Polymorphisms in the lectin pathway of complement activation influence the incidence of acute rejection and graft outcome after kidney transplantation

(1)

Article

Reference

Polymorphisms in the lectin pathway of complement activation influence the incidence of acute rejection and graft outcome after

kidney transplantation

GOLSHAYAN, Déla, et al . & Swiss Transplant Cohort Study

Abstract

There are conflicting data on the role of the lectin pathway of complement activation and its recognition molecules in acute rejection and outcome after transplantation. To help resolve this we analyzed polymorphisms and serum levels of lectin pathway components in 710 consecutive kidney transplant recipients enrolled in the nationwide Swiss Transplant Cohort Study, together with all biopsy-proven rejection episodes and 1-year graft and patient survival.

Functional mannose-binding lectin (MBL) levels were determined in serum samples, and previously described MBL2, ficolin 2, and MBL-associated serine protease 2 polymorphisms were genotyped. Low MBL serum levels and deficient MBL2 diplotypes were associated with a higher incidence of acute cellular rejection during the first year, in particular in recipients of deceased-donor kidneys. This association remained significant (hazard ratio 1.75, 95%

confidence interval 1.18-2.60) in a Cox regression model after adjustment for relevant covariates. In contrast, there was no significant association with rates of antibody-mediated rejection, patient death, early graft dysfunction [...]

GOLSHAYAN, Déla, et al . & Swiss Transplant Cohort Study. Polymorphisms in the lectin pathway of complement activation influence the incidence of acute rejection and graft outcome after kidney transplantation. Kidney International , 2016, vol. 89, no. 4, p. 927-938

PMID : 26924055

DOI : 10.1016/j.kint.2015.11.025

Available at:

http://archive-ouverte.unige.ch/unige:89765

Disclaimer: layout of this document may differ from the published version.

(2)

Polymorphisms in the lectin pathway of

complement activation in ﬂ uence the incidence of acute rejection and graft outcome after

kidney transplantation

De´la Golshayan

^1,2

, Agnieszka Wo´jtowicz

³

, Ste´phanie Bibert

³

, Nitisha Pyndiah

³

, Oriol Manuel

^1,3

, Isabelle Binet

⁴

, Leo H. Buhler

⁵

, Uyen Huynh-Do

⁶

, Thomas Mueller

⁷

, Ju¨rg Steiger

⁸

, Manuel Pascual

^1,2

, Pascal Meylan

^1,3,9

and Pierre-Yves Bochud

^3,9

; and the Swiss Transplant Cohort Study

¹⁰

1Transplantation Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland;²Transplantation Immunopathology Laboratory, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland;³Service of Infectious Diseases, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland;⁴Nephrologie und Transplantationsmedizin, Kantonsspital St. Gallen, St. Gallen, Switzerland;⁵Centre Universitaire Romand de Transplantation, Hôpitaux Universitaires de Genève, Geneva, Switzerland;⁶Department of Nephrology and Hypertension, Inselspital Bern, Bern, Switzerland;

7Department of Nephrology, University Hospital Zurich, Zurich, Switzerland; and⁸Transplantation Immunology and Nephrology, University Hospital Basel, Basel, Switzerland

There are conﬂicting data on the role of the lectin pathway of complement activation and its recognition molecules in acute rejection and outcome after transplantation. To help resolve this we analyzed polymorphisms and serum levels of lectin pathway components in 710 consecutive kidney transplant recipients enrolled in the nationwide Swiss Transplant Cohort Study, together with all biopsy-proven rejection episodes and 1-year graft and patient survival.

Functional mannose-binding lectin (MBL) levels were determined in serum samples, and previously described MBL2,ficolin 2, andMBL-associated serine protease 2 polymorphisms were genotyped. Low MBL serum levels and deficientMBL2diplotypes were associated with a higher incidence of acute cellular rejection during thefirst year, in particular in recipients of deceased-donor kidneys.

This association remained significant (hazard ratio 1.75, 95% confidence interval 1.18–2.60) in a Cox regression model after adjustment for relevant covariates. In contrast, there was no significant association with rates of antibody- mediated rejection, patient death, early graft dysfunction or loss. Thus, results in a prospective multicenter

contemporary cohort suggest thatMBL2polymorphisms result in low MBL serum levels and are associated with acute cellular rejection after kidney transplantation. Since MBL deficiency is a relatively frequent trait in the normal population, ourfindings may lead to individual risk stratification and customized immunosuppression.

Kidney International(2016)89,927–938;http://dx.doi.org/10.1016/

j.kint.2015.11.025

KEYWORDS: acute rejection; cohort study; innate immunity; kidney transplantation; mannose-binding lectin

ª2016 International Society of Nephrology

T

he innate immune system is primarily involved in nat- ural immunity against pathogens and tissue injury. In the setting of solid-organ transplantation (SOT), experimental data have highlighted an important role of innate immune activation early after transplantation, in particular following donor brain death and during ischemia–reperfusion injury in the early stages after surgery.¹^–³Macrophages and dendritic cells act as immune sentinels in peripheral tissues and secondary lymphoid organs, regularly sampling and processing self-antigens to maintain immune homeostasis in steady-state conditions. These innate immune cells express pattern recognition receptors that specifically detect conserved pathogen-associated or danger-associated molec- ular patterns, in the context of local infection or sterile tissue damage, respectively. The activation of surface pattern recognition receptors results in the maturation of the cells into competent antigen-presenting cells that produce proin- flammatory mediators and can initiate an adaptive immune response.⁴^–⁶ Besides surface-bound receptors, soluble pattern recognition receptors have been characterized, including complement-activating mannose-binding lectin (MBL) andficolins (FCNs).⁷Upon ligation, the downstream MBL-associated serine protease (MASP) signaling activates the complement cascade.

MBL and FCN are mainly synthesized in the liver and function as oligomeric molecules assembled with MASP. In circulation, these oligomeric structures bind to surface car- bohydrates of microbes and injured cells, leading to C3b- mediated opsonization and phagocytosis. Single-nucleotide polymorphisms (SNPs) in the genes encoding the lectin pathway proteins determine their functional activity and

Correspondence:Déla Golshayan, Transplantation Center, CHUV, Bugnon 46, 1011 Lausanne, Switzerland. E-mail:dela.golshayan@chuv.ch

9Co-senior authors.

10SeeAppendixfor the members of the Swiss Transplant Cohort Study.

Received 20 June 2015; revised 26 September 2015; accepted 12 November 2015; published online 3 February 2016

(3)

serum levels.⁸ Due to prevalent genetic polymorphisms in both regulatory and coding sequences ofMBL2, up to 25% of the normal population have reduced serum concentrations of MBL and can be classified as functionally deficient.⁹ However, MBL functions may be redundant in healthy individuals, as low levels of functional MBL have been associated with susceptibility to infections mainly in immu- nocompromised individuals.^10–13 As opposed to infectious settings, MBL deficiency has been described to be protective in some glomerular diseases such as IgA nephropathy,^14,15as well as in studies of kidney and heart transplantation where patients with low functional MBL serum levels had a better graft survival.¹⁶^–¹⁹ Moreover, complement deficiency or blockade was shown to limit postischemic reperfusion injury in various organs, and protect against acute renal transplant rejection.²⁰^–²⁴

Conﬂicting data exist on the deleterious versus protective role of the lectin pathway of complement activation in graft rejection, as well as in patient and graft outcomes after SOT.^25–28 Many of the available studies in kidney transplantation are either retrospective, involve small numbers of patients, or have analyzed outcomes under previous eras of immunosuppression and anti-infectious therapies. Therefore, the aim of the present study was to investigate the role of the lectin pathway of complement activation and the possible association between recipient genotype and graft outcomes after kidney transplantation in a large, well-characterized, ongoing prospective cohort. The primary end point was the incidence of biopsy-proven acute rejection episodes at 1 year after transplantation; secondary end points were death- censored graft loss, graft function at 1 year, and patient survival.

RESULTS

Study population

Donor, recipient, and transplant baseline characteristics are described inTable 1. Among the 710 consecutive Caucasian kidney transplant recipients included in the Swiss Transplant Cohort Study (STCS), 672 (94.6%) underwent a kidney-alone transplantation and 38 also received another organ, mainly pancreas (27/38). One hundred twenty-four (17.5%) patients had a previous transplant, of whom 111 (15.6%) were undergoing kidney transplantation for a second or more time.

Deceased donation provided the majority (59.4%) of the organs, with a median deceased donor age of 53 years and median cold ischemia time of 6.5 hours.

Patients were given induction and initial maintenance immunosuppression (prescribed within the ﬁrst week after transplantation) based on international and local guidelines, according to their immunologic risk status. Basiliximab was used in 62.1% of patients, 18.3% received anti-thymocyte globulin, and in 19.6% no induction was prescribed. Ste- roids and mycophenolic acid–based agents were used in

>86% of the patients. Calcineurin inhibitors were started early in 76.7% of the recipients, with a large preference for tacrolimus (FK) as compared to cyclosporine.

Serum MBL levels in kidney transplant recipients correlated with genotypes

The overall median MBL2 serum level measured on the day of transplantation was 678.2 ng/ml (Figure 1). Patients were classified as having“high”(>median) and“low”(#median) MBL levels. The latter group was further stratified into “intermediate” ($10 ng/ml) or “deficient” (<10 ng/ml) MBL levels. As shown in Figure 1, MBL serum levels were largely dependent on the MBL2 genotype in our study population.

The YAYA and YAXA diplotypes resulted in the higher serum MBL levels and were the main genotypes found in our population (total 401 patients, 56.6%) as compared to the remaining XAXA, YAYO, and XAYO (total 266 patients, 37.6%) and YOYO (total 41 patients, 5.8%) diplotypes, which were associated with intermediate and deﬁcient MBL serum levels, Table 1 | Characteristics of the study population at the time of transplantation

Kidney transplant recipients N[710

N (% or IQR)

Donor age (median years; IQR) 53 21.0

Donor gender^aM/F 345/362 48.8/51.2

Recipient age (median years; IQR) 54 20.0

Recipient gender M/F 473/237 66.7/33.4

Type of donor

Living 288 40.6

Deceased 422 59.4

Cold ischemia time^b (median hours; IQR)

6.5 9.2

Multiorgan transplantation 38 5.3

HLA mismatch (>3)^c 292 52.9

HLA-A mismatches^d(0/1/2) 99/326/279 14.1/46.3/39.6 HLA-B mismatches^e(0/1/2) 56/297/351 7.9/41.8/49.9 HLA-DR mismatches^f(0/1/2) 81/358/266 11.5/50.8/37.7 Donor/recipient CMV serostatus^g

D–R– 172 24.6

DþR– 139 19.9

D–Rþ 156 22.3

DþRþ 233 33.3

Induction therapy^h

None 131 19.6

Basiliximab only 414 62.1

Anti-thymocyte globulin 122 18.3

Initial maintenance immunosuppression

Steroids 635 89.4

Tacrolimus/cyclosporine 396/143 56.6/20.1

MPA agents/AZA 614/2 86.5/0.3

mTOR inhibitors 2 0.3

MBL2 levelⁱ(median; IQR) 678.2 1254.9

AZA, azathioprine; CMV, cytomegalovirus; D, donor; F, female; HLA, human leukocyte antigen; IQR, interquartile range; M, male; MBL, mannose-binding lectin; MPA, mycophenolic acid; mTOR, mammalian target of rapamycin; R, recipient.

aData were missing for 3 patients.

bConsidered for deceased donors only.

cData were missing for 158 patients.

dData were missing for 6 patients.

eData were missing for 6 patients.

fData were missing for 5 patients.

gData were missing for 10 patients.

hData were missing for 43 patients.

iData were missing for 1 patient.

(4)

respectively. These data obtained from pretransplant end-stage renal disease patients were in accordance with previous reports in the general population, describing the correlation between the selected SNPs and functional MBL serum levels.⁸

Polymorphisms inMBL2resulting in low MBL serum levels are associated with an increased risk of acute cellular rejection in theﬁrst year after transplantation

The 12-month cumulative incidence of biopsy-proven acute cellular rejection (ACR) and antibody-mediated rejection (AMR) was 0.20 and 0.05, respectively. A total of 144 ACR episodes were recorded during the ﬁrst year, of which 122 had occurred during theﬁrst 6 months after transplantation.

Among the cohort of patients (n ¼ 473) who reached the 3-year follow-up, the incidence of ACR declined over time, with only 16 new ACR episodes reported after theﬁrst year.

Within the ACR episodes during thefirst year, 7 were histo- logically classified as borderline tubulitis. During thefirst year after transplantation, 32 AMR episodes occurred in our study population. Overall, little modification was made in maintenance immunosuppression during the first year as compared to initial therapy. The majority of patients were still on a combination therapy of calcineurin inhibitors (97%, including 76% under FK) and mycophenolic acid–based agents (87%), and 73% received steroids at the 12-month follow-up. The same trend was observed at subsequent yearly follow-up analysis, with, interestingly, still 58% and 50% of patients maintained on steroid treatment at 2 and 3 years after transplantation, respectively.

Low MBL serum levels (# median) were signiﬁcantly associated with the risk of developing ACR during the ﬁrst year after transplantation (log-rank test¼ 0.01, Figure 2a).

The effect of MBL serum levels on ACR episodes was confirmed when analyzing the data based onMBL2genotypes (Figure 2b). This was even more striking when we stratified into“high”versus“intermediate”and“deficient”MBL serum levels (Figure 2c) and corresponding “YAYA” or “YAXA”

versus“XAYO,” “YAYO,” or“XAXA” and“YOYO” diplotypes (Figure 2d). In univariate analysis, the other factors negatively inﬂuencing the risk of ACR included donor age and donor–

recipient human leukocyte antigen (HLA) mismatches (Table 2). Living donation and induction (predominantly anti-thymocyte globulin) as well as steroids and tacrolimus maintenance treatment tended to be protective factors. There was no signiﬁcant association with donor or recipient gender, cold ischemia time (considered for deceased donors only), cytomegalovirus, or BK-related events. A multivariable analysis was performed to account for patient baseline characteristics in MBL serum levels and genotypes, and to correct for other factors including immunosuppressive treatments.

This analysis conﬁrmed that the low MBL status was an independent risk factor for ACR both in stepwise Cox model (hazard ratio [HR] 1.75, 95% conﬁdence interval [CI]

1.18–2.60,P¼0.006;Table 2) and logistic regression model (odds ratio 1.84, 95% CI 1.14–2.98,P¼0.01;Supplementary Table S1online). Besides low MBL levels, the other factor that remained associated with ACR in the multivariable Cox analysis was donor–recipient HLA mismatches (HR 1.79, 95% CI 1.20–2.66,P¼0.004), a risk factor also identiﬁed in the logistic regression analysis together with donor age (odds ratio 1.21, 95% CI 1.02–1.42,P¼0.02) and being on steroids (odds ratio 2.78, 95% CI 1.53–5.06, P < 0.001) (Supplementary Table S1). Data analysis excluding the sub- group of patients with borderline tubulitis yielded similar results (Supplementary Figure S1,Supplementary Tables S2 and S3). There was no association between either MBL serum levels or MBL2 diplotypes and the cumulative incidence of AMR during theﬁrst year after transplantation (data not shown).

Ficolins and MASP are other important components of the lectin pathway, and their serum concentrations were previously shown to correspond to polymorphisms in the respective FCN2 and MASP2 genes.²⁹ We did not ﬁnd any signiﬁcant association between the investigated SNPs and ACR in our study population, except for a trend toward a deleterious effect of theFCN2rs7851696 SNP (HR 1.39, 95%

CI 0.97–2.00, P ¼ 0.07). This, however, did not reach statistical signiﬁcance in the multivariable analysis.

Patient and early graft outcome after kidney transplantation From our starting cohort of 710 patients, 666 patients were followed for at least 1 year, 638 patients for up to 24 months after transplantation, and 473 patients reached the 3-year follow-up time. At 3 years, a total of 22 patients had lost their kidney graft (13 during theﬁrst year), and 42 died with a functioning graft (23 during theﬁrst year).Table 3 summa- rizes the risk factors for death at 1 year after transplantation.

The MBL status at the time of transplantation did not have a Figure 1 | MBL2 serum levels according toMBL2diplotypes

among 710 kidney transplant recipients.The dashed line stands for the overall median MBL2 serum level (678 ng/ml). Patients receiving multiple organs were included except 10 kidney/liver recipients. MBL2 genotyping was missing in 2 of 710 patients.

Correlation between MBL2 levels and MBL2 diplotypes was measured by analysis of variance and non-parametric Kruskal–Wallis rank test (P<0.001).

(5)

signiﬁcant impact on short-term patient survival, norFCN2 and MASP2 genotypes. In the multivariable analysis, donor male gender was signiﬁcantly associated with death (HR 3.76, 95% CI 1.24–11.4,P¼0.02) as well as recipient age (HR 2.22, 95% CI 1.42–3.47, P < 0.001), while immunosuppressive regimens with calcineurin inhibitors or mycophenolic acid–

based agents were protective. We next analyzed death- censored graft loss after transplantation (Table 4) and found a strong association with rejection episodes during the ﬁrst year (HR 6.21, 95% CI 2.71–14.2,P<0.002), as well as with recipient age (HR 2.04, 95% CI 1.48–2.81,P<0.001), and a trend toward association with the MASP2 rs12085877 SNP (HR 2.90, 95% CI 0.85–9.95,P¼0.09), but not withMBL2 orFCN2polymorphisms. Overall, while the low MBL status was associated with ACR, there was no direct association with patient death or graft loss at 1 year in the uni- and multivariable analysis. Analysis of the cohort of patients that reached 3 years follow-up corroborated the results on patient and graft survival obtained at 1 year (data not shown).

Because graft function at 1 year is an important predictor of long-term graft and patient survival after kidney

transplantation,^30,31we analyzed renal function at 1 year and stratiﬁed the patients according to estimated glomerular ﬁltration rate values, setting a cutoff value of 40 ml/min per 1.73 m²to account for moderate to severe graft dysfunction.

As shown inTable 5, graft dysfunction was strongly associated with rejection episodes (ACR and AMR) during theﬁrst year (HR 4.96, 95% CI 2.76–8.91, P < 0.001), as well as with donor age (HR 2.04, 95% CI 1.59–2.61, P < 0.001), and borderline associated with cyclosporine-based maintenance immunosuppression (HR 2.04, 95% CI 1.00–4.19, P ¼ 0.051). The MBL status did not directly affect graft function at 1 year in our study.

DISCUSSION

Over the past years, inconsistent experimental and clinical data have been reported regarding the role of the lectin pathway of complement activation in graft and recipient outcomes after SOT. Using the large prospective multicenter STCS database, we investigated the associations between previously described SNPs in the lectin pathway of complement activation and outcome after kidney transplantation.

Figure 2 | One-year cumulative incidence of acute cellular rejection (ACR) according to MBL2 levels andMBL2diplotypes in 710 kidney transplant recipients.The top panels show stratification into 2 groups, according to MBL2 serum levels (across median,a) orMBL2diplotypes (high-level diplotypes“YAYA”or“YAXA”vs. low-level diplotypes“XAXA,” “YAYO,” “XAYO,”or“YOYO,”b). In order to account for MBL2 deficiency, the 2 lower panels show stratification into 3 groups, according to MBL2 serum levels (“high”vs.“intermediate”and“deficient”levels,c) and diplotypes (high-level“YAYA”or“YAXA”vs. intermediate“XAYO,” “YAYO,”or“XAXA”and deficient“YOYO”diplotypes).Pvalues stand for the log-rank test. Patients receiving multiple organs were included except 10 kidney/liver recipients. Mannose-binding lectin (MBL) genotyping was missing in 2 of 710 patients, and MBL levels were missing in 1 of 710 patients.

(6)

Our main ﬁnding was that MBL2 genotypes resulting in low MBL serum levels strongly correlated with the incidence of ACR in theﬁrst year after transplantation. This effect was still observed when analyzing the data at 36 months (data not shown) and was a more potent predictor of ACR

than differences in immunosuppressive regimens or other traditional risk factors as documented in multivariable analyses. Interestingly, when stratifying the 144 ACR episodes during the ﬁrst year according to histologic severity (70 tubulointerstitial/Banff I, 74 vascular/Banff II and III), Table 2 | Risk factors for acute cellular rejection at 1 year after transplantation

Variable

Univariate model Stratiﬁed multivariable model^f

HR (95% CI) P HR (95% CI) P

Donor age ($median) 1.26 (0.91–1.75) 0.17

Donor male gender 1.15 (0.83–1.60) 0.39

Recipient age ($median) 0.88 (0.63–1.22) 0.43

Recipient male gender 1.10 (0.77–1.56) 0.61

Living donor 0.70 (0.50–0.99) 0.04

Cold ischemia time^a(>median) 1.14 (0.79–1.65) 0.49

HLA mismatch (>3) 1.61 (1.09–2.37) 0.02 1.79 (1.20–2.66) 0.004

Multiorgan transplantation 0.87 (0.41–1.87) 0.73

Donor/recipient CMV serostatus

D–R– reference

DþR– 0.89 (0.56–1.43) 0.64

D–Rþ 0.77 (0.48–1.23) 0.28

DþRþ 0.72 (0.47–1.10) 0.13

CMV viremia/disease 0.99 (0.59–1.67) 0.98

BK viremia/disease 0.75 (0.34–1.64) 0.47

Induction therapy

None reference

Basiliximab only 0.73 (0.49–1.08) 0.12

Anti-thymocyte globulin 0.44 (0.24–0.79) 0.006

Maintenance immunosuppression^b

Steroids 0.68 (0.40–1.15) 0.15

Cyclosporine 1.24 (0.87–1.79) 0.24

Tacrolimus 0.78 (0.55–1.11) 0.17

MPA agents 1.51 (0.74–3.09) 0.26

mTOR inhibitors 0.69 (0.25–1.86) 0.45

MBL2 level

High (>678 ng/ml) reference

Low (#678 ng/ml) 1.54 (1.10–2.14) 0.01 1.75 (1.18–2.60) 0.006

MBL2 diplotypes^c

YAYA or YAXA reference

XAYO, YAYO, or XAXA 1.25 (0.88–1.76) 0.21

YOYO 2.05 (1.16–3.64) 0.01^d

FCN2 genotypes

rs7851696 TT or TG versus GG 1.39 (0.97–2.00) 0.07^e

rs17514136 GG or GA versus AA 1.22 (0.88–1.69) 0.24

rs3124953 AA or AG versus GG 1.01 (0.72–1.41) 0.96

MASP2 genotypes

rs12085877 TT or TC versus CC 1.14 (0.56–2.33) 0.72

rs1033638 AA versus GA or GG 1.03 (0.72–1.45) 0.89

CI, conﬁdence interval; CMV, cytomegalovirus; D, donor; FCN,ﬁcolin; HLA, human leukocyte antigen; HR, hazard ratio; MASP, mannose-binding lectin–associated serine protease; MBL, mannose-binding lectin; MPA, mycophenolic acid; mTOR; mammalian target of rapamycin; R, recipient.

BoldPvalues are signiﬁcant.

aConsidered for deceased donors only.

bTime-dependent covariates.

cMBL genotype O consists of rare allele D (R52C, rs5030737), B (G54D, rs1800450), and C (G57E, rs1800451), respectively, and the wild-type allele of each is referred to as genotype A. MBL genotype Y refers to presence of rare variant C of promoter SNP G221C (rs7096206) and genotype X as wild-type allele G of that polymorphism.

dPvalue after correction for multiple testing (N¼10) is 0.1.

ePvalue after correction for multiple testing (N¼10) is 0.7.

fMultivariable analysis was performed by using stepwise regression model. The proportional hazards assumption was tested by using the stphtest command implemented in Stata. No deviation from this assumption was observed for MBLs. However, since 2 covariables in the multivariable model did not verify this hypothesis (basiliximab, P¼0.0001; steroids,P¼0.05), those were accounted for by using a stratiﬁed multivariable Cox model. Final global proportional hazards assumption testPvalue for multivariable Cox model was 0.84. In addition we have performed univariate and multivariable stepwise logistic regression analysis for ACR at 1 year after transplantation to account for those variables (Supplementary Table S1).

(7)

a significant correlation was found only between low MBL status and tubulointerstitial ACR (Supplementary Figures S2 and S3). Besides low MBL serum levels, the other significant independent risk factors for ACR were donor age and HLA mismatches, as previously described.³² Registry data have highlighted the efficacy of current immunosuppressive

regimens in preventing acute rejection, in particular ACR, at least in low-immunologic-risk recipients.^33–35 Interestingly, these immunosuppressive treatments were developed to target adaptive immune responses, but they only marginally affect innate immunity. Thus, the potential protective role of MBL in preventing ACR may have been better unraveled when Table 3 | Predictors of patients’death at 1 year after transplantation

Variable

Univariate model Stratiﬁed multivariable model^d

HR (95% CI) P HR (95% CI) P

Donor age (per 10-year increase) 1.25 (0.94–1.67) 0.12

Donor male gender 1.62 (0.70–3.74) 0.26 3.76 (1.24–11.4) 0.02

Recipient age (per 10-year increase) 1.71 (1.18–2.50) 0.005 2.22 (1.42–3.47) <0.001

Living donor 0.22 (0.06–0.73) 0.01

HLA mismatch (>3) 0.54 (0.20–1.50) 0.24

Rejection episodes (regrouping ACR and AMR) 3.37 (1.46–7.75) 0.004 Donor/recipient CMV serostatus

D–R– reference

DþR– 1.53 (0.47–5.01) 0.48

D–Rþ 0.91 (0.24–3.39) 0.89

DþRþ 1.23 (0.40–4.91) 0.59

Induction therapy

None reference

Steroids 0.07 (0.03–0.17) <0.001

Cyclosporine 0.33 (0.08–1.40) 0.13 0.09 (0.02–0.50) 0.006

Tacrolimus 0.12 (0.05–0.31) <0.001 0.08 (0.02–0.29) <0.001

MPA agents 0.03 (0.01–0.08) <0.001 0.09 (0.02–0.28) <0.001

mTOR inhibitors 0.00 (0.00–) 1.00

MBL2 level

Low (#678 ng/ml) 0.90 (0.40–2.04) 0.80

MBL2 diplotypes^c

XAYO, YAYO, or XAXA 0.64 (0.26–1.54) 0.32

YOYO 0.00 (0.00–) 1.00

FCN2 genotypes

rs7851696 TT or TG versus GG 0.50 (0.15–1.69) 0.26

MASP2 genotypes

rs12085877 TT or TC versus CC 0.00 (0.00–) 1.00

ACR, acute cellular rejection; AMR, antibody-mediated rejection; CI, conﬁdence interval; CMV, cytomegalovirus; FCN,ﬁcolin; HLA, human leukocyte antigen; HR, hazard ratio;

MASP, mannose-binding lectin–associated serine protease; MBL, mannose-binding lectin; MPA, mycophenolic acid; mTOR; mammalian target of rapamycin; R, recipient.

dMultivariable analysis was performed by using stepwise regression model. The proportional hazards assumption was tested by using the stphtest command implemented in Stata. No deviation from this assumption was observed for MBLs. However, since 2 covariables in the multivariable model did not verify this hypothesis (rejection episodes, P¼0.003; steroids,P¼0.04), those were accounted for by using a stratiﬁed multivariable Cox model. Final global proportional hazards assumption testPvalue for multivariable Cox model was 0.56.

(8)

adaptive immunity was under check by current potent immunosuppressive regimens such as in the present study.

Previous reports either did notfind a significant association between MBL status and acute rejection³⁶or concentrated on long-term graft and patient outcomes.^16,17,26 The finding of

robust and clear-cut association between low MBL status and increased incidence of ACR seen in our study is likely due to size and homogeneity of the prospectively followed cohort being treated with recent-era immunosuppressive protocols.

Moreover, the relatively homogeneous drug prescription, in Table 4 | Predictors of death-censored graft loss at 1 year after transplantation

Variable

Univariate model Multivariable model^f

HR (95% CI) P HR (95% CI) P

Donor age (per 10-year increase) 1.25 (0.96–1.61) 0.09

Recipient age (per 10-year increase) 1.60 (1.16–2.20) 0.004 2.04 (1.48–2.81) <0.001

Living donor 0.23 (0.08–0.66) 0.006

HLA mismatch (>3) 1.11 (0.46–2.67) 0.82

Rejection episodes (regrouping ACR and AMR) 4.14 (1.89–9.07) <0.001 6.21 (2.71–14.2) <0.002 Donor/recipient CMV serostatus

D–R– reference

DþR– 1.45 (0.52–3.99) 0.47

D–Rþ 0.80 (0.25–2.51) 0.70

DþRþ 0.97 (0.36–2.60) 0.95

Induction therapy

None reference

Steroids 0.12 (0.05–0.25) <0.001 0.10 (0.05–0.23) <0.001

Cyclosporine 0.54 (0.19–1.55) 0.25

Tacrolimus 0.21 (0.10–0.45) <0.001 0.30 (0.13–0.70) 0.005

MPA agents 0.05 (0.03–0.12) <0.001 0.09 (0.04–0.20) <0.001

mTOR inhibitors 0.00 (0.00–) 1.00

MBL2 level

Low (#678 ng/ml) 0.86 (0.41–1.81) 0.69

MBL2 diplotypes^c

XAYO, YAYO, or XAXA 0.62 (0.27–1.42) 0.26

YOYO 1.02 (0.24–4.39) 0.97

FCN2 genotypes

MASP2 genotypes

rs12085877 TT or TC versus CC 2.24 (0.68–7.40) 0.19 2.90 (0.85–9.95) 0.09^e

rs1033638 AA versus GA or GG 0.42 (0.16–1.09) 0.07^d

ACR, acute cellular rejection; AMR, antibody-mediated rejection; CI, conﬁdence interval; CMV, cytomegalovirus; D, donor; FCN,ﬁcolin; HLA, human leukocyte antigen;

HR, hazard ratio; MASP, mannose-binding lectin–associated serine protease; MBL, mannose-binding lectin; MPA, mycophenolic acid; mTOR; mammalian target of rapamycin;

R, recipient.

fMultivariable analysis was performed by using stepwise regression model. The proportional hazards assumption was tested by using the stphtest command implemented in Stata. In this model no violation from the proportional hazards assumption was observed. Final global proportional hazards assumption testPvalue for multivariable Cox model was 0.39.

(9)

particular regarding maintenance treatment during the ﬁrst year, may have unmasked the effects of innate immunity, such as the MBL pathway of the complement system.

The lectin pathway of complement activation plays an important role in host defenses against microorganisms, and

unambiguous data exist on the harmful role of low MBL levels in the incidence and severity of bacterial, viral, and fungal infections after transplantation.^13,37,38 Indeed, local complement activation via the lectin pathway leads to the production of C3b that mediates opsonization and Table 5 | Predictors of graft dysfunction (estimated glomerularﬁltration rate£40 ml/min per 1.73 m²) at 1 year after transplantation

Univariate model Multivariable model

OR 95% CI P OR 95% CI P

Donor age (per 10-year increase) 1.99 (1.63–2.42) <0.001 2.04 (1.59–2.61) <0.001

Recipient age (per 10-year increase) 1.17 (1.01–1.36) 0.04

Living donor 0.45 (0.29–0.71) 0.0006

HLA mismatch (>3) 1.56 (0.94–2.59) 0.09

Rejection episodes (regrouping ACR and AMR) 2.74 (1.76–4.26) <0.001 4.96 (2.76–8.91) <0.001

ACR 2.58 (1.61–4.11) <0.001

Donor/recipient CMV serostatus

D–R– reference

DþR– 1.21 (0.63–2.32) 0.57

D–Rþ 1.32 (0.72–2.45) 0.37

DþRþ 1.19 (0.67–2.12) 0.56

Induction therapy

None reference

Maintenance regimen^b

Steroids 1.75 (1.05–2.91) 0.03

Cyclosporine 1.55 (0.93–2.57) 0.09 2.04 (1.00–4.19) 0.051

MPA agents 0.40 (0.23–0.69) 0.001 0.22 (0.10–0.48) <0.001

Tacrolimus 0.66 (0.41–1.08) 0.10

mTOR inhibitors 2.60 (0.83–8.09) 0.10

MBL2 level

Low (#678 ng/ml) 1.06 (0.70–1.61) 0.80

MBL2 diplotypes^c YAYA or YAXA

XAYO, YAYO, or XAXA 1.17 (0.76–1.80) 0.48

YOYO 0.41 (0.12–1.39) 0.15

FCN2 genotypes

rs3124953 AA or AG versus GG 0.65 (0.41–1.01) 0.055^d 0.45 (0.24–0.84) 0.01^e

MASP2 genotypes

rs12085877 TT or TC versus CC 1.05 (0.42–2.65) 0.92

ACR, acute cellular rejection; AMR, antibody-mediated rejection; CI, conﬁdence interval; CMV, cytomegalovirus; D, donor; FCN,ﬁcolin; HLA, human leukocyte antigen; MASP, mannose-binding lectin–associated serine proteases; MBL, mannose-binding lectin; MPA, mycophenolic acid; mTOR; mammalian target of rapamycin; OR, odds ratio; R, recipient.

(10)

phagocytosis of microbes by macrophages and dendritic cells.

Experimental data have shown that the lectin pathway could also activate the complement cascade in the context of sterile tissue inflammation such as ischemia–reperfusion injury following SOT.^39,40 This leads to the formation of C3 convertase and downstream complement cascade activation, resulting in the local production of the terminal component C5b-C9 (membrane attack complex) that contributes to tissue damage. Accordingly, inhibition of MBL using a mono- clonal antibody resulted in reduced cardiac damage in a rat model of myocardial ischemia–reperfusion.⁴¹ Recently, data from experimental models of renal ischemia–reperfusion injury suggested an additional direct cytotoxic effect of circulating MBL on tubular epithelial cells.⁴²Thus, high MBL serum levels may be deleterious in the setting of severe ischemia–reperfusion injury early after SOT, as well as contribute to delayed graft function. Organs or tissues may, however have different susceptibilities to ischemia–reperfusion injury, therefore be differentially affected by MBL serum levels, in part explaining some discrepancies on the reported role of this pathway in various experimental and clinical settings and across organs. In our cohort, up to 40% of the kidneys came from living donors, and, even for deceased donors, the median cold ischemia time was relatively short (reflecting the small distances between centers in Switzerland). Unfortunately, we could not investigate delayed graft function appropriately using the STCS database. How- ever, the fact that our cohort possibly experienced less initial graft damage may in part explain why we did not find a significant association between MBL serum levels and graft dysfunction or loss at 1 year, as opposed to some studies which preponderantly included deceased donors.^16,17

Besides its role in host defenses against pathogens, MBL was shown to directly bind to late apoptotic and necrotic cells in vitro, facilitating phagocytosis by macrophages and dendritic cells.^43,44 These findings indicate a role of the lectin pathway in tissue homeostasis and immune tolerance. Low MBL serum levels have been associated with autoimmune diseases such as systemic lupus erythematosus or rheumatoid arthritis.^12,45–48 In a recent study in SOT, early surveillance biopsies performed at 3 to 6 months after kidney transplantation were analyzed for the degree of inflammation and cell death in the tubulointerstitial compartment. The authors concluded that low recipient serum MBL levels were associated with increased inflammation and apoptosis.⁴⁹Interest- ingly and in line with our data, the same authors had previously described that subclinical rejection after kidney transplantation was associated with low serum MBL levels.⁵⁰ Thus, depending on the microenvironment, MBL may exert a protective or a deleterious role, and this could partly explain the discrepancies observed in various experimental and clinical studies. By binding to injured tissues in the acute in- flammatory setting of ischemia–reperfusion injury after transplantation, MBL would aggravate tissue damage and potentiate alloantigen presentation. In the absence of overt inflammation, such as at later stages after transplantation,

MBL could contribute to the sequestration and clearance of graft-derived apoptotic cells by quiescent immature macrophages and dendritic cells, thus limiting host alloresponses.

The activation of the lectin pathway leads to the formation of C3 convertase and the downstream complement cascade, resulting in the production of membrane attack complex.

This pathway may thus be implicated in endothelial injury and AMR. However, we did notfind any association between MBL serum levels and AMR, but our study was limited by relatively low numbers of events during the first year. While the lectin pathway may contribute to its pathogenesis, the initiation of AMR is probably more dependent on the presence of circulating donor-specific alloantibodies and the activation of the classical pathway of complement.

The MBL status at the time of transplantation did not have a signiﬁcant direct impact on patient and graft 1- and 3-year survival as well as on 1-year graft function in our prospective study, nor FCN2 and MASP2 genotypes. In fact, acute rejection episodes and in particular ACR were the most important determinants of graft outcome after transplantation in our multivariable analysis (Tables 4 and 5, data not shown at 3 years). Recently, Bayet al.have used a lower median cutoff value (#454 ng/ml) and have shown that low serum MBL levels were associated with decreased 5-year death-censored graft survival.²⁶ Interestingly, the strongest association was seen in non-immunized patients receiving a kidney from a deceased donor.²⁶When stratifying our data, we also observed a strongest association between low MBL status and ACR in recipients of deceased as compared to living donors (Supplementary Figures S4 and S5). These results supported the concept of MBL being involved in the clearance of graft- derived damaged and dying cells, a process that would be more needed in the context of cadaveric organs and prolonged cold ischemia time to limit host alloreactivity. On the other hand, our results are in contradiction with studies from Berger et al., who found that low pretransplant MBL serum levels (<400 mg/ml) were protective for patient and graft survival.^16,17 However, these studies were performed in smaller cohorts and in previous eras using other immunosuppressive regimens. Moreover, with longer follow-up periods (>10 years in these studies), other non-immunologic confounders may have been involved. Indeed, low MBL serum levels were also shown in the general population to be associated with accel- erated atherosclerosis,⁵¹ a major risk factor for patient and graft loss after transplantation.

In conclusion, using the large and well-characterized contemporary prospective STCS cohort, we found that low MBL serum levels at the time of transplantation were associated with an increased risk of ACR after kidney transplantation, in particular in recipients of deceased donors. As acute rejection episodes were predictors of worse graft outcome after kidney transplantation, our ﬁndings are potentially important and they should be conﬁrmed in other prospective cohorts and studied for other solid organs.

MBL deﬁciency being a relatively frequent trait in the normal population, screening and MBL substitution could be

(11)

considered in the future as a preventive or therapeutic approach in relevant clinical settings after SOT.

MATERIALS AND METHODS Study design

The STCS is a prospective multicenter cohort including all SOT activity performed in Switzerland from May 2008 onward.⁵² The overall acceptance rate of participation in the STCS is 95%. For this study, we included all Caucasian consecutive kidney transplant recipients transplanted from May 2008 to March 2011. The STCS has been approved by the ethics committees of all centers, and all patients have given written informed consent for participation.

Data collection

Patient- and transplant-specific data were collected at baseline (day of transplantation), then prospectively at 6 and 12 months, and yearly after using standardized case report forms. Baseline donor and recipient data included gender, age, ethnicity, diagnostic of nephropathy, preexisting comorbidities, and HLA immunologic status. Transplant-related data included type of donor, cold ischemia time, and immunosuppressive induction treatment. Follow-up data included graft loss, patient outcome, weight, serum creatinine, immunosuppressive treatments, rejection episodes, and infectious and noninfectious events (mainly cardiovascular and oncologic). All biopsies (per protocol and per cause) were recorded in the STCS database and scored according to Banff’09 update classification⁵³by each center’s reference pathologist. For our study, we classified the rejection episodes as follows: biopsy-proven acute cellular rejection (ACR) regrouping borderline changes and acute T cell–mediated rejection, and acute and chronic active humoral rejection grouped as antibody-mediated rejection (AMR). Cytomegalovirus events were classified as either asymptomatic replication (viremia) or disease (replication with corresponding signs/symptoms).^54–56 Cytomega- lovirus antiviral strategies depended on donor–recipient serostatus but varied among centers, using either a 3- to 6-month prophylactic strategy or a preemptive polymerase chain reaction screening approach.⁵⁵Polyomavirus BK occurrences were classified as viremia or probable/proven disease (viremia>10,000 copies twice, biopsy- proven nephropathy, or both).

Study end points

The primary end point was the incidence of biopsy-proven acute rejection at 1 year after transplantation. As secondary end points, we assessed patient and death-censored graft loss (deﬁned as need for dialysis or retransplantation), and graft function at 1 year as estimated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula.

DNA extraction and genotyping

DNA was extracted from blood samples collected from each patient at the time of transplantation, as described.⁵⁷To limit the risk of population stratiﬁcation bias, genetic analyses were performed in Caucasian patients only. Candidate gene genotyping was performed using a custom-made genotyping platform (VeraCode GoldenGate Genotyping Assay on BeadXpress, Illumina, San Diego, CA) or competitive allele-speciﬁc PCR (KASP) technology (LGC Genomics, Teddington, UK).⁵⁷ Based on previous literature, the following 6 SNPs in the MBL2 gene were selected for analysis: rs5030737, rs11003125, rs1800450, rs1800451, rs7095891, and rs7096206. The following 4 SNPs were assessed for the FCN2 gene: rs7851696, rs17514136, rs3124953, and rs3124952; and 3 for theMASP2gene:

rs56392418, rs12085877, and rs1033638. To check for genotyping accuracy, Hardy–Weinberg equilibrium and pairwise linkage disequilibrium were calculated with the genhwi and the pwld pro- grams in Stata 13.1 software (StataCorp, College Station, TX). SNP frequencies and linkage disequilibrium were compared to those found in general Caucasian population (from public genetic data- bases such as HapMap Project or 1000 Genomes Project). MBL haplotypes were inferred using PHASE 2.1 software (University of Washington, Seattle, WA). MBL genotype O was deﬁned as presence of rare allele D (R52C, rs5030737), B (G54D, rs1800450), and C (G57E, rs1800451), respectively. Presence of the wild-type allele of each was referred to as genotype A. MBL genotype Y was deﬁned as presence of rare variant C of promoter SNP G221C (rs7096206) and genotype X as wild-type allele G of that polymorphism.

MBL serum levels

A serum sample was obtained on the day of transplantation from 710 consecutive Caucasian kidney transplant recipients enrolled in the STCS for the determination of functional MBL levels by enzyme- linked immunosorbent assay (MBL Oligomer ELISA Kit, Bioporto Diagnostics, Hellerup, Denmark).

Statistical analysis

Statistical analyses were performed by using Stata 13.1 software.

Associations were assessed by using the log-rank test and in multivariable stepwise Cox regression models for covariates that were independently associated with each end point, with censoring at death or lost to follow-up date at 12 or 36 months post-transplantation. The P value of 0.05 was considered to be significant. The proportional hazards assumption was tested by using stphtest and stcoxkm pro- grams in Stata 13.1 software afterfitting a Cox model. For variables that showed violation of the proportional hazards assumption, we performed a stratified Cox model by using the strata option in Stata.

The incidence (time toﬁrst rejection episode) and type of rejection were analyzed according to genotypes as categorical variables, as well as according to MBL functional levels as continuous variables. Cor- relation between MBL2 levels and MBL2 diplotypes was measured by a 1-way analysis of variance and non-parametric Kruskal–Wallis test.

Other previously described determinants of rejection and graft outcome were considered as relevant covariates in multivariable analysis. In particular, immunosuppressive drugs were considered in a time-dependent manner, accounting for all changes during follow-up.

DISCLOSURE

All the authors declared no competing interests.

ACKNOWLEDGMENTS

This study has been conducted within the framework of the STCS supported by the Swiss National Science Foundation (grant 33CS30- 148512) and the Swiss University Hospitals and Transplant Centres.

We thank all patients who participated in the STCS, the central and local data managers, as well as all the doctors, nurses, and investigators involved. We are grateful to Valentin Rousson (Social and Preventive Medicine, University of Lausanne) for statistical advice. DG and MP were supported by an unrestricted grant from Astellas. DG was supported by the Fondation Pierre Mercier pour La Science and Fondation Medi-CAL Futur. PYB was supported by the Swiss National Science Foundation (grant 324730-144054), the Leenaards Foundation, the Santos-Suarez Foundation, and the Loterie Romande. PYB is recipient of a Mérieux Research Grant. This project was also supported by a grant from the Emma Muschamp

Foundation and the Fondation Lausannoise pour la Transplantation d’Organes.