronegative—­recipient 58 (15.1)
varying covariates Cox model to identify predictive variables as-
seronegative: D−/R−
sociated with the risk of first episode rejection and (b) logistic
Recipient seropositive: R+ 255 (66.4)
generalized estimating equation model using an exchangeable
Donor seropositive—­recipient 71 (18.5)
covariance matrix and a robust covariance matrix estimator
seronegative: D+/R−
(sandwich estimator) to account for within-­s ubject correlation
Immunology
and obtain population-­averaged odds-­r atios. A P-­v alue < .05 was
HLA A/B/DR/DQ mismatch—­mean 384 6.4 (1.3)
considered statistically significant. Statistical analysis was per- (SD)
formed with STATA 15.1 software (StataCorp).
Pre-­formed DSA (MFI ≥500)—­no. (%) 384 176 (45.8)
ATG induction therapy—­no. (%) 384 372 (96.9)
First immunosuppressive regimen 384
3 | R E S U LT S
CYA + MMF + corticosteroids 357 (93.0)
TAC + MMF + corticosteroids 27 (7.0)
3.1 | Characteristics of patients
Abbreviations: ATG, Antithymocyte globulines; CMV, cytomegalovirus;
A total of 384 patients were included. A flowchart is provided in CYA, cyclosporine; DSA, donor-­specific antibodies; ECMO,
extracorporeal membrane oxygenation; HTx, heart transplantation;
Figure 1. The characteristics of patients are detailed in Table 1.
LVAD, left ventricular assist devices; MCS, mechanical circulatory
Fifty-­eight (15.1%), 255 (66.4%), and 71 (15.5%) patients were clas- support; MFI, mean fluorescence intensity; MMF, mycophenolate
sified as having low (donor seronegative, recipient seronegative mofetif; TAC, tacrolimus, TAH, total artificial hearts.
[D−/R−]), intermediate (recipient seropositive [R+]), and high (donor
seropositive, recipient seronegative [D+/R−]) risk of CMV infec-
tion. Patients were mostly men (78.7%) and their mean age at 3.2 | Epidemiology of cytomegalovirus infections
transplant was 49.9 ± 12.2 years. All patients received induction
therapy, which was mostly based on ATG infusions (96.9%). The During the study period, a total of 6388 CMV PCR tests were per-
routine immunosuppressive regimen was based on a triple therapy formed (16.4 ± 4.8 per patient, Table 2). Among them, 1223 (19.2%) and
with CNI (cyclosporine: n = 357, 93%), mycophenolate mofetil, and 396 (6.2%) CMV loads were found to be greater than or equal to 500
corticosteroids. and 10 000 IU/mL, respectively. Median duration of CMV load ≥ 500
BOUTOLLEAU et al. | 5 of 9

TA B L E 2 Number of tests performed during the first-­year 3.4 | Primary endpoint: association between
post-­transplant cytomegalovirus infection ≥ 500 UI/mL and
6-­ allograft rejection
0-­1 mo 1-­3 mo 3-­6 mo 12 mo

CMV loads 1281 1725 1385 1997 We did not find any association between CMV infection with a viral

EMBs 191 961 946 1396 load ≥ 500 IU/mL and allograft rejection (univariable analysis: OR
0.94, 95% CI [0.61, 1.45], P = .78, multivariable analysis: OR 0.86,
Abbreviations: CMV, Cytomegalovirus; EMBs, endomyocardial biopsies.
95% CI [0.55, 1.33], P = .85, Table 4 and Figure 2A). After resampling
procedures, CMV infections ≥ 500 UI/mL were identified as an in-
and ≥ 10 000 IU/mL was 31 days (interquartile range—­IQR = 21-­47) dependent risk factor for allograft rejection in only 5% of bootstrap
and 15 days (IQR = 8-­28), respectively. A total of 284 (72.1%) and 194 samples. Assuming an odds ratio of CMV infections of 2, the power
(49.2%) patients had at least one CMV load ≥ 500 and ≥ 10 000 IU/ of the analysis of the primary endpoint was calculated at 0.82 (95%
mL, respectively. The risk of CMV replication depended on the time CI = 0.79-­0.84, simulation-­based power calculation).
post-­transplant and the CMV serostatus at transplant (Figure S2A-­D).
The risk was particularly high and early in seropositive recipients (R+),
reflecting the absence of universal prophylaxis in this group. Among 3.5 | Secondary endpoints
these patients, 152 patients (59.6%) were diagnosed with CMV in-
fection ≥ 10 000 UI/mL. All of them were treated with full-­dose val- We did not find any association between CMV infection with a viral
ganciclovir. An additional limited group of 8 patients (3.1%) received load ≥ 10 000 IU/mL and allograft rejection (univariable analysis: OR
full-­dose valganciclovir for rapidly increasing CMV viral load (range: 0.61, 95% CI [0.26, 1.47], P = .78, multivariable analysis: OR 0.58,
7000-­9999 UI/mL). On the other hand, the risk of replication was de- 95% CI [0.49, 1.18], P = .85, Figure 2B).
layed beyond 3 months in D+/R− patients (prophylaxis strategy). A We then performed complementary analyses to account for a
total of 14 cases of CMV disease were diagnosed in 14 patients (3.6%) possible remnant effect of CMV infection on the risk of rejection.
during the first-­year post-­transplant. Remnant effects at 1 or 2 weeks after the last CMV load above the
threshold values were not associated with an increased risk of al-
lograft rejection (Figure 2A,B). We observed a trend toward an in-
3.3 | Epidemiology of allograft rejection creased risk of rejection when considering a 4-­week remnant effect
for CMV loads ≥ 500 IU/mL (OR 1.23, 95% CI [0.88, 1.72], P = .23).
During the first year, a total of 3494 EMBs were performed (9.3 ± 3.0 However, when adjusting for important risk factors for rejection, in-
per patient). Cellular rejections classified as 1R1B, 1R2, and ≥ 2R cluding recipient age, number of HLA mismatches, and presence of
were diagnosed in 154 (4.4%), 6 (0.2%), and 26 (0.7%) EMBs, re- DSA, the impact of CMV infection on the risk of rejection appeared
spectively. Antibody-­mediated rejections pAMR1(I+), pAMR1(H+), to be neutral (OR 1.03, 95% CI [0.72, 1.47], P = .89). Similar results
and pAMR2-­3 were diagnosed in 9 (0.26%), 55 (1.6%), and 3 (0.1%) were found when considering a 3-­month period at risk following
of EMBs, respectively. The detailed rejection diagnoses are given in CMV infection.
Table 3. Overall, allograft rejection was present in 246 (7.04%) EMBs We did not find any association between CMV infec-
and diagnosed in 149 patients (38.8%). The risk of rejection was tion ≥ 500 UI/mL and the CAV ≥ grade 1 or CAV ≥ grade 2 after uni-
highest in the early post-­transplant setting and decreased over time variable analyses (Figure 3A,B) or after adjustment for donor age and
during the first year (Figure S3A,B). Severe rejection was diagnosed gender (CAV ≥ grade 1: CMV infections: HR = 1.05, 95%CI = 0.66-­
in 29 EMBs (0.8%) from 28 patients (7.3%). 1.66; CAV ≥ grade 2: CMV infections: HR = 1.03, 95%CI = 0.49-­
2.17). Similar results were found for CMV infection ≥ 10 000 UI/ml.

TA B L E 3 Detailed results of endomyocardial biopsies during the

first-­year post-­transplant
3.6 | Sensitivity analyses
pAMR0 pAMR1(I+) pAMR1(H+) pAMR2-­3

ACR 0R 2660 9 31 2 Sensitivity analyses were performed to test the consistency of our
ACR 1R1A 381 0 17 1 results across various clinical scenarios. Firstly, we aimed to account
ACR 1R1B 120 0 4 0 for CMV serologic status at transplant; however, CMV status was
ACR 1R2 4 0 1 0 not an independent risk factor for rejection in our cohort (D−/R−,

ACR 2R-­3R 22 0 2 0 reference; R+, OR 1.00, 95% CI [0.63, 1.60]; D+/R−, OR 0.80, 95%
CI [0.43, 1.47]; P = .39) and no interaction between the variables
Classification of allograft rejection according to the International
CMV status and CMV loads ≥ 500 IU/mL was found. Secondly, we
Society for Heart and Lung Transplantation guidelines (22-­24).
Abbreviations: ACR, acute cellular rejection; pAMR, pathologic analyzed the association between CMV disease and allograft rejec-
antibody-­mediated rejection. tion. We did not observe any increase in the risk of allograft rejection
6 of 9 | BOUTOLLEAU et al.

following CMV disease (OR = 0.95, 95%CI [0.10, 9.98], P = .96). This

.006
<.001
value

.06
.35

.05
analysis was limited by the small number of CMV diseases in our

P
cohort (n = 14). Third, we did not observe any significant increase in
the risk of severe rejection during CMV infection ≥ 500 IU/mL (OR

0.86 [0.74-­1.00]

1.24 [1.06-­1.44]
0.98 [0.97-­0.99]

1.41 [0.97-­2.05]
1.18, 95% CI [0.50, 2.5], P = .45).

1.19 [0.82-­1.76]
Finally, very similar results were found when using alternative
OR [95%CI] statistical models (time varying covariates Cox model—­Figure S4—­
+3 mo

and logistic generalized estimating equation model—­Figure S5).

P value

.008
<.001

.053
4 | D I S CU S S I O N
.89

05

Based on a deeply phenotyped and contemporary cohort of 384

Abbreviations: CI, confidence interval; CMV, cytomegalovirus; DSA, donor-­specific antibodies; HTx, heart transplantation; MM, mismatch; OR, odds ratio.
1.44 [1.00-­2.09]
0.86 [0.75-­1.00]

1.23 [1.06-­1.43] HTx recipients and on the analysis of the results of more than 6300
1.03 [0.72-­1.47]
0.98 [0.97-­0.99]

CMV loads and 3400 EMBs performed during the first year post-­
OR [95%CI]

transplant, we did not observe any significant increase in the risk

+4 wks

of allograft rejection during or weeks after CMV infection. These

Association between CMV infection ≥ 500 UI/mL and allograft rejection. Multivariable mixed effect logistic regression

results were reinforced by internal validation procedures, a poste-

riori calculation of the power of the analysis of the primary end-
P value

.008
<.001

.053

point and by sensitivity analyses exploring various clinical scenarios.

.05
.51

Additionally, we provided contemporary data on the epidemiology

of both CMV infections and allograft rejections during the first year
post-­transplant. Our results highlight the high incidence of CMV
1.44 [1.00-­2.09]
0.87 [0.59-­1.30]

0.87 [0.75-­1.00]

1.23 [1.05-­1.43]
0.98 [0.97-­0.99]

infections in R + patients treated with a preemptive strategy. Our

OR [95%CI]

study has several strengths including (a) the large sample size and
+2 wks

number of CMV load measurements and EMBs analyzed, (b) the ap-
plication of a statistical model adapted to the analysis of longitudinal
data, (c) the adjustment of the association between CMV infection
P value

.008
<.001

.053

and rejection for major risk factors for allograft rejections, and (d)
.05
.42

the internal validation procedure that reinforced our results.

Whether CMV infection remains an independent risk factor for
allograft rejection in the era of potent antiviral therapies is an im-
0.83 [0.54-­1.29]

1.44 [1.00-­2.09]
0.87 [0.75-­1.00]

1.23 [1.05-­1.43]
0.98 [0.97-­0.99]

portant question in order to improve the stratification of the risk of

OR [95%CI]

allograft rejection. HTx offers a unique opportunity to answer this

+ 1 wk

question. On the one hand, HTx recipients are at high risk of CMV in-
fections.6 On the other hand, protocol-­based EMB monitoring, with
numerous EMBs performed during the first year post-­transplant,
P value

.008
<.001

.053

which is specific to HTx compared to other solid organ transplanta-

.05
.49

tions, allows an in-­depth analysis of the association between CMV

During CMV infection

infection and allograft rejection over time. 20

The principle of CMV-­induced allograft rejection has a strong bi-
1.44 [1.00-­2.09]
0.86 [0.55-­1.33]

0.87 [0.75-­1.00]

1.23 [1.05-­1.43]
0.98 [0.97-­0.99]

ological background. CMV infects endothelial cells, where it is able

OR [95%CI]

to replicate and persist during latency. This infection leads to the

upregulation of adhesion molecules on endothelial cells, thereby
promoting proinflammatory processes. This direct non-­immune
pathway promotes allograft endothelial inflammation25,26 resulting
CMV infection ≥500 UI/mL
Time post-­HTx (per 1-­week

HLA A-­B-­DR-­DQ MM (per

in increased immune cell recruitment and anti-­HLA antigen expres-

Recipient age (per 10-­yr

sion on the surface of endothelial cells, thus increasing the risk of

Predictive variables

1 MM increment)

allograft rejection. 27,28 Importantly, experimental data suggest that

Pre-­formed DSA

although antiviral therapies do limit CMV replication and dissemina-

increment)

increment)
TA B L E 4

tion, they do not attenuate the inflammatory potential of infected

endothelial cells. 29 CMV may also increase the risk of rejection
through indirect effects, particularly on the immune system, by
BOUTOLLEAU et al. | 7 of 9

F I G U R E 2 Association between cytomegalovirus infection and allograft rejections (mixed-­effect logistic regression). Forest plots
representing the odds ratio and 95% confidence interval of CMV infection on the risk of allograft rejection. Two different thresholds to
define CMV infection were used: (A) CMV viral load ≥ 500 IU/mL and (B) CMV viral load ≥ 10 000 IU/mL. Four different at-­risk periods were
analyzed: i) only during CMV load greater than or equal to the threshold value, ii) during CMV load greater than or equal to the threshold
value + 1 wk after, iii) during CMV load greater than or equal to the threshold value + 2 wks after, iv) during CMV load greater than or
equal to the threshold value + 4 wks after, and iv) during CMV load greater than or equal to the threshold value + 3 mo after. Univariable
(dashed line) and multivariable (solid line) analyses are provided. In multivariable analyses, the effect of CMV infection on the risk of allograft
rejection was adjusted for recipient age, number of HLA mismatches, presence of donor-­specific antibodies, and time post-­transplant. CMV
infections ≥ 500 and ≥ 10 000 IU/mL were not independent risk factors for allograft rejection in our cohort, even after accounting for a
possible remnant effect of CMV infection on the risk of rejection

F I G U R E 3 Association between
CMV infection ≥ 500 UI/mL and cardiac
allograft vasculopathy. Cardiac allograft
vasculopathy-­free survival. A, Cardiac
allograft vasculopathy ≥ grade 1 according
to international guidelines. B, Cardiac
allograft vasculopathy ≥ grade 2. We
did find any association between CMV
infections ≥ 500 UI/mL and cardiac
allograft vasculopathy. CAV, cardiac
allograft vasculopathy; CI, confidence
interval; HR, hazard radio; HTx, heart
transplantation

enhancing both humoral immunity and T-­cell responses. 26,30 In par- a contemporary cohort was provided by Stern et al They reported a
ticular, CMV infection elicits a strong and sustained cellular immune twofold increase in the risk of allograft rejections after CMV infec-
response from the host that targets the vascular endothelium.31,32 tions across all solid organ transplants, despite the use of potent an-
However, the clinical implication of CMV as an independent risk tiviral therapies.16 In the current study, we report discordant results
factor for allograft rejection in the era of potent antiviral therapies that may be explained by several differences between the two stud-
and prophylaxis/preemptive strategies remains debated. The stron- ies. First, the routine EMBs monitoring protocol during the first year
gest evidence of an increased risk of rejection after CMV infection in in our cohort was intense, pre-­defined, and comparable between
8 of 9 | BOUTOLLEAU et al.

patients with or without CMV infections. On the opposite, most of D I S C LO S U R E

the patients included in the previous study were non-­HTx recipi- The authors declare no conflicts of interest.
ents, and biopsies were mostly driven by a clinical indication and not
protocol based. Second, we applied a statistical model adapted to AU T H O R C O N T R I B U T I O N
the analysis of longitudinal data. Third, we were able to analyze the David Boutolleau contributed to the acquisition, analysis, and in-
precise duration of CMV infections. Fourth, we were able to adjust terpretation of data for the work and drafted the work. Guillaume
the effect of CMV infections for the most important risk factors for Coutance contributed to the acquisition, analysis, and interpretation
allograft rejection including the number of HLA mismatches and the of data for the work and drafted the work. Eva Désiré contributed
presence of circulating anti-­HLA DSA. Finally, our study included a to the acquisition of data for the work and drafted the work. Adrien
large single-­center homogeneous cohort of heart transplant recipi- Bouglé contributed to the acquisition and interpretation of data for
ents treated with the same CMV and immunosuppressive protocol the work and revised the manuscript critically for important intel-
during all the study period. lectual content. Nicolas Bréchot contributed to the acquisition and
Our current preemptive therapy for CMV-­positive recipients is interpretation of data for the work and revised the manuscript criti-
6
not supported by current international guidelines. However, the cally for important intellectual content. Pascal Leprince contributed
paucity of data in the field of HTx makes it difficult to recommend to the conception of the work and revised the manuscript critically
a unique strategy. In the context of an ATG-­based induction ther- for important intellectual content. Shaida Varnous contributed to
apy, we faced an important number of early post-­HTx valganciclovir-­ the design and the conception of the work and revised the manu-
induced neutropenia and recurrence of CMV infections following script critically for important intellectual content. All authors gave
the usual short-­term preemptive therapy. We therefore apply this their final approval of the version to be published and agreed to be
delayed and sustained preemptive strategy for CMV-­positive recip- accountable for all aspects of the work.
ients. Despite this, we did not find any association between CMV
infections and allograft rejection or CAV and a limited number of ORCID
CMV diseases were diagnosed. These data may result in more com- Guillaume Coutance https://orcid.org/0000-0003-4939-2175
fort with low-­level viremia in the CMV-­positive HTx population that
support a hybrid approach of both universal prophylaxis and pre- REFERENCES
emptive therapy. 1. Aissaoui N, Morshuis M, Maoulida H, et al. Management of end-­
Our study should be interpreted within the context of its limita- stage heart failure patients with or without ventricular assist
device: an observational comparison of clinical and economic out-
tions. First, we reported the absence of association between CMV
comes. Eur J Cardio-­Thorac Surg Off J Eur Assoc Cardio-­Thorac Surg.
infections and the risk of allograft rejection, which could be as a re- 2018;53(1):170-­177. https://doi.org/10.1093/ejcts/​ezx258
sult of type II error. However, the power of the statistical analysis of 2. Vierecke J, Gahl B, de By TMMH, et al. Results of primary biven-
the primary endpoint was greater than 0.80 to detect an OR of 2. tricular support: an analysis of data from the EUROMACS reg-
istry. Eur J Cardio-­Thorac Surg Off J Eur Assoc Cardio-­Thorac Surg.
Nevertheless, our study was underpowered to analyze the associ-
2019;56(6):1037-­1045. https://doi.org/10.1093/ejcts/​ezz173
ation between the association between CMV infections and severe 3. Mehra MR, Canter CE, Hannan MM, et al. The 2016 international society
rejection episodes (ie, ≥2R and/or ≥pAMR2). Second, our study did for heart lung transplantation listing criteria for heart transplantation: a
not include an external validation control group. Different antiviral and 10-­year update. J Heart Lung Transplant Off Publ Int Soc Heart Transplant.
2016;35(1):1-­23. https://doi.org/10.1016/j.healun.2015.10.023
immunosuppressive strategies may have resulted in a different epide-
4. Khush KK, Cherikh WS, Chambers DC, et al. The International Thoracic
miology of CMV infections. Third, the small number of CMV diseases Organ Transplant Registry of the International Society for Heart
or severe allograft rejections limit the analysis of the impact of CMV and Lung Transplantation: Thirty-­fifth Adult Heart Transplantation
in these particular settings. Importantly, we only considered CMV dis- Report-­2018; Focus Theme: Multiorgan Transplantation. J Heart
eases requiring hospitalization for intravenous antiviral therapies, cor- Lung Transplant Off Publ Int Soc Heart Transplant. 2018;37(10):1155-­
1168. https://doi.org/10.1016/j.healun.2018.07.022
responding mostly to tissue-­invasive CMV diseases since the presence
5. Razonable RR, Humar A. Cytomegalovirus in solid organ transplant
of a viral syndrome at the time of CMV infection was not prospectively recipients—­Guidelines of the American Society of Transplantation
collected in our database. Last, we did not analyze the impact of late Infectious Diseases Community of Practice. Clin Transplant.
CMV infections since that CMV load and EMB are not routinely per- 2019;33(9):e13512.
6. Kotton CN, Kumar D, Caliendo AM, et al. The third international
formed beyond the first year post-­transplant at our center.
consensus guidelines on the management of cytomegalovirus in
In conclusion, CMV infections were frequent but not associated solid-­o rgan transplantation. Transplantation. 2018;102(6):900-­
with an increased risk of allograft rejection in our contemporary co- 931. https://doi.org/10.1097/TP.00000​0 0000​0 02191
hort of HTx recipients. 7. Sagedal S, Nordal KP, Hartmann A, et al. The impact of cy-
tomegalovirus infection and disease on rejection episodes
in renal allograft recipients. Am J Transplant Off J Am Soc
AC K N OW L E D G M E N T S Transplant Am Soc Transpl Surg. 2002;2(9):850-­856. https://doi.
We acknowledge Saadia Jerbi and Pascale Weber for their assis- org/10.1034/j.1600-­6143.2002.20907.x
tance with data collection and Guillaume Rondeau for his help and 8. Grattan MT, Moreno-­C abral CE, Starnes VA, Oyer PE,
Stinson EB, Shumway NE. Cytomegalovirus infection is
support.
BOUTOLLEAU et al. | 9 of 9

associated with cardiac allograft rejection and atherosclerosis. 22. Jarcho J, Naftel DC, Shroyer TW, et al. Influence of HLA mismatch
JAMA. 1989;261(24):3561-­3566. on rejection after heart transplantation: a multiinstitutional study.
9. Loebe M, Schüler S, Zais O, Warnecke H, Fleck E, Hetzer R. The Cardiac Transplant Research Database Group. J Heart Lung
Role of cytomegalovirus infection in the development of coro- Transplant Off Publ Int Soc Heart Transplant. 1994;13(4):583-­595;
nary artery disease in the transplanted heart. J Heart Transplant. discussion 595–­596.
1990;9(6):707-­711. 23. Wever-­Pinzon O, Edwards LB, Taylor DO, et al. Association of
10. Humar A, Lebranchu Y, Vincenti F, et al. The efficacy and safety recipient age and causes of heart transplant mortality: impli-
of 200 days valganciclovir cytomegalovirus prophylaxis in high-­ cations for personalization of post-­t ransplant management-­A n
risk kidney transplant recipients. Am J Transplant Off J Am Soc analysis of the International Society for Heart and Lung
Transplant Am Soc Transpl Surg. 2010;10(5):1228-­1237. https://doi. Transplantation Registry. J Heart Lung Transplant Off Publ Int Soc
org/10.1111/j.1600-­6143.2010.03074.x Heart Transplant. 2017;36(4):407-­417. https://doi.org/10.1016/j.
11. Lowance D, Neumayer H-­H, Legendre CM, et al. Valacyclovir for healun.2016.08.008
the prevention of cytomegalovirus disease after renal transplan- 24. Kontopantelis E, Springate DA, Parisi R, Reeves D. Simulation-­
tation. International Valacyclovir Cytomegalovirus Prophylaxis based power calculations for mixed effects modeling: ipdpower in
Transplantation Study Group. N Engl J Med. 1999;340(19):1462-­ Stata. J Stat Softw. 2016;74(12):1–­25.
1470. https://doi.org/10.1056/NEJM1​99905​13340​1903 25. Shabir S, Smith H, Kaul B, et al. Cytomegalovirus-­associated CD4(+)
12. Reischig T, Jindra P, Mares J, et al. Valacyclovir for cytomegalovi- CD28(null) cells in NKG2D-­dependent glomerular endothelial in-
rus prophylaxis reduces the risk of acute renal allograft rejection. jury and kidney allograft dysfunction. Am J Transplant Off J Am Soc
Transplantation. 2005;79(3):317-­324. https://doi.org/10.1097/01. Transplant Am Soc Transpl Surg. 2016;16(4):1113-­1128. https://doi.
tp.00001​50024.01672.ca org/10.1111/ajt.13614
13. Potena L, Holweg CTJ, Chin C, et al. Acute rejection and cardiac 26. Kaminski H, Fishman JA. The cell biology of cytomegalovirus:
allograft vascular disease is reduced by suppression of subclinical implications for transplantation. Am J Transplant Off J Am Soc
cytomegalovirus infection. Transplantation. 2006;82(3):398-­4 05. Transplant Am Soc Transpl Surg. 2016;16(8):2254-­2269. https://doi.
https://doi.org/10.1097/01.tp.00002​29039.87735.76 org/10.1111/ajt.13791
14. Potena L, Valantine HA. Cytomegalovirus-­associated allograft 27. Mori DN, Kreisel D, Fullerton JN, Gilroy DW, Goldstein DR.
rejection in heart transplant patients. Curr Opin Infect Dis. Inflammatory triggers of acute rejection of organ allografts.
2007;20(4):425-­431. https://doi.org/10.1097/QCO.0b013​e3282​ Immunol Rev. 2014;258(1):132-­144. https://doi.org/10.1111/
59c33b imr.12146
15. Méndez-­Eirín E, Barge-­C aballero E, Paniagua-­Martín MJ, et al. 28. Lefaucheur C, Loupy A, Zeevi A. Complement-­binding anti-­HLA an-
Incidencia, factores de riesgo e impacto pronóstico de la infección tibodies and kidney transplantation. N Engl J Med. 2014;370(1):85-­
por citomegalovirus tras el trasplante cardiaco. Medicina Clínica. 86. https://doi.org/10.1056/NEJMc​1313506
2020;154(10):381-­387. 29. Waldman WJ, LeClaire JD, Knight DA. T-­cell activation response
16. Stern M, Hirsch H, Cusini A, et al. Cytomegalovirus serology and to allogeneic CMV-­infected endothelial cells is not prevented
replication remain associated with solid organ graft rejection and by ganciclovir or foscarnet: implications for transplant vascu-
graft loss in the era of prophylactic treatment. Transplantation. lar sclerosis. Transplantation. 2002;73(2):314-­318. https://doi.
2014;98(9):1013-­1018. https://doi.org/10.1097/TP.00000​0 0000​ org/10.1097/00007​890-­20020​1270-­0 0032
000160 30. Gatault P, Al-­Hajj S, Noble J, et al. CMV-­infected kidney grafts
17. Berry GJ, Burke MM, Andersen C, et al. The 2013 International drive the expansion of blood-­borne CMV-­specific T cells re-
Society for Heart and Lung Transplantation Working stricted by shared class I HLA molecules via presentation on donor
Formulation for the standardization of nomenclature in the cells. Am J Transplant Off J Am Soc Transplant Am Soc Transpl Surg.
pathologic diagnosis of antibody-­m ediated rejection in heart 2018;18(8):1904-­1913. https://doi.org/10.1111/ajt.14672
transplantation. J Heart Lung Transplant Off Publ Int Soc Heart 31. Borchers AT, Perez R, Kaysen G, Ansari AA, Gershwin ME. Role of
Transplant. 2013;32(12):1147-­1162. https://doi.org/10.1016/j. cytomegalovirus infection in allograft rejection: a review of pos-
healun.2013.08.011 sible mechanisms. Transpl Immunol. 1999;7(2):75-­82. https://doi.
18. Stewart S, Winters GL, Fishbein MC, et al. Revision of the 1990 org/10.1016/s0966​-­3274(99)80023​-­9
working formulation for the standardization of nomenclature in 32. Chanouzas D, Sagmeister M, Dyall L, et al. The host cellular immune
the diagnosis of heart rejection. J Heart Lung Transplant Off Publ response to cytomegalovirus targets the endothelium and is associ-
Int Soc Heart Transplant. 2005;24(11):1710-­1720. https://doi. ated with increased arterial stiffness in ANCA-­associated vasculitis.
org/10.1016/j.healun.2005.03.019 Arthritis Res Ther. 2018;20(1):194. https://doi.org/10.1186/s1307​
19. Mehra MR, Crespo-­L eiro MG, Dipchand A, et al. International 5-­018-­1695-­8
Society for Heart and Lung Transplantation working formula-
tion of a standardized nomenclature for cardiac allograft vas-
culopathy-­2010. J Heart Lung Transplant Off Publ Int Soc Heart S U P P O R T I N G I N FO R M AT I O N
Transplant. 2010;29(7):717-­727. https://doi.org/10.1016/j.
Additional supporting information may be found online in the
healun.2010.05.017
20. Costanzo MR, Costanzo MR, Dipchand A, et al. The International
Supporting Information section.
Society of Heart and Lung Transplantation Guidelines for the care
of heart transplant recipients. J Heart Lung Transplant Off Publ Int Soc
Heart Transplant. 2010;29(8):914-­956. https://doi.org/10.1016/j. How to cite this article: Boutolleau D, Coutance G, Désiré E,
healun.2010.05.034 et al. Association between cytomegalovirus infection and
21. Coutance G, d’Orio V, Belin L, et al. Favorable outcome of an exclu- allograft rejection in a large contemporary cohort of heart
sively posttransplant prophylactic strategy after heart transplan-
transplant recipients. Transpl Infect Dis. 2021;00:e13569.
tation in recipients with high immunological risk. Transplantation.
2019;103(7):1439-­1449. https://doi.org/10.1097/TP.00000​0 0000​ https://doi.org/10.1111/tid.13569
002503