This is an observational study conducted in the First Department of Nephrology of Aristotle University of Thessaloniki, in collaboration with the Department of Immunology, National Histocompatibility Center, of General Hospital of Thessaloniki “Hippokration”, from June 2021 to June 2024. Four groups of patients were included in the study; patients with CKD stage V not yet in dialysis (estimated glomerular filtration rate - eGFR: <15 mL/min/1.73 m²) – CKD-V group; patients in chronic maintenance hemodialysis for at least 1 year – HD group; recently transplanted patients (1 year since kidney transplantation) – rKTx group; long transplanted patients (>17 years since kidney transplantation) – lKTx group. The cut-off of 17 years was the average predicted 3rd quantile of living and deceased donors graft survival in years, as reported in a large study published in 2021 [8]. All transplanted patients had received an ABO and HLA compatible graft, with no detectable anti-HLA antibodies against donor’s HLA molecules.

Demographic, clinical and laboratory characteristics, medication, and also, hemodialysis prescription data (for those on HD), were recorded on the day of enrollment. For all patients IgG antibodies levels against cytomegalovirus were measured. Patients eligible for the study were Caucasians, adults, with a minimum of 12-month follow-up in our department. All patients should be in stable treatment for the last 3 months, with well controlled hypeparathyroidism and anemia. Patients undergoing hemodialysis were on a chronic maintenance thrice a week program with Kt/V >1.2, with satisfactory regulation of biochemical parameters, according to international guidelines. For rKTx and lKTx patients, eGFR should be >30 mL/min/1.73 m² calculated using the equation CKD-EPI 2021 for creatinine. Patients were excluded in case of a recent infection or vaccination (<3 months), active malignancy, autoimmune, or hematologic disorder (<5 years), or prior immunosuppressive therapy within the past year (excluding steroids, MMF, and cyclosporine or tacrolimus for transplant patients). Moreover, we excluded transplanted patients receiving agents other than those mentioned in the following section, such as everolimus or belatacept, or patients on a corticosteroid sparing scheme. Forty-nine age- and sex-matched healthy volunteers were used as a control group (CG). In all patient groups and the CG, we assessed immune phenotype with flow cytometry in peripheral blood samples, as described in the relevant section.

All transplanted patients had received standardized immunosuppressive treatment according to the Transplant Clinic protocol. Induction therapy consisted of basiliximab or anti-thymocyte globulin (ATG), according to the following. The standard induction regimen consisted of two doses of basiliximab 20 mg (administered 60 min before and on day 4 post-transplant) and two doses of methylprednisolone 500 mg, 12 and 1 h before transplantation. Patients in medium immunological risk, for example patients with history of donor specific antibodies (DSA) positivity, high expected cold ischemia time or marginal donor, received induction therapy with 1.5 mg/kg ATG for 4 days to a total dose of 6 mg/kg. Maintenance therapy included methylprednisolone (125 mg intravenously on day one and then 16 mg orally until day 14, with a gradual tapering until day 42), tacrolimus (plasma levels 6–8 ng/mL for the first year), and mycophenolate mofetil (2 g per day until day 14 and then 1 g daily). Patients with T-cell mediated rejection episodes received ATG, whereas patients with antibody mediated rejection episodes received corticosteroids and/or plasma exchange, depending on the indication.

All study participants signed an informed consent prior to enrollment. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Bioethics Committee of the Aristotle University of Thessaloniki (no. 2348/24-11-2019)

Immune phenotype examination was performed with flow cytometry on whole blood samples collected in K2EDTA tubes, within a short period after collection and in any case in no more than 12 h. Analysis was performed on a Navios EX Flow Cytometer (Beckman Coulter), according to the manufacturer’s instructions, using the following conjugated antibodies: CD45 PC7 (Clone J33), CD3 FITC (Clone UCHT1), CD16‐CD56 PE [CD16: Clone 3G8, CD56: Clone N901(NKH-1)], CD4 APC (Clone 13B8.2), CD8 PC5.5 (Clone B9.11), CD4 FITC (Clone SFCI12T4D11), CD25 PC5 (Clone B1.49.9), and FOXP3 PE (Clone 259D, intracellular staining). Percentages of CD4⁺ (CD3⁺CD4⁺), CD8⁺ (CD3⁺CD8⁺), regulatory (CD4⁺CD25⁺FOXP3⁺) T lymphocytes as well as NK cells (CD3⁻CD16⁺CD56⁺) and CD28⁻ (CD4⁺CD28⁻ and CD8⁺CD8⁻) T cells were determined. The total lymphocyte count was also determined from a blood count in the same sample. Gating strategy is shown in Figure 1 and was manually adjusted in each case.

Singlets discrimination was performed with plotting forward scatter integral (FS-INT) versus forward scatter Time of Flight (FS-TOF). Events deviating from the singlet line were considered cell aggregates and discarded from analysis. Compensation settings were determined before experimental analysis, by using single stained control samples. One sample for each conjugated antibody was analyzed separately to measure the spectral overlap of each fluorochrome’s emission into other detectors and mathematical compensation was applied and adjusted manually to correct the spillover. No viability check was needed, as the analysis was performed in fresh blood samples.

Statistical analysis was performed with Statistical Package for the Social Sciences (SPSS) for Windows, version 25 (IBM, NY, United States). Continuous variables are presented as median (25th – 75th percentile). Quantitative variables were compared with Mann-Whitney U-test between two groups, or with Kruskal-Wallis H-test among multiple groups. The χ² test was used to compare differences in non-quantitative variables between two groups. Correlation between two continuous variables was assessed using the Spearman test. The result was considered significant when R was greater than the critical value for the degrees of freedom. Linear univariate regression was used to predict simple models after logarithmic transformation of the data, and linear multivariate analysis was applied to find independent factors associated with a variable. Statistical significance was considered at a p value of <0.05. Finally, in cases of comparison of more than two different populations, the Bonferroni correction was applied.

Given the use of non-parametric tests between CG, HD and lKTx patients and for a significance level of 0.005 (after Bonferroni correction) with 80% power, the expected effect size for the variable “CD8⁺CD28⁻ T cells percentage” was estimated based on a previous similar study [9], corresponding to a moderate effect (Cohen’s d ≈ 0.48, corresponding to r ≈ 0.23). Based on this effect size, the adequate sample size for the lKTx group was calculated to N = 121 patients. Due to practical limitations, the sample size for lKTx patients (N = 26) in this study is smaller than the calculated one. Accordingly, the findings of the present study referring to the lKTx group should be interpreted with caution and considered exploratory.

In total, 438 patients were included in the study, aged 54 (42–62) years. Out of them 56 were patients with CKD-V, 207 patients on HD for at least 1 year, 149 rKTx patients (1 year post-transplantation), and 26 lKTx patients (>17 years post-transplantation). The sex and median age of the patients did not differ among groups [58 (39–69) years for CKD-V, 55 (43–62) years for HD, 51 (41–57) years for rKTx and 58 (45–66) years for lKTx patients, p = 0.788] The control group included 49 healthy individuals aged 53 (39–64) years. Patients’ demographics are demonstrated in Table 1. In the rKTx group, 22 patients had received ATG as an induction treatment, as described above. Moreover, seven patients had a rejection episode during the first year, 5 out of which had biopsy-proven T cell mediated rejection and received ATG for treatment. All these patients had received ATG for induction. The rest two patients had antibody mediated rejection and were treated with plasma exchange, corticosteroids and intravenous immunoglobulin. In the lKTx group, one patient had a T cell mediated rejection episode and he was treated with ATG both for induction and rejection. Another one patient had an antibody mediated rejection and was also treated with corticosteroids and plasma exchange. Moreover, in the same group, two patients were not on calcineurin inhibitors at the time of flow cytometry analysis, for three and 5 years, because of chronic toxicity. Moreover, three of the patients in the rKTx cohort had a PCR proven CMV reactivation. All of these patients were already CMV seropositive at transplantation.

Table 2 shows the differences of lymphocytic cell subsets in patients’ groups, compared to the CG. In the following comparisons of lymphocyte subpopulations, the level of statistical significance was set at p < 0.005, after Bonferroni correction. Total lymphocyte percentage was reduced in patients with CKD-V compared to CG, 20.2 (14.4–25.8) vs. 29.7 (24.4–38.1)%, respectively, p < 0.001, without further substantial deterioration after dialysis initiation [19.9 (15.3–23.7)% in HD, p = 0.83 compared to CKD-V]. Kidney transplantation resulted in an early increase in total lymphocyte percentage during the first year, 24.7 (20–32.6) vs. 19.9 (15.3–23.7)% in rKTx and HD respectively, p < 0.001, with no increase lKTx 25.5 (21.9–35)%, p = 0.27 compared to rKTx and p = 0.16 compared to CG (Table 2). Similar kinetics were observed in CD4⁺ subpopulation proportion (Figure 2A), however, with CD8⁺ T lymphocytes percentage presenting a gradual upward tendency which was significant in the rKTx group compared to CG, CKD-V and HD [29.6 (24.3–38.3) in rKTx vs. 21.9 (17.1–31.1)% in CG, p < 0.001, vs. 23.3 (20.1–31.9)% in CKD-V p < 0.001, vs. 24.8 (19.1–30.6)% in HD, p < 0.001, (Figure 2B). Similar kinetics were observed in the total counts of total lymphocytes and CD4⁺ T cells, whereas the total count of CD8⁺ T cells was reduced in HD in comparison to HC and peaked in the lKTx group (Supplementary Table S1; Supplementary Figure S1).

HD patients presented a significant increase of NK proportion compared to CG [17.6 (11.8–24.3) vs. 10.9 (9.4–15.8)% respectively, p = 0.009]. In the rKTx group, NK were found to have significantly decreased in comparison to HD [8.9 (5.6–14.9) vs. 17.6 (11.8–24.3)%, p < 0.001], without difference from CG (p = 0.07) and lKTX patients (p = 0.70) (Figure 3A). The count of NK cells was found to reach a nadir in the group of rKTx, but the fluctuations were not significant after Bonferroni correction (Supplementary Table S1; Supplementary Figure S2).

Treg percentage was found to be within normal range in CKD-V patients [6 (3.7–7.9) vs. 5.9 (4.4–7.4)%, for CKD-V and CG, respectively, p = 0.76] but showed a decline after HD initiation, 4.4 (2.9–5.9)%, p < 0.001 compared to CKD-V. One year after kidney transplantation, Treg percentage remained stable [4.2 (3.2–5.3)%, p = 0.30, in comparison to HD, but a further decline was observed in long-term transplantation 2.8 (1.7–3.7)%, p < 0.001 in comparison to rKTx (Figure 3B). The Treg count decreased gradually from CG to HD, following the total lymphocyte decrease. However, a slight but significant increase was observed in Treg count after transplantation, despite the drop in their proportion (Supplementary Table S1; Supplementary Figure S2).

The percentage of CD28⁻ T cell subpopulations increased in HD patients compared to CG, with CD8⁺CD28⁻ increasing already before dialysis initiation, in the latest stages of CKD-V, 38 (24.4–49.8), 49.3 (24.2–63.7) and 45.5 (28.4–58.9)% for CG, CKD-V and HD respectively, p = 0.049 for CKD-V vs. CG and p = 0.026, for HD vs. CG. In the CD4⁺CD28⁻ subpopulation, the increase was observed only after HD initiation, 3.5 (1.4–6.8), 3.8 (1–11.9) and 6 (2.1–13.2)%, for CG, CKD-V and HD respectively, p = 0.64 for CKD-V vs. CG and p = 0.02 for CKD-V vs. HD. However, the above changes were not significant after Bonferonni correction. Kidney transplantation resulted in restoration of CD4⁺CD28⁻ and CD8⁺CD28⁻ to normal levels already from the first year after transplantation. However, in the group of long-term transplanted patients, an increase in both CD28^‐ subpopulations was observed to levels considerably higher even than from those of HD patients, CD4⁺CD28^‐: 12.6 (4.7–27.6) vs. 6 (2.1–13.2)%, for lKTx and HD respectively, p = 0.006, CD8⁺CD28^‐: 68.4 (54.4–90.3) vs. 45.5 (28.4–58.9)% for lKTx and HD respectively, p < 0.001 (Figure 4). The steep increase of CD28⁻ T cell subsets in lKTx patients was also significant when total cell counts were examined (Supplementary Table S1).

Immune phenotype differed significantly between male and female patients in the HD group. Women had an increased total lymphocyte proportion in comparison to men [21.4 (17.5–25.9) vs. 18.1 (14.6–22.4)%, p < 0.001] which was due to increased CD4⁺ T cell proportion [46.5 (38.1–53.8) vs. 42.3 (36–49)%, p = 0.038], but not CD8⁺ T cells [24.3 (20.3–29.2) vs. 25.3 (18.1–31.6)%, p = 0.99)]. Moreover, NK were decreased in women on HD [14.5 (10.3–20.7) vs. 19.3 (14.4–26.6)%, p < 0.001], with no differences in the rest of analyzed cell subsets. Sex remained significant after inclusion of age and CMV status in a multivariate model only for total lymphocytes percentage (95% CI: 0.025–0.321, p = 0.023) and NK (95% CI: -0.587 – -0.071, p = 0.013) but not for CD4⁺ T cells percentage (95% CI: -0.034 – 0.182, p = 0.178). In comparison, in the CG, only CD4⁺ T cell proportion differed between male and female patients [46.2 (39.2–53.6) vs. 43.9 (37.5–47.9)% for females and males respectively, p = 0.028].

Patients’ age was a significant predictor of the percentage of CD8⁺CD28⁻ T cells in all patient groups studied, as well as in healthy volunteers, except for the lKTx group (Figure 5). The linear regression data are shown in Table 3. In kidney transplant patients, both rKTx and lKTx, age was an independent predictor of the percentage of CD4⁺CD28⁻ T cells (Table 3). In the rKTx group, age was an independent predictor for several T cell populations, such as total lymphocytes percentage (r = −0.31, R² = 0.097, 95% CI: −0.275 – −0.088, p < 0.001), NK percentage (r = 0.26, R² = 0.071, 95% CI: 0.066–0.276, p = 0.002) CD4⁺ T lymphocytes count (r = −0.4, R² = 0.16, 95% CI: −15.34 – −6.92, p < 0.001) and Tregs count (r = −0.31, R² = 0.09, 95% CI: −0.685 – −0.216, p < 0.001). In the lKTx group, NK proportion depended also on the age (r = 0.49, R² = 0.21, 95% CI: 0.073–0.643, p = 0.016).

Anti-CMV IgG antibodies levels could predict the percentage of CD4⁺CD28⁻ and CD8⁺CD28⁻ T lymphocytes in HD patients, (r = 0.29, R² = 0.08, 95% CI: 0.001–0.007, p = 0.011 and r = 0.28, R² = 0.08, 95% CI: 0.000–0.03, p = 0.015 respectively). However, in multivariate regression including age, the antibody levels did not remain statistically significant for the prediction of CD8⁺CD28⁻ T lymphocytes and age remained the only independent factor. Correlation of anti-CMV IgG antibody levels with the percentage of CD4⁺CD28⁻ T lymphocytes, (r = 0.31, R² = 0.1, 95% CI: 0.002–0.007, p = 0.002) was also observed in rKTx patients but not in the lKTx group (r = 0.071, R² = 0.1, 95% CI: -0.002 – 0.029, p = 0.072). In multivariate analysis, both antibody levels and age remained independent factors for the percentage of CD4⁺CD28⁻ T lymphocytes. Finally, eGFR was not a predictor of the percentage of CD28⁻ T lymphocytes in either of the groups of transplanted patients.

Dialysis vintage (DV) was associated negatively to the total lymphocytes proportion in HD patients (r = −0.18, R² = 0.035, 95% CI: -0.002 – 0.000, p = 0.016), affecting mainly CD4⁺ T cells (r = −0.22, R² = 0.052, 95% CI: −0.002 – 0.000, p = 0.003), but not CD8⁺ T cells (r = 0.06, R² = 0.004, 95% CI: −0.001 – 0.001, p = 0.468). NK were positively associated to DV (r = 0.19, R² = 0.038, 95% CI: 0.000–0.004, p = 0.017), while Tregs decreased as DV increased (r = −0.17, R² = 0.028, 95% CI: -0.003 – 0.000, p = 0.031). CD4⁺CD28⁻ T cells proportion was also affected (r = 0.17, R² = 0.03, 95% CI: 0.000–0.009, p = 0.035), but not CD8⁺CD28⁻ T cells proportion (r = −0.08, R² = 0.007, 95% CI: -0.001 – 0.003, p = 0.297). However, in the multivariate models including sex, age and CMV status, DV was significant for none of the analyzed cell subsets. This effect of DV in total lymphocytes and CD4⁺ T cells was also retained during the first year of transplantation, but was not significant in multivariate analysis or lKTx patients.

Estimated GFR was a predictor of CD4⁺ and CD8⁺ T cell proportion (r = 0.21, R² = 0.045, 95% CI: 0.001–0.005, p = 0.012 and r = 0.17, R² = 0.029, 95% CI: −0.006 – 0.000, p = 0.043, respectively) and Tregs count (r = 0.21, R² = 0.047, 95% CI: 0.002–0.012, p = 0.01) in rKTx patients, but this correlation did not remain significant in multivariate analysis including age and anti-CMV antibodies levels.

History of ATG treatment in the transplant patient groups had impact on phenotype of T cells. In the rKTx groups N = 22 patients had received ATG for induction therapy. These patients had a lower CD4⁺ T cell proportion in comparison to patients who had not received ATG [36.9 (29.6–45.9) vs. 47.7 (40.2–53.9)% respectively, p = 0.003], as well as more CD8⁺ T cells [38 (25.9–46.3) vs. 28.9 (24.1–36.8)% respectively, p = 0.037], and more CD8⁺CD28⁻ T cells [49 (37–69) vs. 37 (24–57)% respectively, p = 0.013] (Supplementary Table S2). However, in the multivariate linear model with age, sex and CMV status as confounding factors, ATG treatment significantly affected only the CD4⁺ T cell proportion (95% CI: -0.367 – -0.09, p = 0.002), along with age, but neither of CD8⁺ or CD8⁺CD28⁻ T cells, and age was a significant factor for CD8⁺CD28⁻ T cells percentage. In the lKTx group no difference was found in the immune phenotype according to ATG treatment (N = 5 patients had received ATG, data on Supplementary Table S2). In order to increase the generalizability of the study and eliminate the impact of ATG treatment on lymphocytic subsets, we performed the comparisons of the examined subsets among the patients groups, after excluding the ATG treated patients. The kinetics of subpopulations were not substantially altered, as shown in the Supplementary Table S3.