The primary endpoint of this study was to investigate whether, in the course of DGF, due to ATN, there is an immune-mediated insult to the allograft delivered by the recipient’s T lymphocytes. To this purpose, we established the biopsy-proven diagnosis of ATN following transplantation in KTrs as the clinical outcome of DGF.

Thirty-one consecutive first-time KTrs were enrolled in this unicentre prospective observational study at the Immunology Department of the Clinical University Hospital “Virgen de la Arrixaca” (HUVA) in Murcia, Spain. Additionally, 17 healthy volunteers were also included as controls. For patients and control subjects, a sample consisting of 10 mL of whole peripheral blood was drawn in a sodium heparin container by venipuncture phlebotomy. This protocol of blood extraction was followed at different time points within the first-year post-transplantation (first and second-week; first, second, third and sixth-month; first-year) as shown in Fig. 1. Socio-demographic data (age, sex) from donors and recipients, alongside clinical, pathological and immunological data were collected in a unified database. Post-transplant complications were also registered (AR, opportunistic infections).

Fig. 1.

Study design. Kidney transplant recipients enrolled in this prospective observational study were appointed for sample collection at different time points during the first-year post-transplantation. Following blood extraction, samples were activated in vitro using polyclonal stimulus to assess T lymphocytes for several surface markers using multi-parametric flow cytometry.

Inclusion criteria for participation in the study were as follows: ABO match, first-time kidney transplant, immunosuppressive therapy based on tacrolimus (TRL) with or without mycophenolate mofetil (MMF), as well as HIV-negativity. Paediatric, re-transplant, or simultaneous kidney-pancreas or liver-kidney transplant patients were excluded.

All patients gave informed consent for their samples to be manipulated and stored for current and future research before recruitment to the study. The study was conducted following the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the Clinical University Hospital “Virgen de la Arrixaca” (HUVA) (PI15/01370).

Patients enrolled in this study were prospectively monitored for percentages of CD4${}^{+}$ and CD8${}^{+}$ T lymphocytes as well as measuring the upregulation of various important activation and functional markers (CD25, CD38, CD154, and CD95) applying a modified protocol used by Barten MJ et al. [27] and as previously described elsewhere [26, 28]. Briefly, 10 mL of whole peripheral blood (WPB) was drawn for in vitro stimulation purposes according to our own validated standard operating procedure for T lymphocyte stimulation. WPB samples were diluted with RPMI-1640 (1:10) (BioWhittaker®, Lonza, Belgium) and added into flat-bottom 24-well tissue culture microtiter plates. Patient samples were polyclonally activated with lectin mitogenic concanavalin A (Con-A), which is known for its ability to interact with specific components of the T-cell receptor [29] (1 mg/mL; Sigma-Aldrich®, St. Louis, Missouri, USA), reaching a final concentration of 15 µg/mL and a final volume of 2 mL per well. Cell-culture microtiter plates were incubated in a humidified 5% CO${}_2$ incubator at 37 °C for 72 hours (Fig. 1).

Upon in vitro stimulation, cell cultures were prepared for multi-colour stain using a combination of mouse anti-human monoclonal antibodies (MoAb) (BD® Biosciences BD, San Jose, CA, USA). Stimulated samples were incubated with a panel of MoAb according to the manufacturer’s instructions following erythrocyte lysis by adding 2 mL of 1$\times$ lysis buffer (BD FACS™ Lysing Solution). The panel of MoAb included a mouse IgG1${}_{\text{K}}$ anti-human CD3-FITC, CD4-APC, CD8-PerCPCy™ 5.5, CD25-PE, CD38-PE, CD69-PE and CD95-PE (BD® Biosciences BD, San Jose, CA, USA) and mouse IgG1${}_{\text{K}}$ anti-human CD154-PE (Beckman Coulter®, Marseille, France). In all cases, isotype control antibodies were used to assess the positivity of each fluorochrome; IgG1-FITC and IgG1-PE (Beckman Coulter®, Marseille, France), mouse IgG1${}_{\text{K}}$-APC and mouse IgG1${}_{\text{K}}$-PerCP (BD® Biosciences BD, San Jose, CA, USA). Mean fluorescence intensity (MFI) was used as a relative measurement of molecule expression. Finally, the patient’s samples were analysed in a flow cytometer (FACScanto™ II, BD® Biosciences BD, San Jose, CA  USA) following the acquisition of 50,000 events from the activated CD3${}^{+}$ T lymphocyte gate. Data analysis was carried out with BD FACSDiva™ Software 6.1.3 (FACScanto™ II, BD® Biosciences BD, San Jose, CA  USA) for the percentage and level of expression of CD25, CD38, CD69, CD95, and CD154 on CD4${}^{+}$ and CD8${}^{+}$ T lymphocytes, as shown in Fig. 1.

The recipients enrolled in the present study were classified into two study groups named as follows, with acute tubular necrosis (ATN) or without ATN (NoATN). Seven patients (22.6%) experienced DGF due to ATN within the first-year post-transplantation and were therefore included in the ATN study group, whereas the remaining patients (n = 24, 77.4%) were included in the NoATN study group. At our institution, KTrs experiencing DGF with oliguria are subjected to a “for-cause biopsy”. Socio-demographic, clinical, and immunological characteristics, as well as the outcomes, were compared between the two groups.

In this context, renal allograft biopsies are commonly evaluated by experienced pathologists based on the current Banff classification [30, 31, 32]. There is no current definition of ATI endorsed by the Banff classification. Currently, ATI is included as a diagnosis in Banff diagnostic category 6, which is characterised by the lack of other apparent causes but histological evidence of acute tissue injury in the context of a diagnosis of acute or active ABMR with antibody-mediated changes [30]. Therefore, ATI was diagnosed from patient biopsies with DGF clinically manifested as oliguria that did not show antibody-mediated changes but others unrelated to acute or chronic rejection. Variable tissue changes used to diagnose ATI were described as vacuolisation, brush border loss, and pyknotic nuclei of tubular epithelial cells with signs of tubular dilatation with or without tubulitis, or the presence of any of the following, interstitial inflammation, glomerulitis, tubular atrophy, and luminal casts [33, 34].

KTrs received immunosuppression therapy based on our institution’s standard practice of care. Accordingly, induction therapy consisted of either rabbit anti-thymocyte globulin or anti-CD25 monoclonal antibody-based therapy. Therefore thymoglobulin was provided to high-immune-risk patients, whereas basiliximab was given to low-immune-risk patients. From our data, 87.1% (n = 27) received basiliximab and 12.9% (n = 4) received thymoglobulin.

Maintenance immunosuppression therapy was based on the administration of either tacrolimus (TRL, Prograf®, Astellas Pharma, United Kingdom) with a target dose of 5 mg/day or mycophenolate mofetil (MMF, CellCept®, Roche Pharma, Switzerland) with a target dose of 2000 mg/day. All patients in this study were under the same immunosuppressive conditions. Additionally, KTrs also received methylprednisolone as the main corticosteroid-based therapy (Dacortin® 20 mg/day). Steroids were gradually reduced to 5 mg/day from day 1 until day 90 post-transplantation. Due to the number of side effects associated with glucocorticoids, patients with stable graft function were weaned from steroids at the sixth-month post-transplantation. A total of 90.3% (n = 28) of KTrs received double immunosuppression therapy based on TRL and MMF, whereas the remaining patients were on a monotherapy regimen, either with TRL (3.2%, n = 1) or MMF (6.5%, n = 2).

Demographic, clinical, and immunological data were collected in a unified database (Microsoft Access 11.0; Microsoft Corporation, Seattle, WA, USA), and statistical analysis was performed using the SPSS 20.0 software (SPSS Inc., Chicago IL, USA). The Kolmogorov–Smirnoff test was used to assess whether the demographic and clinical patient data, as well as the percentages and MFI data of the T lymphocyte subsets adjusted to parametric distribution. All variables were distributed nonparametrically. Therefore, quantitative data were presented using the median with an interquartile range of 25 and 75, whereas qualitative data were presented as absolute and relative frequencies.

Pearson’s X${}^2$ and two-tailed Fisher’s exact tests were carried out to detect differences comparing bivariant categorical variables between groups, whereas the Kruskal–Wallis test was applied for variables with more than two categories. Odds ratios (OR) and 95% confidence intervals (CI) were used to calculate the specificity, sensibility, as well as positive and negative predictive values of our models. Additionally, the U Mann–Whitney test was used to compare unpaired continuous variables, and Spearman’s rank-order correlation coefficient test was used to measure the strength and direction of associations existing between nonparametric paired continuous variables.

Receiver operating characteristic (ROC) curves were used to identify the optimal cut-off points for those surrogate biomarkers deemed significant to stratify patients at high risk of ATN. Cut-off points were calculated based on the best Youden-index (sensitivity + specificity-1) [35]. The area under the ROC curve (AUC) was analysed as follows, an area of 0.6–0.7 was considered acceptable; an area of 0.7–0.8, excellent; and an area of $>$0.8, outstanding [36].

Any demographic, clinical, and immunological variable statistically significant at the univariate pre-transplant cross-sectional analysis as well as any known variable with clinical importance, was finally assessed in a backward stepwise multivariate logistic regression [37]. Multivariable logistic regression analysis was applied to confirm positive associations. Relative hazard ratios (HR) and their 95% confidence intervals (CI) were calculated to estimate the likelihood of the occurrence of ATN. A level of p $\le$ 0.05 was accepted as statistically significant.

Within the first-year post-transplantation, 22.6% of KTrs (n = 7) were diagnosed with DGF due to biopsy-proven ATN versus 77.4% (n = 24) who did not. Furthermore, from the ATN group, only one patient (14.3%) additionally developed AR, whereas 25% (n = 6) of recipients from the NoATN group did not reject the allograft during the study. Nevertheless, neither the presence nor the absence of ATN was correlated with the occurrence of AR in our cohort of patients (p = 0.551, OR = 1.750, 95% CI = 0.251–12.207 and OR = 0.875, 95% CI = 0.598–1.280, respectively).

Overall, the median [IQR] donor age was 55 [42–61], and the median [IQR] recipient age was 53 [41–59] with no significant impact on ATN (p = 0.679 and p = 0.125, respectively) despite age being previously associated with DGF and ATN. However, some intrinsic and extrinsic immunological and clinical characteristics of the patients differed significantly from the ATN and NoATN study groups. For instance, the presence of preformed anti-HLA antibodies (42.9% vs. 16.7%, p = 0.012, OR = 3.750, 95% CI = 1.295-10-862), and the maximum (6.42 $\pm$ 1.37 vs. 4.46 $\pm$ 1.27, p = 0.005), as well as current (4.71 $\pm$ 1.56 vs. 3.21 $\pm$ 0.87, p = 0.032) panel reactive antibody (PRA), showcased significant differences between patients diagnosed with and without ATN. Moreover, the dose of MMF (mg/day) was significantly higher in NoATN than in ATN (1890.83 $\pm$ 30.56 vs. 1500.25 $\pm$ 0.89, p = 0.006). Post-transplant immunosuppressive therapy was also significantly correlated with ATN, suggesting that in patients treated with a double therapeutic regimen (TRL + MMF), 78.57% (n = 22) had stable allograft function within the first-month post-transplant, compared with 21.43% (n = 2) who developed ATN (p = 0.019). Finally, renal function was significantly impaired in patients with ATN compared with patients without ATN, as showed by serum creatinine (13.06 $\pm$ 7.07 vs. 2.48 $\pm$ 0.25, p = 0.000000143) and glomerular filtration rate (32.71 $\pm$ 6.18 vs. 63.78 $\pm$ 3.32, p = 0.00012). All remaining patient characteristics were not significant. These data are depicted in Table 1.

As ATN is most frequently diagnosed within the first days following surgery, we focused our stratification analysis on different CD4${}^{+}$ and CD8${}^{+}$ T lymphocyte subsets from day one to one-month post-transplant. We calculated the mean of the distribution for NoATN vs ATN of CD4${}^{+}$CD95${}^{+}$ (Fig. 2A, 44.31 $\pm$ 1.74 vs. 36.59 $\pm$ 3.68, p = 0.0446) and CD8${}^{+}$CD95${}^{+}$ (Fig. 2B, 24.31 $\pm$ 3.71 vs. 53.82 $\pm$ 1.26, p = 0.000168) T lymphocytes from 7, 15 and 30 days post-transplantation, where both showed statistically significant differences. No differences were observed between either of the T lymphocyte subsets and the control group. Furthermore, the percentages of CD4${}^{+}$CD25${}^{+}$, CD8${}^{+}$CD25${}^{+}$, CD4${}^{+}$CD38${}^{+}$, CD8${}^{+}$CD38${}^{+}$, CD4${}^{+}$CD154${}^{+}$, CD8${}^{+}$CD154${}^{+}$, CD4${}^{+}$CD69${}^{+}$ and CD8${}^{+}$CD69${}^{+}$ T lymphocytes from both study groups showed no differences at any other time point during the study (Supplementary Table 1).

Fig. 2.

Stratification analysis in KTrs within the first-month post-transplantation. (A) Differences in the percentage of CD4${}^{+}$CD95${}^{+}$ T lymphocytes* between ATN, NoATN and controls. (B) Differences in the percentage of CD8${}^{+}$CD95${}^{+}$ T lymphocytes* between ATN, NoATN and controls. ATN, KTrs with acute tubular necrosis; NoATN, KTrs without acute tubular necrosis; KTrs, kidney transplant recipients. *The mean percentages of CD4${}^{+}$CD95${}^{+}$ and CD8${}^{+}$CD95${}^{+}$ T lymphocytes at 7, 15 and 30 days post-transplantation were used in this analysis.

The expression, measured as MFI, of the different surface markers, did not differ much from what we observed regarding the distribution of T cell subsets. Mean MFI values were calculated for the different T lymphocyte subsets and used in the analysis. Albeit not seen as statistically significant following the univariate stratification analysis, the expression of CD25 on CD8${}^{+}$ T lymphocytes showed a trend towards a higher MFI in ATN than in NoATN patients (Fig. 3A, 2704.44 $\pm$ 825.37 vs 1632.12 $\pm$ 476.17, p = 0.058). CD8${}^{+}$CD69${}^{+}$ T lymphocytes showed differences between both study groups, where patients with ATN had statistically significantly higher MFI values than patients without ATN (Fig. 3B, 4495.10 $\pm$ 604.85 vs. 3985.43 $\pm$ 549.94, p = 0.035). Furthermore, the level of CD95 again indicated an inverse expression depending on the T lymphocyte subset in such a way that patients with ATN had a lower expression of CD95 on CD4${}^{+}$ (Fig. 3C, 4495.10 $\pm$ 604.85 vs. 3985.43 $\pm$ 549.94, p = 0.004) compared with CD8${}^{+}$ (Fig. 3D, 4567.11 $\pm$ 642.20 vs 2861.08 $\pm$ 575.36, p = 0.012) T lymphocytes. None of the T lymphocyte subsets differed significantly from controls, as shown in Fig. 3.

Fig. 3.

Stratification analysis in KTrs within the first-month post-transplantation. (A) Differences in the MFI of CD8${}^{+}$CD25${}^{+}$ T lymphocytes* between ATN, NoATN and controls. (B) Differences in the MFI of CD8${}^{+}$CD69${}^{+}$ T lymphocytes* between ATN, NoATN and controls. (C) Differences in the MFI of CD4${}^{+}$CD95${}^{+}$ T lymphocytes* between ATN, NoATN and controls. (D) Differences in the MFI of CD8${}^{+}$CD95${}^{+}$ T lymphocytes* between ATN, NoATN and controls. ATN, KTr with acute tubular necrosis; NoATN, KTr without acute tubular necrosis; KTr, kidney transplant recipients; MFI, median fluorescent intensity. *Mean MFI of CD8${}^{+}$CD25${}^{+}$, CD8${}^{+}$CD69${}^{+}$, CD4${}^{+}$CD95${}^{+}$ and CD8${}^{+}$CD95${}^{+}$ T lymphocytes at 7, 15 and 30 days post-transplantation were used in this analysis.

Following the univariate stratification analysis, we carried out a study to validate, from the pool of significant T lymphocyte subsets, those of importance as surrogate biomarkers for ATN in KTrs within the first-month post-transplantation.

The AUC analysis showed that out of all CD4${}^{+}$ and CD8${}^{+}$T lymphocyte subsets, only a few were capable of stratifying statistically significant recipients at high risk of ATN with reasonable sensitivity and specificity (Fig. 4). Therefore, we tested the capability of our four different predictive models, where we found that CD25 and CD69 expression on CD8${}^{+}$ T lymphocytes correlated directly with the risk of ATN. Accordingly, 84.8% of patients with MFI values of CD8${}^{+}$CD25${}^{+}$1015.20 did not develop ATN compared with 15.2% that, regardless of being below the cut-off, developed ATN (OR = 4.738, 95% CI = 1.662–13.508, p = 0.002). Similarly, 76.2% of KTrs with an MFI of CD8${}^{+}$CD69${}^{+}$$\ge$2489.05 developed ATN, whereas 23.8% of patients developed ATN despite being stratified as low risk (OR = 3.491, 95% CI = 1.151–10.590, p = 0.022).

Fig. 4.

ROC curve analysis within the first-month post-transplant. (A) AUC${}^{\rfloor }$, sensitivity and specificity of CD8${}^{+}$CD25${}^{+}$ T lymphocytes that accurately determined the cut-off${}^{\mathrm{‡}}$ for the MFI value to be used as a surrogate biomarker to stratify KTrs at high risk of ATN. (B) AUC${}^{\rfloor }$, sensitivity and specificity of CD8${}^{+}$CD69${}^{+}$ T lymphocytes that accurately determined the cut-off${}^{\mathrm{‡}}$ for the MFI value to be used as a surrogate biomarker to stratify KTrs at high risk of ATN. (C) AUC${}^{\rfloor }$, sensitivity and specificity of CD4${}^{+}$CD95${}^{+}$ T lymphocytes that accurately determined the cut-off${}^{\mathrm{‡}}$ for the MFI value to be used as a surrogate biomarker to stratify KTrs at high risk of ATN. (D) AUC${}^{\rfloor }$, sensitivity and specificity of CD8${}^{+}$CD95${}^{+}$ T lymphocytes that accurately determined the cut-off${}^{\mathrm{‡}}$ for the MFI value to be used as a surrogate biomarker to stratify KTrs at high risk of ATN. ${}^{\rfloor }$AUC = values considered in this study: 0.6–0.7 as acceptable; 0.7–0.8 as excellent, and $>$0.8 as outstanding. ${}^{\mathrm{‡}}$Youden index (sensitivity + specificity – 1) was carried out to obtain the most accurate cut-off values for the MFI of CD8${}^{+}$CD25${}^{+}$, CD8${}^{+}$CD69${}^{+}$, CD4${}^{+}$CD95${}^{+}$, and CD8${}^{+}$CD95${}^{+}$ T lymphocytes capable of stratifying patients at high risk of ATN. AUC, area under the curve; 95% CI, 95% confidence interval; ROC, receiver operating characteristic curve; KTrs, kidney transplant recipients; ATN, acute tubular necrosis.

In addition, the expression of CD95 on CD4${}^{+}$ or CD8${}^{+}$ T lymphocytes differed in stratifying KTrs. A total of 88.6% of patients with MFI values of CD4${}^{+}$CD95${}^{+}$$>$1581.98 did not develop ATN compared with 11.4% who experienced otherwise (OR = 14.393, 95% CI = 4.434–46.721, p = 0.0000587). Contrarily, MFI values of CD8${}^{+}$CD69${}^{+}$4257.28 were capable of stratifying patients at low risk of ATN, showing that 84.1% of KTrs with levels below the cut-off were free of ATN, whereas 15.9% were diagnosed with ATN (OR = 3.644, 95% CI = 1.310–10.132, p = 0.011).

The specificities, sensitivities, and positive and negative predictive values of the four different predictive models for ATN are depicted in Table 2.

Recipients who experienced ATN had significantly lower glomerular filtration rates (GFR) compared with those who displayed stable allograft function within the first-month post-transplantation (Fig. 5A, 32.71 $\pm$ 6.18 vs 63.78 $\pm$ 3.32, p = 0.000120). Furthermore, the impairment in renal function was accompanied by a significant increase in serum creatinine levels, as shown in Fig. 5B (13.06 $\pm$ 7.07 vs 2.48 $\pm$ 0.25, p = 0.000000143).

Fig. 5.

Analysis of the allograft function and its association with T lymphocytes. (A) Differences in the GFR (mL/min/1.73 m${}^2$) between KTrs with and without ATN. (B) Differences in the concentration of serum creatinine (mg/dL) between KTrs with and without ATN. (C) Differences in the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio between KTrs with and without ATN. (D) AUC${}^{\rfloor }$, sensitivity and specificity of CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio that accurately determined the cut-off${}^{\mathrm{‡}}$ value to be used as a surrogate biomarker to stratify KTrs at high risk of ATN. (E) Logarithmic-scale correlation (log-log) plot analysis between the concentration of serum creatinine and the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio with the first-month post-transplantation. (F) Logarithmic-scale correlation (log-log) plot analysis between GFR (mL/min/1.73 m${}^2$) and the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio with the first-month post-transplantation. ${}^{\rfloor }$AUC = values considered in this study: 0.6–0.7 as acceptable; 0.7–0.8 as excellent, and $>$0.8 as outstanding. ${}^{\mathrm{‡}}$Youden index (sensitivity + specificity – 1) was carried out to obtain the most accurate cut-off values for the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio capable of stratifying patients at high risk of ATN. GFR, glomerular filtration rate; KTr, kidney transplant recipient; ATN, acute tubular necrosis; AUC, area under the curve; 95% CI, 95% confidence interval; ROC, receiver operating characteristic curve. **** means a p value $\le$ 0.0001.

Interestingly, we calculated the ratio of CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ and compared it against both study groups showing that ATN patients had an inversion of this ratio compared with NoATN patients (Fig. 5C, 0.56 $\pm$ 0.09 vs 1.09 $\pm$ 0.31, p = 0.023).

Furthermore, we assessed the CD4${}^{+}$CD95${}^{+}$/ CD8${}^{+}$CD95${}^{+}$ ratio using a ROC curve analysis to ascertain the cut-off point to stratify KTrs at high risk of allograft failure in the absence of rejection. Recipients were tested before transplantation and within the first-month post-transplantation. Albeit the ROC curve analysis at baseline did not reach statistical significance (AUC = 0.470, 95% CI = 0.211–0.729, p = 0.813), it was different when used in the first-month post-transplantation. A CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio $\le$0.32 accurately stratified KTrs at high risk of DGF due to ATN (Fig. 5D, AUC = 0.664, 95% CI = 0.522–0.806, p = 0.023) with reasonable specificity and sensitivity, as shown in Table 2. In fact, after stratifying KTrs according to the calculated cut-off point, 88.9% of recipients who fell into the high-risk group for DGF due to a CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio $\le$0.32 were diagnosed with ATN in the first-month post-transplantation, as opposed to 11.1% who did not develop ATN irrespective of a cut-off $\le$0.32. In our cohort, recipients with a CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio $\le$0.32 had a 5-fold higher risk of ATN over other recipients (OR = 5.00, 95% CI = 1.64–15.21, p = 0.005).

Correlation analysis showed that the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio was inversely proportional to serum creatinine [Fig. 5E, r${}_{\mathrm{\text{s}}}$ (8) = 0.226, p = 0.033] whereas for GFR, the CD4${}^{+}$CD95${}^{+}$/CD8${}^{+}$CD95${}^{+}$ ratio showed a positive correlation [Fig. 5F, r${}_{\mathrm{\text{s}}}$ (8) = 0.378, p = 0.000239]. In both cases, an increase in CD8${}^{+}$CD95${}^{+}$ T lymphocytes over CD4${}^{+}$CD95${}^{+}$ T lymphocytes was associated with poor allograft function.

Following the stratification analysis, we then analysed whether the differences showed by CD8 T cell subsets between patients with ATN and stable graft function could be used to stratify them into two different groups based on the MFI expression of CD25, CD69 and CD95 in CD8 T cells to ascertain the risk of ATN. ROC curve analysis was applied to find the best cut-off values for CD8${}^{+}$CD25${}^{+}$, CD8${}^{+}$CD69${}^{+}$, and CD8${}^{+}$CD95${}^{+}$ (Fig. 4).

An MFI $\ge$923 for CD8${}^{+}$CD25${}^{+}$ (AUC = 0.616, 95% CI = 0.484–0.748, p = 0.048), 2747 for CD8${}^{+}$CD69${}^{+}$ (AUC = 0.652, 95% CI = 0.537–0.767, p = 0.021) and 2638 for CD8${}^{+}$CD95${}^{+}$ (AUC = 0.657, 95% CI = 0.540–0.774, p = 0.017) were capable of stratifying KTrs at high risk of ATN within the first-month post-transplantation.

Following univariate analysis, those patients and donor characteristics of clinical importance that showed, by any means, an association with poor allograft function within the first-month post-transplantation were input into a multivariate logistic regression model to ascertain their independency as risk factors for ATN. Certainly, various variables kept their statistical significance toward a poorer outcome by influencing the odds of developing ATN.

From the multivariate analysis, 45.3% of the variation in the development of ATN in KTrs cannot be explained by our predictors (covariables). Our regression model accurately predicted ATN development with high sensitivity and specificity, showing that 95.65% of patients from the low-risk pool did not develop ATN, whereas 80.95% of KTrs stratified as high risk indeed developed ATN within the first-month post-transplantation. Donor age had a protective effect on the risk of ATN, showing that for each year increase in donor age, the recipient was 0.847 times less likely to develop ATN compared with patients transplanted with younger donors (HR: 0.847, 95% CI = 0.728–0.984, p = 0.003). Similarly, regarding renal function measured as GFR, a protective effect was also observed. Particularly, patients who maintained stable allograft function in the first-month after transplantation had a 6.2% reduction in the relative risk of ATN (HR: 0.938, 95% CI = 0.882–0.997, p = 0.039). Furthermore, serum creatinine showed a significant detrimental effect on the primary outcome, as recipients with higher levels were 1.08% more likely to develop ATN following transplantation (HR: 0.938, 95% CI = 0.882–0.997, p = 0.039).

Regarding the immunologic characteristics of the patients, several variables contributed independently to the odds of developing ATN within our predictive model. There were independent pre- and post-transplant variables that affected the risk associated with ATN. Recipients with preformed anti-HLA antibodies were 5.726 times more likely to develop ATN than those without antibodies (HR: 5.726, 95% CI = 1.313–24.977, p = 0.020).

The most important variables we aimed to analyse were the different T lymphocyte subsets to determine any possible underlying immune-mediated pathogenic mechanism causing acute tubular allograft injury. In this respect, we were able to evaluate their potential use as surrogate biomarkers of ATN and, indeed, our predictive model significantly correlated three out of the four T lymphocyte subsets as independent factors contributing to increased risk of ATN. Thus, KTrs with CD8${}^{+}$CD25${}^{+}$ MFI $\ge$1015.20 within the first-month post-transplantation had a 5.212-fold increased risk of ATN compared with those with levels below the cut-off point (HR: 5.212, 95% CI = 1.295–20.975, p = 0.020). Likewise, KTrs with CD8${}^{+}$CD95${}^{+}$ MFI $\ge$4257.28 were 3.693 times more likely than recipients with levels 4257.28 to develop ATN (HR: 3.693, 95% CI = 1.928–14.689, p = 0.034). The main predictor and contributor to the increased risk of ATN was the CD4${}^{+}$CD95${}^{+}$ T lymphocyte population, demonstrating that the lower the decrease below an MFI of 1581.98, the higher the probability of relative risk of ATN, increasing by a factor of 12.376-fold (HR: 12.376, 95% CI = 3.104–49.339, p = 0.000363). All these data are shown in Table 3.