Presence of FoxP3⁺ Regulatory T Cells Predicts Outcome of Subclinical Rejection of Renal Allografts

Abstract

Subclinical rejection (SCR) of renal allografts refers to histologic patterns of acute rejection despite stable renal function. The clinical approach to SCR is controversial; it would be helpful to identify biomarkers that could determine whether the identified cellular infiltrates were detrimental. For investigation of whether the presence of FoxP3⁺ regulatory T cells (Treg) could help determine the functional importance of tubulointerstitial infiltrates observed in 6-mo protocol biopsies, 37 cases of SCR were evaluated. The presence of FoxP3⁺ Treg discriminated harmless from injurious infiltrates, evidenced by independently predicting better graft function 2 and 3 yr after transplantation. Furthermore, the FoxP3⁺ Treg/CD3⁺ T cell ratio positively correlated with graft function at 2 yr after transplantation, suggesting that an increasing proportion of Treg within the global T cell infiltrate may facilitate renal engraftment; therefore, immunostaining for FoxP3⁺ Treg in patients with SCR on protocol biopsies may ultimately be useful to identify patients who may require alterations in their immunosuppressive regimens.

Kidney transplantation is the elective treatment for all patients with ESRD.¹ Despite the significant improvement in the understanding of alloimmune mechanisms for graft injury and the appearance of new immunosuppressants, graft and patient survival have not increased as expected in the past decade; death with a functioning allograft as a result of cardiovascular disease and cancer, both aggravated by immunosuppressive drug adverse effects, and the development of interstitial fibrosis and tubular atrophy in the graft remain as the most relevant causes of graft loss.²

Actually, the assessment of allograft histology through prospective protocol biopsies was originally carried out for monitoring the advent of histopathologic lesions in clinically stable allografts³; however, although they have provided relevant insights regarding the natural history of histopathologic graft lesions over time, it still remains uncertain to what extent these lesions are specific or not for the antidonor alloimmune response. Notably, a main concern is the identification of histologic lesions of acute rejection in a high proportion of protocol biopsies (ranging from 15 up to 40%) in patients with well-functioning grafts during the first 6 mo after transplantation.^4–7 The presence of these asymptomatic tubulointerstitial cellular infiltrates has been defined as subclinical rejection (SCR).⁸ Currently, inconclusive results have been shown by several studies evaluating the predictive value of SCR on graft outcome.^5,9–12 Furthermore, this controversy is even more evident when the renal effect of treating or not treating SCR is analyzed^6,7,11,13; therefore, we lack reliable biomarkers to discriminate whether these cellular infiltrates represent formal injurious rejection or, conversely, are merely protective T cell infiltrates necessary for graft acceptance. Indeed, there is increasing evidence showing that immune responses are controlled by an antigen-specific T cell subset population with regulatory function (Treg) expressing the transcription factor FoxP3, which is the “master switch” for the development and function of Tregs.^14,15 These Treg have been shown to be capable of suppressing antidonor cytotoxic alloresponses. Moreover, some experimental and human studies have demonstrated that graft infiltration by these Treg is of relevance for graft acceptance.^16–21; therefore, we investigated whether the presence of FoxP3⁺ Treg within tubulointerstitial renal allograft infiltrates in patients with a diagnosis of SCR in protocol biopsies could be a useful functional biomarker for discriminating harmful cellular infiltrates from those playing an active role for graft acceptance.

RESULTS

Relevant demographic and clinical characteristics at the time of protocol biopsy are depicted in Table 1. Immunosuppression was analyzed taking into account whether they received induction therapy (Thymoglobulin [rATG; Genzyme, Madrid, Spain] to 18 patients and basiliximab [Simulect; Novartis, Basel, Switzerland] to five patients). The maintenance immunosuppressive regimen was based either on calcineurin inhibitor (CNI) or on the mammalian target of rapamycin inhibitor sirolimus (SRL; Rapamune; Wyeth, Madrid, Spain). Thymoglobulin was given as induction therapy to 13 patients who were on SRL and to three patients who were receiving a CNI as maintenance immunosuppressant. Basiliximab was given to one patient on SRL and to four patients who were on a CNI. At the time of the protocol biopsy (6 mo), patients were on either SRL or a CNI-based regimen. None was on both immunosuppressants. Three patients who were on SRL at 6 mo were switched to tacrolimus, and none from the CNI group was changed to SRL. At 3 yr, no patients had died and two had lost their graft at months 8 and 16 after transplantation, respectively.

Table 1.

Baseline demographic data and clinical characteristics at the time of protocol biopsy^a

Variable	Presence of FoxP3+ Treg(n = 25)	Absence of FoxP3+ Treg(n = 12)	P
Donor age (yr; mean ± SD)	39 ± 12	43 ± 15	NS
Donor gender (M/F)	17/8	8/4	NS
Recipient age (yr; mean ± SD)	45 ± 12	42 ± 17	NS
Recipient gender (M/F)	18/7	8/4	NS
Cause of ESRD			NS
glomerular	8	4
diabetes	1	0
unknown	10	2
APKD	1	2
interstitial	3	1
nephrosclerosis	2	3
No. of transplant (1/2)	23/11	2/1	NS
HLA mismatches A, B, DR (mean ± SD)	3.2 ± 0.8	2.9 ± 0.8	NS
Cold ischemia time (h; mean ± SD)	19.9 ± 4.5	23.6 ± 7.0	NS
DGF (no/yes)	17/8	7/5	NS
BPAR (no/yes)	16/9	10/2	NS
Immunosuppression
Induction therapy (no/yes; n = 23)	6/19	8/4	0.027
SRL based (no/yes; n = 14)	11/14	12/0	0.001
Non–SRL based (no/yes; n= 23)	15/11	3/12	NS
Serum creatinine (μmol/L; mean ± SD)	140 ± 74	138 ± 41	NS
eGFR (ml/min; mean ± SD)	60.9 ± 26.0	53.9 ± 24.0	NS
Proteinuria (g/24 h; mean ± SD)	0.55 ± 1.00	0.55 ± 1.10	NS
Acute Banff ’05 score (BLc/IA/IB)	15/9/1	9/2/1	NS
Chronic Banff ’05 score (0/I/II)	14/8/3	5/4/3	NS
CD3+ T cells/high-power field (mean ± SD)	12.9 ± 8.1	14.6 ± 10.0	NS

^aBPAR, biopsy-proven acute rejection; APKD, adult polycystic kidney disease; DGF, delayed graft function; MMF, mycophenolate mofetil; BLc, borderline changes.

Among the 37 patients, none had presence of C4d in the peritubular capillaries. Twelve (32.5%) did not show any evidence of FoxP3⁺ Treg, and 25 (67.5%) had presence of FoxP3⁺ Treg within the infiltrates in both the interstitium and the tubules. Double immunofluorescence labeling identified that the majority of FoxP3⁺ Treg were CD4⁺CD25⁺ and few were CD8⁺ T cells (Figure 1). The percentage of FoxP3⁺ Treg/CD3⁺ T cells per high-power field ranged from 0.7 to 52.0%.

Figure 1.

Representative tubulointerstitial cellular infiltrates of patients with FoxP3⁺ Treg within cellular infiltrates. (A and B) CD3⁺ (A) and FoxP3⁺ cells (B) in the same interstitial cellular infiltrate, respectively. (C) A negative FoxP3 staining of an interstitial cellular infiltrate. (D) Red arrows show positive FoxP3⁺ Treg in tubules of a patient with the diagnosis of SCR. (E) Double-immunofluorescence labeling in a renal interstitial infiltrate; red, CD3⁺ T cells; green, FoxP3⁺ cells. (F) Higher power resolution of an infiltrate; red color, CD8⁺ T cells; green, FoxP3⁺ cells. Magnifications: ×200 in C and E; ×400 in A, B, D, and F.

We studied the FoxP3 transcription factor as a binary variable to analyze whether the presence or absence of FoxP3⁺ Treg in patients with SCR was associated with relevant clinical data. As shown in Table 1, only SRL and the antecedent of induction therapy were significantly associated with infiltration of FoxP3⁺ Treg in the graft. When different immunosuppressive regimens were analyzed (Table 2), only the combination of rATG+SRL was significantly associated with presence of FoxP3⁺ Treg within graft infiltrates. On the contrary, patients who were on a CNI and had not received any induction therapy had significantly less presence of FoxP3⁺ Treg among these infiltrates than the rest.

Table 2.

Immunosuppressive regimens and presence of FoxP3⁺ Treg

Immunosuppressive Regimens	Presence of FoxP3+ Treg(n = 25)	Absence of FoxP3+ Treg(n = 12)	P
rATG + SRL (yes/no; n = 13)	13/12	0/12	0.002
rATG + CNI (yes/no; n = 5)	2/22	2/10	NS
Anti-CD25 + SRL (yes/no; n= 1)	1/24	0/12	NS
Anti-CD25 + CNI (yes/no; n= 4)	3/22	1/11	NS
No induction + CNI (yes/no; n= 14)	6/19	8/4	0.013

No relationship was found between the presence of FoxP3⁺ Tregs and other relevant clinical, demographic, and histologic characteristics, such as the antecedent of acute rejection, serum creatinine, estimated GFR (eGFR), proteinuria, and both acute and chronic Banff ’05 histologic lesions within the different renal compartments (interstitium, tubules, glomeruli, and vascular). Likewise, no differences were found regarding acute histologic scores with the global T cell (CD3⁺) infiltration between FoxP3⁺ and FoxP3⁻ patients. Furthermore, we observed that 21 (56.8%) patients had a diffuse T cell infiltrate pattern and 16 (43.2%) had a nodular distribution. Fifteen (71%) of the 21 patients with a diffuse distribution had presence of FoxP3⁺ Tregs, and 10 (62%) of 16 patients with a nodular pattern had presence of FoxP3⁺ Treg (NS). We could not observe any association between these two histologic patterns and graft function evolution or different immunosuppression in this group of patients.

Regarding graft function evolution, patients with FoxP3⁺ Treg within renal infiltrates had significantly better graft function (both serum creatinine and eGFR) compared with those without FoxP3⁺ Treg at 2 and 3 yr after transplantation (Figure 2, A and B). No differences were observed regarding levels of proteinuria between both groups (data not shown).

Figure 2.

FoxP3⁺ Treg and renal function. (A) Patients with FoxP3⁺ Treg within renal infiltrates had significantly better serum creatinine compared with those without Treg at 2 yr, and this difference was maintained 3 yr after transplantation (118.6 ± 26.5 versus 155.9 ± 46.7 μmol/L [P = 0.006] and 128.6 ± 42.0 versus 167.0 ± 47.4 μmol/L [P = 0.039], respectively). (B) Patients with FoxP3⁺ Treg within renal infiltrates displayed significantly better eGFR than those without Treg at 2 and 3 yr after transplantation (72.4 ± 20.0 versus 48.1 ± 15.8 [P = 0.001] and 72.4 ± 20.0 versus 50.1 ± 17.0 ml/min [P = 0.007], respectively). (C) The percentage of FoxP3⁺ Treg among total CD3⁺ T cells (FoxP3⁺ Treg/CD3⁺ T cells) in the 25 patients with presence of FoxP3⁺ Treg infiltrating the graft was positively correlated with the eGFR at 2 yr after transplantation (r = 0.36, P = 0.03).

When percentage of FoxP3⁺ Treg among total CD3⁺ T cells infiltrating the graft was analyzed, a significantly positive correlation was found with the eGFR at 2 yr after transplantation (r = 0.36, P = 0.03; Figure 2C). In addition, the assessment of chronic histologic lesions (interstitial and tubular chronic scores ≥2) showed that patients with no chronic damage had a significantly higher percentage of FoxP3⁺ Treg/CD3⁺ T cells than those with any chronic histologic damage (14.2 ± 16.3 versus 8.3 ± 8.1%; P = 0.035). Again, no association was observed with different acute histologic scores and the percentage of FoxP3⁺ Treg/CD3⁺ T cells.

When we evaluated whether immunosuppression could influence graft function evolution, we observed that patients on SRL had significantly better eGFR than those who were not receiving SRL at 2 and 3 yr after transplantation (Figure 3A). Thus, because SRL was associated with presence of FoxP3⁺ Treg, it could be argued that the better graft function evolution achieved among FoxP3⁺ biopsies was related to the non-nephrotoxic effect of SRL; however, when graft function evolution was analyzed within patients who were not receiving SRL, we also found that patients with presence of FoxP3⁺ Treg had significantly better graft function at 2 and 3 yr after transplantation than those without (Figure 3B).

Figure 3.

Renal function according immunosuppression and effect of FoxP3⁺ Treg on renal function in patients not receiving SRL. (A) Patients on SRL had significantly better eGFR than those not receiving SRL at 2 and at 3 yr after transplantation (78.3 ± 21.0 versus 55.1 ± 18.2 ml/min [P = 0.003] and 89.7 ± 11.0 versus 56.1 ± 18.0 ml/min [P = 0.003], respectively). (B) Among patients not receiving SRL, patients with FoxP3⁺ Treg within graft infiltrates had significantly better graft function at 2 and at 3 yr after transplantation than those without FoxP3⁺ Treg (63.0 ± 17.1 versus 44.5 ± 5.0 ml/min [P = 0.02] and 62.0 ± 20.0 versus 52.3 ± 16.6 ml/min [P = 0.04], respectively).

Subsequently, we performed a Kaplan-Meier analysis considering event graft loss or achieving a significant decrease of renal function, regarded as eGFR <40 ml/min. As shown in Figure 4, patients with FoxP3⁺ Treg had significantly lower risk for event than patients without FoxP3⁺ Treg. Univariate and multivariate Cox regression analyses were performed to analyze clinical, analytical, and histologic data associated with the achievement of an eGFR <40 ml/min. As shown in Table 3, in the univariate analyses, 6-mo serum creatinine and presence of FoxP3⁺ Treg within graft infiltrates were associated with outcome, whereas induction therapy was marginally significant (P = 0.07). Other relevant variables, such as interstitial fibrosis and tubular atrophy, acute rejection, delayed graft function, cold ischemia time, donor and recipient age, CNI-based or SRL-based immunosuppression, and HLA class I and II mismatches, were not risk factors for declining renal function in patients with the diagnosis of SCR in 6-mo protocol biopsy. Likewise, in the multivariate analyses, only 6-mo serum creatinine and presence of FoxP3⁺ Treg were independent predictor factors for maintaining an eGFR ≥40 ml/min.

Figure 4.

Kaplan-Meier estimates of achieving eGFR <40 ml/min per 1.73 m² in patients without presence of FoxP3⁺ Treg within the graft.

Table 3.

Variables associated with eGFR >40 ml/min by univariate and multivariate Cox regression analysis adjusting for timing of the protocol biopsy^a

Variable	Category	Univariate Analysis			Multivariate Analysis
		RR	CI 95%	P	RR	CI 95%	P
Induction therapy	Y/N	1.280	0.900 to 14.300	0.0700	0.820	0.210 to 5.400	0.9000
6-mo SCr		0.016	1.000 to 1.020	0.0016	0.018	1.020 to 1.008	<0.0100
Presence of FoxP3+ Treg	Y/N	1.440	1.050 to 16.900	0.0410	1.920	1.120 to 41.500	0.0370

^aCI, confidence interval; SCr, serum creatinine.

DISCUSSION

Presence of histologic patterns of acute rejection in patients with well-functioning renal allografts is a controversial topic in clinical transplantation. Herein, we show that presence of FoxP3⁺ Treg within asymptomatic cellular infiltrates in 6-mo protocol biopsies may be a reliable biomarker for distinguishing an ongoing rejection/inflammatory process from a safe/protective condition. This conclusion is supported by the better graft function evolution achieved at both 2 and 3 yr after transplantation in patients with FoxP3⁺ Treg. Furthermore, the FoxP3⁺ Treg/CD3⁺ T cell ratio was positively correlated with graft function at 2 yr after transplantation, suggesting that not only the presence of Treg but also its proportion regarding the global T cell infiltrate is of relevance for facilitating renal engraftment. This fact would support the results of previous experiences^6,7 in which after not treating the so-called SCR, neither increase of interstitial fibrosis nor progressive loss of graft function was observed. A potential mechanistic explanation that could clarify this process is that donor-antigen recognition by Treg directly in the graft would be necessary for developing a donor-specific hyporesponsive state, mediated by the suppressive activity of these Treg.²² Accordingly, in a recent study, we showed that presence of FoxP3⁺ Treg within tubulointerstitial infiltrates in a group of stable renal transplant patients in 6-mo protocol biopsies was associated with peripheral donor-specific hyporesponsiveness, which was mediated by the antidonor suppressive activity of FoxP3⁺ Treg.¹⁷

It has been described that some immunosuppressants such as SRL and rATG are more likely to play a role in the development and expansion of Treg in peripheral blood.^16,23 Furthermore, a recent interesting experimental report showed how SRL induced expression of TGF-β played a role in the recruitment of FoxP3⁺ Treg from the naive pool whereas CNI have an opposite effect and prevent recruitment of Treg.²⁴ Here, we also show for the first time that patients who are on SRL and/or are receiving induction therapy (mostly with rATG) have significantly higher expression of FoxP3⁺ Treg within clinically stable renal allografts. Nonetheless, certain patients on CNI did also present Treg within graft infiltrates, and, interestingly, they also displayed better graft function evolution. Thus, this feature may support the hypothesis that presence of FoxP3⁺ Treg in patients with the diagnosis of SCR yields better graft function evolution even in patients treated with standard immunosuppression under CNI.

It is very important to differentiate from the current report previous experiences that studied FoxP3⁺ Treg in patients undergoing clinical acute rejection; that is, when there is ongoing renal graft dysfunction, thus analyzing a completely different immunologic scenario. First, Muthukumar et al.²⁵ reported indirect evidence of FoxP3⁺ Treg infiltrating the graft by measuring FoxP3 mRNA in urine from patients with clinically and biopsy-proven acute rejection. They showed that levels of serum creatinine and FoxP3 mRNA in urinary cells were independent predictors of both reversal of acute rejection and graft failure. In contrast, Veronese et al.,²⁶ again in patients undergoing biopsy-proven acute rejection, did not observe that presence of FoxP3⁺ Treg exerted any beneficial effect on graft outcome. Interestingly, Grimbert et al.²⁷ recently reported that patients with clinical dysfunction and BLc had significantly higher levels of mRNAFoxP3/mRNAGranzymeB ratio in the graft than patients with IA acute rejection; therefore, it is likely that during clinical acute rejection, Treg would counterbalance graft-destructive effector function of cytotoxic T cells rather than have a preventive or protolerogenic function. In fact, it has been well documented that under inflammatory conditions, Treg may not be able to develop their antidonor suppressive activity.^28,29 Herein, we analyzed a completely different clinical situation, which is that of patients with stable renal function but with histologic signs of acute rejection. Remarkably, just the absence of any FoxP3⁺ Treg within these infiltrates acts as an independent predictor factor for poorer graft outcome.

In conclusion, our study provides new insight suggesting that evidence of FoxP3⁺ Treg within tubulointerstitial infiltrates in clinically stable renal allografts may be able to differentiate harmless from detrimental cellular infiltrates. Outstandingly, distinguishing these two different patterns of infiltrates has relevant clinical consequences and perhaps can explain the controversial findings about prognosis and treatment of SCR. Hence, absence of FoxP3⁺ Treg in patients with SCR may identify patients who would benefit from increasing immunosuppression.

CONCISE METHODS

Patients

Since 1988, prospective protocol biopsies have been systematically performed on renal transplant patients who were from our hospital and gave informed consent. Biopsies were selected from those previously graded and diagnosed blindly for SCR accordingly to the Banff ’05 criteria³⁰ in absence of any clinical information. SCR was defined as presence of acute interstitial and tubular score ≥1. Chronic histologic changes were graded as chronic interstitial and tubular scores ≥2.

Inclusion criteria: patients with a transplanted renal organ; biopsies performed at 6 mo after transplantation; serum creatinine <300 μmol/L; proteinuria <1 g/24 h; stable renal function (defined as variability of serum creatinine of <15% during 2 wk before and after biopsy); adequate tissue sample (considered as presence of at least 10 glomerular and two arterial sections); and immunosuppression consisting of at least one of the following: Cyclosporine/tacrolimus, sirolimus, or mycophenolate mofetil. For this study, nonexhausted paraffin blocks from protocol biopsies done at 6 mo after transplantation were selected. A total of 170 biopsies were performed on 170 patients suitable to be included. Among them, 37 cases fit the diagnosis of SCR and were selected for this study.

Clinical data were obtained from our local transplant database. Graft function was analyzed at the time of biopsy and 1, 2, and 3 yr after transplantation by serum creatinine (μmol/L), eGFR estimated by the Cockroft-Gault formula (ml/min), and proteinuria (g/24 h).

Immunohistochemistry

In all cases, FoxP3 (the 86D/D6 mouse mAb; CNIO, Madrid, Spain), CD3 (Master Diagnostica, Madrid, Spain), and C4d (Biomedica, Madrid, Spain) immune staining in formalin-fixed, paraffin-embedded tissues was performed as described previously.¹⁷ The number of positive FoxP3⁺ and CD3⁺ T cells in the cortex, perivascular areas, and corticomedullary junction aggregates were first counted by using symmetric square power fields (×400; Leica Geosystems, Barcelona, Spain). Then, the proportion of FoxP3⁺ Treg among the CD3⁺ T cells (FoxP3⁺/CD3⁺) per field was also scored. Also, all infiltrates evaluated in each biopsy were classified according to the different infiltrate pattern, such as nodular or diffuse. The nodular pattern was defined as the presence of a cluster of cells of at least the size of a tubular cross-section.

Data were given by presence of FoxP3⁺ Treg in a protocol biopsy and by percentages of FoxP3/CD3 per field. Also, double immunofluorescence labeling was done (Leica TCS-SL), combining the intranuclear FoxP3 with four different T cell surface markers: CD25, CD4, CD8, and CD3 (Master Diagnostica, Vision Biosystem Novocastra), as described previously.¹⁷ All of the histologic analysis was done in a blinded manner.

Statistical Analysis

Results are expressed as means ± SD. Comparison between groups was performed by means of χ² test for categorical data. The one-way ANOVA or t test was used for normally distributed data, and the nonparametric Kruskal-Wallis or Mann-Whitney U test was used for non-normally distributed variables. Renal function expressed either by serum creatinine or eGFR was considered the outcome variable of the study. Graft survival was also estimated using Kaplan-Meier survival method, considering event when eGFR ≤40 ml/min and differences between the groups were established using the log-rank test. To decide which was the most sensible significant cutoff GFR value to be used as an event in the Kaplan-Meyer analyses, we made a sensitivity/specificity receiver operating characteristic curve test, using different values of GFR (30, 40, and 50 ml/min). The most sensible value was a GFR of 40 ml/min (which had a 74% sensitivity; 95% confidence interval 0.52 to 0.93; P = 0.04). Univariate and multivariate Cox model was used to evaluate risk factors for eGFR ≤40 ml/min. All P values were two-tailed, and the statistical significance level was defined as P < 0.05.

DISCLOSURES

None.