Effector Antitumor and Regulatory T Cell Responses Influence the Development of Nonmelanoma Skin Cancer in Kidney Transplant Patients

Abstract

Background 

Chronic immunosuppression promotes nonmelanocytic squamous cell carcinoma (SCC) after kidney transplantation. Adaptive and innate immunity play a key role controlling tumor growth and are influenced by different immunosuppressive agents. We hypothesized that functional impairment of tumor-specific T cell responses due to calcineurin inhibitors (CNI) could contribute to SCC development, whereas conversion to mammalian target of rapamycin inhibitors (mTOR-i) could recover this protective immune response.

Methods 

Peripheral tumor-specific T cell responses against main SCC-derived antigens using the IFN-γ enzyme-linked immunospot assay and intratumor (IT) and circulating immune phenotypes (CD4 + T, CD8 + T, CD20 + B, CD56 + NK, FOXP3 + regulatory T [Treg] cells) were explored in a cross-sectional analysis in 59 kidney transplant patients with SCC on CNI (KT-CNI-SCC) or mTOR-i (KT-mTORi-SCC), 25 nontransplants developing SCC (NoKT-SCC) and 6 healthy controls. Moreover, 25 KT-CNI-SCC were switched to mTOR-i and evaluated after 12 months.

Results 

Kidney transplant patients showed lower IT infiltrates and tumor-specific T cell responses than NoKT-SCC, and intratumoral and circulating FOXP3 + Treg cells were higher in KT-mTORi-SCC (P < 0.05). Tumor-specific T cell responses were significantly lower in KT-CNI-SCC than KT-mTORi-SCC and NoKT-SCC and predicted SCC relapses (area under the curve = 0.837; P < 0.05). One-year after mTOR-i conversion, a significant increase in FOXP3 + Treg cell numbers and tumor-specific T cell responses were observed, reaching similar levels than KT-mTORi-SCC and NoKT-SCC patients.

Conclusions 

Tumor-specific T cell responses are strongly impaired in CNI-treated patients but recover after mTOR-i conversion, reducing SCC relapses.

In Brief

In a cross-sectional study of 59 kidney transplant recipients with a squamous cell carcinoma (on CNI or mTOR inhibitor-based immunosuppression), the authors demonstrate weaker tumor-specific T cell responses in CNI-treated patients who recover after conversion to mTOR inhibitors.

Kidney transplant recipients exhibit significantly higher incidence of malignancies with a much more aggressive behavior than the general population.^1-3 Nonmelanocytic skin cancers (NMSC) and particularly squamous cell carcinomas (SCC) are the most common cancers, occurring in up to 50% of solid organ transplant patients.^4-7 There are several well-characterized risk factors influencing the development of SCC, such as sun exposure, previous SCC, age at transplantation and duration and burden of immunosuppression.^8-10

Modulation of the immune system seems to significantly differ according to different immunosuppressants. In fact, patients on calcineurin inhibitors (CNI) and azathioprine (AZA) more likely develop posttransplant malignancies,^11,12 whereas transplant recipients receiving inhibitors of the mammalian target of rapamycin (mTOR-i) display significantly lower risk of malignancies in general^13-15 and of NMSC if used either as main immunosuppressive agent¹⁶ or after conversion from an established CNI-based immunosuppressive regimen.^4,17,18 Remarkably, although the rationale for using mTOR-i to treat malignancies has basically focused on the direct growth inhibitory effect on tumor cells, the relevant pleiotropic immune modulatory properties of mTOR-i renders these immunosuppressive agents of great value for being investigated in the setting of organ transplantation.

Innate and adaptive immunities have been shown to play a major role in cancer growth and progression. In this regard, high numbers and functionally active natural killer (NK) cells have been described to provide a protective role in the control of tumor growth and metastasis both in animal models as well as in humans.^19-21 Importantly, tumor-specific effector T cell responses have also been described to be essential for the protection against the development and dissemination of malignancies^22,23 and the induction and expansion of effective cytotoxic antitumor T cells has become a major goal of vaccines against cancer.^24,25 Conversely, Foxp3+ regulatory T (Treg) cells are known to facilitate the emergence and spread of certain types of malignancies, including SCC.^26-31

Herein, we postulated that a functional impairment of tumor-specific T cell responses due to chronic CNI exposure could strongly contribute to SCC development and recurrence but such deleterious immune state could potentially be attenuated after conversion to mTOR-i. Therefore, we designed a prospective, one-arm, interventional pilot study in which we assessed changes in different innate and adaptive immune cell phenotypes both circulating and within tumor tissue as well as in tumor-specific T cell responses against main SCC-expressing antigens in kidney transplant patients developing SCC receiving CNI drugs and at 1 year after mTOR-i conversion. Although we confirm that the number of IT Foxp3 + Treg cells and NK cells correlate with the risk of developing new SCC in CNI-treated patients, a significant functional impairment of antitumor T cell immunity was observed, which was recovered 12 months after mTORi conversion.

MATERIALS AND METHODS

Study Design and Patients of the Study

Ninety patients were evaluated in this study. All patients included in the study gave written informed consent to participate, and the study was approved by the IRB at Bellvitge University Hospital and at Hospital Germans Trias i Pujol (ethics committee number EPA022/11). As shown in Figure 1, the study was divided in 2 phases. In the first phase, we made a cross-sectional analysis of different circulating and IT immune-phenotypes as well as tumor-specific effector T cell responses in a group of 59 kidney transplant patients with history of nonmelanoma SCC receiving either CNI (KT-CNI-SCC; n = 39) or mTOR-i–based immunosuppression (KT-mTORi-SCC; n = 20). KT-mTORi-SCC had been on a mTOR-i immunosuppressant without any CNI drug for at least 5 years before the development of the SCC. KT-mTORi-SCC had been on SRL, some since the initial period after transplantation and other after conversion due to CNI-related nephrotoxicity. The different immunosuppressive drugs combinations as well as drug trough levels are shown in Table 1. Also, 25 nontransplant patients with SCC (NoKT-SCC) and 6 healthy controls (HC) were evaluated as nonimmunosuppressed controls.

FIGURE 1: Flowchart of the study.

TABLE 1: Immunosuppressive regimens and trough levels in KT-SCC patients at time of the study

In the second phase of the study, 25 of the 39 patients receiving CNI-based immunosuppression with SCC (KT-CNI-SCC) who satisfied clinical criteria for safety conversion were switched to the mTOR-i Everolimus (Certican, Novartis, Basel, Switzerland) and assessed at 1 year for changes on circulating immunephenotypes and tumor-specific T cell responses. Patients eligible to be converted to a mTOR-i after the diagnosis of SCC, were those showing a stable allograft function, low levels of proteinuria (<1 g/d) and no recent episodes of allograft rejection. Five of 25 were switched back to the previous CNI-based medication due to drug-related adverse events (2 peripheral edemas, 1 severe hypercholesterolemia, 1 progressive proteinuria, and 1 mTOR-i–related pneumonitis); therefore, 20 KT-CNI-SCC could be studied 1 year after mTOR-i conversion. Patients of the study were recruited at Bellvitge University Hospital and at Hospital Germans Trias i Pujol in Barcelona, Spain. SCC lesions were treated by surgical excision with margins in all cases. The study was approved by both local Ethics Committees and conducted according to the Declaration of Helsinki and Istanbul.

Immune Cell Phenotypes of Skin Cancer Infiltrates

Samples from SCC biopsies were collected for phenotypic analysis of different immune cellular infiltrates. SCC paraffin-embedded samples were stained by a staining enzyme immunoassay technique using the FoxP3 (clone 236A/E7; Abcam, Cambridge, UK), CD8 (clone C8/144B; Dako UK Ltd., Ely, UK), CD20 (clone L-26; Master Diagnostica, Granada, Spain), and CD56 (clone BC56C04; A. Menarini Diagnostics UK, Wokingham, UK) antibodies following a previously established protocol.^31,32 The number of positive FoxP3+, CD8+, CD20+, and CD56+ cells in all tumoral areas were counted using symmetric square high-resolution power fields (number of cells/field (mm³)) (×400; Leica Geosystems, Barcelona, Spain), (Figure S1, SDC, https://links.lww.com/TP/B428).

Phenotypic Characterization of Circulating Lymphocytes in Peripheral Blood

Blood samples were collected and peripheral blood mononuclear cells (PBMCs) were isolated by standard Ficoll density gradient centrifugation and cryopreserved for further functional and phenotypic studies. Different lymphocyte subpopulations were characterized by flow cytometry, as previously described.³³ Multiple combinations of monoclonal antibodies (Becton Dickinson, Mountain View, CA) were used for staining PBMCs and were analyzed by a FACSscan CANTO II (Becton Dickinson). Cell markers used were CD3+, CD4+ and CD8+ for T cells, CD56+ for Natural-Killer cells, CD19+ for B lymphocytes, and CD4+ CD25+ CD127− FoxP3+ for Treg cells. The acquired data were analyzed using the FACS DIVA software (BD Biosciences, San Jose, CA). The total lymphocyte count from routine hematology laboratory testing was used to calculate absolute cell counts (Figure S2, SDC,https://links.lww.com/TP/B428). Data are given in number of cells/μL.

Assessment of Tumor-Specific T Cell Responses in Peripheral Blood

Frequencies of effector T cell responses were evaluated against different tumor-specific peptides expressed by epithelial cell carcinomas such as in SCC. Peptides tested were 15 amino acid length sequences with 11 amino acid overlap of melanoma antigen-encoding genes 1 and 3 (MAGE-A1, Swiss-Prot ID: P43355 and MAGE-A3, Swiss-Prot ID: P43357),³⁴ p53 (Swiss-Prot ID: Q2XN98),³⁵ Survivin (Swiss-Prot Acc. no. O15392),³⁶ and hTERT (UniProtKB Acc. no. O14746)³⁷ antigens (Miltenyi Biotec, Germany). Tumor-specific T cell responses were evaluated using an IFN-γ enzyme-linked immunospot (ELISPOT) assay, and frequencies of antigen-specific IFN-γ–producing T cells were counted as previously described.³⁸ Briefly, 3 × 10⁵ PBMCs were stimulated with cancer-associated peptides (1 μg/mL) in duplicate wells. Also, PBMCs were added together with medium plus DMSO (Amresco, Solon OH) (4 μL/mL) alone as a negative control and with PHA (Sigma-Aldrich, Madrid, Spain) as a positive control. The resulting spots were counted using a computer-assisted ELISPOT reader (Autoimmune Diagnostika, AID Elispot Reader, Germany). Results were expressed as number of cells secreting IFN-γ per 3 × 10⁵ after subtracting the number of spots observed in the respective negative controls. Frequencies of IFN-γ–producing cells against mean ELISPOT values against all 5 tumor antigens were also assessed.

Statistical Analysis

Data are presented as mean ± standard deviation. The appropriate test for each analysis, depending on whether the variable conditions met normality or not, was used. Groups of patients were compared using the χ² test for categorical variables and the one-way ANOVA analysis of variance (using Bonferroni correction to make the statistical analysis much more stringent) or Student t test for normally distributed data for quantitative variables, and the nonparametric Kruskal-Wallis or Mann-Whitney U test for non-normally distributed variables. Bivariate correlation analyses were done using Pearson or Spearman test for nonparametric variables. A sensitivity/specificity receiver operating characteristic (ROC) curve test was done to determine the predictive probability of the combination of tumor-specific effector T cell responses against all SCC antigens for predicting the advent of tumor relapse. The statistical significance level was defined as 2-tailed P less than 0.05.

RESULTS

Study Population

Figure 1 shows the flowchart of the study design. Main demographic variables of the patients are depicted in Table 2. Nontransplant patients with SCC (NoKT-SCC) and HC were matched for age and sex to all transplant recipients with SCC (KT-SCC). Likewise, KT-SCC receiving either mTOR-i or CNI drugs were comparable regarding main clinical characteristics, such as cause of chronic kidney disease (CKD) needing previous immunosuppressive therapy, such as primary glomerulonephritis or renal vasculitis, number of previous transplants, cumulative time of immunosuppression, use of polyclonal antibodies, number of donor/recipient HLA mismatches, and incidence of biopsy-proven acute rejection (BPAR) requiring rescue immunosuppressive treatment. Differently to KT-CNI-SCC patients, only 2 KT-mTORi-SCC developed 1 new tumor 13 months after the first diagnosis, whereas KT-CNI-SCC displayed significantly higher SCC relapses (mean, 2.31 ± 2.1; range, 0-13). No differences were observed among KT-CNI-SCC and KT-mTORi-SCC patients regarding tumor parameters, such as tumor diameter, depth, and Clark level, but NoKT-SCC patients displayed higher tumor size and depth as well as Clark level as compared with transplant individuals.

TABLE 2: Baseline demographics

Tumor Infiltrating and Circulating Immune Cell Subsets in Kidney Transplant Patients and Nontransplant Patients Developing SCC

A different but consistently detectable degree of all type of cellular infiltrates from the adaptive and innate immunity (Foxp3 + Treg cells, CD8 + T cells, CD20 + B cells, CD56 + NK cells) was observed in SCC tumor biopsies in all patients. As illustrated in Figures 2A and B, a positive correlation was observed among all cell subsets infiltrating tumor tissues, illustrating an intensive local inflammatory response in both KT-SCC and NoKT-SCC patients. Nonetheless, NoKT-SCC showed significantly higher IT cellular infiltrates of all cell subsets than transplant individuals (KT-SCC) (Figure 2C).

FIGURE 2: Analysis of tumor-cell infiltrates in patients with SCC. A, Positive correlation between all different cell subtypes infiltrating tumor tissue samples in KT-SCC. IT CD8+ T cells numbers positively correlated with CD20+ B cells (P = 0.009), FoxP3 + Treg cells (P = 0.006), CD56+ NK cells (P = 0.072), and number of CD20+ B cells correlated with FoxP3+ cell subsets (P = 0.019), R = 0.25. B, Positive correlation between different IT cell subsets in NoKT-SCC. A positive correlation was observed between CD8+ T cell numbers and CD20+ B cells (P = 0.002), FoxP3+ Treg cells (P < 0.001), CD56+ NK cells (P = 0.021), as well as between CD20+ and FoxP3+ Treg cells (P = 0.001) (R = 0.3) in NoKT-SCC. 2C: NoKT-SCC showed significantly higher IT cellular infiltration compared to KT-SCC (74.67 ± 45.43 vs 31.63 ± 33.54, P = 0.003 for CD8+ T cells; 110.33 ± 134 vs 38.12 ± 46, P = 0.05 for CD20 B cells; 28.87 ± 27.41 vs 15.71 ± 13.07, P = 0.013 for FoxP3 + Treg cells and 4.46 ± 2.55 vs 4.08 ± 4.69 for CD56+ NK cells, P = 0.06).

Conversely, no differences were observed in the percentage of CD4+ and CD8+ T cells, CD20+ B cells, CD56 + NK cells, and FoxP3 + Treg cells circulating in peripheral blood between KT-SCC, NoKT-SCC, and HC (Figure S3, SDC,https://links.lww.com/TP/B428).

MTORi-Treated Patients and Nontransplant Individuals Developing SCC Show Higher IT and Circulating Foxp3 + Treg Cells Than Patients Developing SCC on CNI-Based Immunosuppression

KT-CNI-SCC and KT-mTORi-SCC showed similar CD20 + B and CD56 + NK cells within IT infiltrates, whereas infiltrating CD8+ T cell numbers were higher in KT-mTORi-SCC (38.41 ± 38.88 CD8 + T cells vs 37.47 ± 60.73 CD20 + B cells, P = 0.33; 3.79 ± 5.13 vs 4.73 ± 3.57 CD56 + NK cells, P = 0.52, and 24.32 ± 24.35 vs 48.20 ± 45.10 CD8+ T cells, P = 0.02, in KT-CNI-SCC and KT-mTORi-SCC patients, respectively) (Figure 3A). Likewise, both groups of transplant patients displayed similar numbers of circulating CD20 + B cells, CD56 + NK cell as well as CD8+ and CD4+ T cell subsets (561.55 ± 269.94 vs 404.77 ± 235.08 CD8+ T cells, P = 0.14; 673.03 ± 424.86 vs 724.58 ± 449.61 CD4+ T cells, P = 0.77; 75.72 ± 55.21 vs 74.60 ± 63.88 CD20 + B cells, P = 0.96 and 1080.65 ± 497.41 vs 805.75 ± 402.90 CD56 + NK cells, P = 0.15 in KT-CNI-SCC and KT-mTORi-SCC patients, respectively) (Figure 3B).

FIGURE 3: Analysis of cellular infiltrates and circulating in peripheral blood in kidney transplant patients receiving CNI or mTORi-based immunosuppression. A, No differences were observed in IT CD20 + B cells and CD56 + NK cells but different CD8 + T cell infiltrates between KT-CNI-SCC and KT-mTORi-SCC patients. B, No differences in CD4+ and CD8+ T cells, CD20 + B cells and CD56 + NK cells subsets between KT-CNI-SCC and KT-mTORi-SCC patients in peripheral blood. C, The numbers of tumor-infiltrating Foxp3 + Treg cells in KT-mTORi-SCC and NoKT-SCC patients were significantly higher than that in KT-CNI-SCC. D, The numbers of circulating Foxp3 + Treg cells in KT-mTORi-SCC and NoKT-SCC patients were significantly higher than that in KT-CNI-SCC.

Conversely, KT-CNI-SCC showed significantly lower numbers of Foxp3 + Treg cells both within tumor infiltrates and in peripheral blood as compared to KT-mTORi-SCC (22.40 ± 18.07 vs 12.76 ± 8.98, P = 0.05 within tumor infiltrates and 12.10 ± 6.88 vs 5.50 ± 3.33, P = 0.037 in peripheral blood for KT-mTORi-SCC and KT-CNI-SCC, respectively) and than NoKT-SCC (28.87 ± 27.40 vs 12.76 ± 8.98 P = 0.042, within tumor infiltrates and 10.77 ± 6.88 vs 5.50 ± 3.33, P = 0.045 in the circulation for noKT-SCC and KT-CNI-SCC, respectively) (Figures 3C-D).

CNI-Treated Patients Developing SCC Display Significantly Weaker Tumor-Specific T Cell Responses Than Patients on mTOR-i and Than Nontransplant Patients Developing SCC

Tumor-specific effector T cell responses against main SCC-derived antigens (MAGE-A1, MAGE-A3, p53, Htert, and Survivin) were detectable in all patients developing SCC, regardless of the type of immunosuppression, but not in HC (Figure 4A). KT-mTORi-SCC showed higher T cell responses than KT-CNI-SCC against most SCC antigens (P = 0.08, P = 0.106, P = 0.172, P = 0.05, and P = 0.202 for MAGE-A1, MAGE-A3, p53, hTERT, and Survivin, respectively). NoKT-SCC patients showed similarly high antitumor T cell frequencies as compared with KT-mTORi-SCC, but significantly higher than KT-CNI-SCC (P = 0.05, P = 0.10, P = 0.05, P = 0.05, and P = 0.08 for MAGE-A1, MAGE-A3, p53, hTERT, and Survivin, respectively). Remarkably, mean frequencies of IFN-γ–producing T cells against all 5 tumor antigens within KT-mTORi-SCC were significantly more robust than that in KT-CNI-SCC (32.35 ± 28.63 vs 6.7 ± 6.13 mean IFN-γ spots/3 × 10⁵ PBMC against all SCC antigens, P < 0.001 in KT-mTORi-SCC and KT-CNI-SCC, respectively) and numerically higher than NoKT-SCC, although not statistically different (32.35 ± 48.63 vs 19.64 ± 12.56 mean IFN-γ spots/3 × 10⁵ PBMC, P = 0.87 in KT-mTORi-SCC and NoKT-SCC, respectively). Of note, tumor-specific T cell frequencies in NoKT-SCC were also significantly higher than that in KT-CNI-SCC (19.64 ± 12.56 vs 6.7 ± 6.13 mean IFN-γ spots/3 × 10⁵ PBMC against all SCC antigens, P = 0.03) (Figure 4B).

FIGURE 4: Tumor-specific IFN-γ producing T cell frequencies against SCC antigens. Tumor-specific effector T cell responses against main SCC-derived antigens (MAGE-A1, MAGE-A3, p53, Htert, and Survivin) were detectable in all patients developing SCC, regardless of the type of immunosuppression, but not in HC. A, In general, KT-mTORi-SCC showed higher T cell responses than KT-CNI-SCC against most SCC antigens (40.50 ± 54.53 vs 11.25 ± 10.16, P = 0.08; 38.25 ± 57.69 vs 9.15 ± 10.40, P = 0.106; 38 ± 71.99 vs 7.25 ± 13.74, P = 0.172; 23.5 ± 35.90 vs 1 ± 3.46, P = 0.05 and 21.5 ± 43.71 vs 4.25 ± 6.57 IFN-γ spots/3 × 10⁵ PBMC, P = 0.202 for MAGE-A1, MAGE-A3, p53, hTERT and Survivin, respectively). NoKT-SCC patients showed similarly high antitumor T cell frequencies as compared to KT-mTORi-SCC (31.81 ± 32.89, 24.50 ± 27.91, 23.60 ± 23.49, 10.30 ± 15.59 and 13.80 ± 16.84 IFN-γ spots/3 × 10⁵ PBMC, P = NS for MAGE-A1, MAGE-A3, p53, hTERT and Survivin, respectively) but significantly higher than KT-CNI-SCC (11.25 ± 10.16 vs 31.81 ± 32.89, P = 0.05; 9.15 ± 10.40 vs 24.50 ± 27.91, P = 0.10; 7.25 ± 13.74 vs 23.60 ± 23.49, P = 0.05; 1 ± 3.46 vs 10.30 ± 15.59, P = 0.05 and 4.25 ± 6.57 vs 13.80 ± 16.84 IFN-γ spots/3 × 10⁵ PBMC, P = 0.08 for MAGE-A1, MAGE-A3, p53, hTERT and Survivin, respectively). HC individuals showed almost no detectable T cell responses against SCC-derived antigens (5 ± 4.89, 3 ± 1.89, 4.5 ± 4.13, 2 ± 2.44 and 1.5 ± 1.64 IFN-γ spots/3 × 10⁵ PBMC for MAGE-A1, MAGE-A3, p53, Htert and Survivin, respectively). B, Antitumor effector T cell responses in patients developing SCC. Mean frequencies of IFN-γ–producing cells against SCC antigens were significantly more vigorous among KT-mTORi-SCC and NoKT-SCC patients than that in KTCNI-SCC. ^a P < 0.001 between HC and all other groups.

SCC Relapse in CNI-Treated Patients Is Associated With Impaired Tumor-Specific T Cell Responses

When the impact of tumor-specific effector T cell responses was evaluated in relation to the incidence of tumor relapses, transplant patients with tumor relapses showed significantly lower antitumor IFN-γ–producing T cell frequencies than those without (6 ± 2.6 vs 12.77 ± 1.68 mean IFN-γ spots/3 × 10⁵ PBMC against all SCC antigens, P = 0.001) (Figure 5). Indeed, tumor-specific T cell responses against all SCC antigens were lower in patients with SCC relapses as compared with patients without tumor recurrence (Figure S4, SDC,https://links.lww.com/TP/B428).

FIGURE 5: Antitumor effector T cell responses and tumor relapse in KT-CNI-SCC patients. A, Mean antitumor T cell responses were significantly higher in patients with SCC relapses as compared to patients without. B, ROC curve analysis of tumor-specific T cell responses for the prediction of SCC relapses. Frequencies of tumor-specific T cell responses predicting the presence of SCC relapse (AUC, 0.714; 95% CI, 0.423-1; P = 0.18; AUC, 0.694; 95% CI, 0.393-0.994; P = 0.22; AUC, 0.633; 95% CI, 0.326-0.940; P = 0.40; AUC, 0.510; 95% CI, 0.197-0.824; P = 0.16; AUC, 0.541; 95% CI, 0.225-0.857; P = 0.16 for MAGE-A1, MAGE-A3, P53, Htert, and Survivin, respectively). C, Combined predictive probability of SCC relapse of all tumor-specific T cell responses. The predictive probability of tumor relapse combining all tumor-antigen T cell responses in a binary logistic regression analysis showed high accuracy (AUC, 0.837; 95% CI, 0.624-0.99; P = 0.035). ROC, receiver operating characteristic.

Furthermore, KT-CNI-SCC showed a positive correlation between the number of tumor-infiltrating Foxp3 + Treg cell and the number of tumor relapses (R = 0.6, P = 0.001), whereas a weak but negative correlation was also found with the number of tumor-infiltrating CD56 + NK cells (R = −0.2; P = 0.001). Conversely, we did not find any association between the number of all cellular phenotypes circulating in peripheral blood and the number of tumor relapses (data not shown).

Although the number of Foxp3 + Treg cells and CD56 + NK cells either infiltrating the tumor or in the circulation did not accurately predict the risk of SCC relapses (area under the curve [AUC] < 0.6) (data not shown), the frequencies of tumor-specific T cell responses performed significantly better (AUC > 0.7) (Figure 5B). We then determined the predictive probability of tumor relapse combining all tumor-antigen T cell responses in a binary logistic regression analysis. As illustrated in Figure 5C, the combination of the overall antitumor T cell responses significantly increased the accuracy for new SCC development (AUC, 0.837; 95% confidence interval [CI], 0.624-0.99; P = 0.035). In addition, no correlation was found in tumor infiltrates, numbers of different cell subpopulations in peripheral blood, and tumor-specific T cell responses according to different tumor relapse stratification (none, 1-3 and >3 relapses) between KT-CNI-SCC and KT-mTORi-SCC patients (data not shown).

MTOR-i Conversion Induces a Significant Increase of Circulating Foxp3 + Treg Cells and Tumor-Specific T Cell Responses in Kidney Transplant Patients

Twenty of the 39 KT-CNI-SCC patients could be successfully switched to an mTORi-based regimen and evaluated after 1 year. As shown in Table 3, there were no BPAR events, and allograft function remained stable over the 12-month period of follow-up. Interestingly, only 1 patient developed an SCC relapse at month 9 after conversion.

TABLE 3: Baseline demographics

No significant changes were observed in circulating CD4+ and CD8+ T cells, CD20 + B cells and CD56 + NK cells at 12 months after mTOR-i conversion (data not shown). Conversely, the numbers of Foxp3 + Treg cells significantly increased as compared with baseline (5.50 ± 3.33 vs 10.08 ± 5.52, P = 0.035), achieving the same levels than NoKT-SCC (10.08 ± 5.52 vs 10.77 ± 6.88, P = 0.80) and KT-mTORi-SCC patients (10.08 ± 5.52 vs 12.10 ± 6.88, P = 0.71), but numerically higher than HC (10.08 ± 5.52 vs 5.94 ± 4.77, P = 0.07) (Figure 6A). Tumor-specific T cell responses against all SCC antigens significantly increased in KT-CNI-SCC as compared with baseline 12 months after mTORi conversion (11.25 ± 10.16 vs 25.9 ± 18.06, P = 0.02; 9.75 ± 10.40 vs 23.45 ± 18.09, P = 0.03; 7.25 ± 13.74 vs 16.36 ± 14.78, P = 0.14; 1 ± 3.46 vs 6 ± 8.59, P = 0.07; 4.25 ± 6.57 vs 5.72 ± 8.84, P = 0.65 IFN-γ spots/3 × 10⁵ PBMC for MAGE-A1, MAGE-A3, P53, Htert, and Survivin) (Figure 6B).

FIGURE 6: Changes in circulating Foxp3 + Treg cells and antitumor T cell responses in KT-CNI-SCC at 1 year after conversion to mTor-i as compared with KT-mTori-SCC, NoKT-SCC and HC. A, Changes in peripheral blood immunophenotype after mTORi conversion. B, Changes of tumor-specific effector T cell responses against SCC-derived antigens after mTORi conversion. Tumor-specific T cell responses in KT-CNI-SCC significantly increased as compared with baseline 12 months after mTORi conversion (11.25 ± 10.16 vs 25.9 ± 18.06, P = 0.02; 9.75 ± 10.40 vs 23.45 ± 18.09, P = 0.03; 7.25 ± 13.74 vs 16.36 ± 14.78, P = 0.14; 1 ± 3.46 vs 6 ± 8.59, P = 0.07; 4.25 ± 6.57 vs 5.72 ± 8.84, P = 0.65 IFN-γ spots/3 × 10⁵ PBMC for MAGE-A1, MAGE-A3, P53, Htert and Survivin, at baseline and at 12-month after conversion, respectively), whereas no differences were observed as compared to KT-mTORi-SCC and NoKT-SCC. C, Changes of effector T cell responses against tumor antigens after mTORi conversion. Mean antintumor T cell responses against all tumor antigens in SCC significantly increased over baseline responses (under CNI) 12 months after conversion.

Furthermore, at 12-month after mTOR-i conversion, a significant generalized recovery of antitumor T cell responses was observed (15.49 ± 9.39 vs 6.7 ± 6.13, P = 0.014 IFN-γ spots/3 × 10⁵ PBMC), reaching similar antitumor T cell frequencies as NoKT-SCC (15.49 ± 9.39 vs 19.64 ± 12.56 IFN-γ spots/3 × 10⁵ PBMC, respectively, P = 0.25) and closer to KT-mTORi-SCC (15.49 ± 9.39 vs 32.35 ± 48.63 IFN-γ spots/3 × 10⁵ PBMC, respectively, P = 0.26) (Figure 6C).

DISCUSSION

Nonmelanoma SCC is the most frequent cancer after solid-organ transplantation, and it portends relevant morbidity and negatively impacts the quality of life of kidney transplant recipients.^4,5 Although the advent of SCC has classically been linked to a state of overimmunosuppression, no data have been reported so far about the impact of tumor-specific effector T cell responses in kidney transplant patients developing SCC under different immunosuppression. In this work, we show for the first time that antitumor effector T cell immunity is significantly compromised in kidney transplant patients receiving chronic immunosuppression based on CNI drugs as compared with kidney transplant recipients treated with mTOR-i and as compared with nonimmunosuppressed individuals also developing SCC. Remarkably, although on the one hand, we confirm previous findings showing the differential expression of Foxp3 + Treg cells and CD56 + NK cells both locally within the tumor tissue as well as circulating in peripheral blood, we here also show that 1 year after mTOR-i conversion, tumor-specific effector T cell frequencies significantly recover, achieving similar levels as non-immunosuppressed individuals and as kidney transplant recipients developing SCC under mTOR-i. Of note, frequencies of tumor-specific effector T cells against main SCC-expressing antigens may serve as a predictive biomarker for the risk of tumor relapse among CNI-treated kidney transplant patients.

Recent studies have shown the predictive value of different immune phenotypes either circulating or directly within tumor tissue for subsequent tumor recurrence.^32,39 In our work, we also observed an important active inflammatory process within the tumor tissues, involving diverse cell subsets from both the innate and adaptive immunities. Interestingly, as compared with nonimmunosuppressed individuals also developing SCC, kidney transplant patients showed significantly lower cellular infiltrates, suggesting a direct effect of burden immunosuppression reducing the inflammation degree. Of note, when transplant patients with SCC were analyzed according to the type of immunosuppression, mTORi-treated patients displayed significantly higher Foxp3 + Treg cell numbers both infiltrating the tumor and also in the circulation as compared with patients on CNI drugs. In fact, when CNI-treated patients with SCC were evaluated 1 year after conversion to mTOR-i, the number of Foxp3 + Treg cells achieved similar levels than transplant recipients with SCC on mTOR-i and significantly higher than healthy individuals. Altogether, these findings support the notion of the capacity of mTOR-i to expand Foxp3 + Treg cells both in vivo and in vitro.^33,40,41 Furthermore, and in agreement with previous published data, we found a positive correlation between the number of tumor-infiltrating Foxp3 + Treg cells and the number of SCC relapses, whereas a negative correlation between the number of CD56 + NK cells and the number of SCC relapses in patients on CNI.^28,29 Interestingly, in a similar approach to our study, although in a smaller cohort of kidney transplant patients, Carrol et al³⁰ evaluated the kinetics of Foxp3 + Treg cells and CD56 + NK cells in peripheral blood 6 months after mTOR-i conversion from a CNI- or AZA-based regimen. Interestingly, conversion to mTOR-i revealed that high preexisting Foxp3 + Treg cell phenotype could eventually predict those transplant recipients who continue to develop SCC.

Although the main rationale for using mTOR-i to treat malignancies has focused on their direct, growth-inhibitory effects, the expansion, and generation of antigen-specific memory effector T cells under the influence of mTOR-i has gained considerable interest in the last years. Here, using the sensitive IFN-γ ELISPOT assay,³⁸ against a broad range of most frequently expressed immunogenic antigens in SCC,^34-37 we found an important impairment of antitumor cellular immunity against most SCC-expressed antigens in patients treated with CNI, which was significantly recovered after conversion to mTOR-i, achieving similar levels than transplant patients on mTOR-i than nonimmunosuppressed individuals also developing SCC. Mean antitumor T cell frequencies more accurately illustrated the overall antitumor effector cellular immunity and discriminated those transplant patients with lower incidence of tumor relapses with high accuracy. Importantly, healthy individuals without SCC did not show any detectable antigen-specific T cell frequencies, emphasizing the specificity of the antigens used. Our data are in line with recent relevant studies that have elegantly shown the capacity of mTOR-i to uncover a novel role of mTORC1 in determining functional fates of CD4+ and CD8+ T cells, favoring transition of effector CD8+ T cell responses to memory^33,42,43 and thus, be of great value for the development of vaccines against intracellular infections⁴⁴ as well as against tumors.^45,46 Also, targeting immune-checkpoint cellular signals, such as PD-1 or CTLA-4 receptors to disrupt immune cell-inhibition and restore antitumor immunity, has been shown to be an effective therapeutic strategy for certain cancers.^47,48

In line with our findings, Bottomley and colleagues⁴⁹ recently described that transplant recipients on CNI drugs displaying high proportion of senescent, terminally differentiated CD57^highCD8 + T cells,⁵⁰ were more likely to develop SCC and higher recurrence ratios than those with low proportion of this T cell subset. Altogether, one might speculate that mTOR-i agents might drive the enhancement of tumor-specific effector T cell responses, although we cannot rule out whether this increase might be a consequence of CNI cessation. Nevertheless, although not statistically significant, KT-mTORi-SCC patients showed numerically higher antitumor T cell responses than noKT-SCC, suggesting a direct role of mTORi in this regard.

There are some limitations in our study, which mainly rely in the somehow low number of patients evaluated after conversion to a mTORi-based immunosuppressive regimen. Nonetheless, the consistency of the results obtained within this group of patients, fitting with the same observations found in kidney transplant patients developing SCC under mTOR-i, reinforces our preliminary data and stands the bases for subsequent confirmatory studies. Also, the impossibility to have the complete clinical data of nontransplant patients developing SCC, particularly about the number of SCC recurrences, due to the absence of clinical follow-up in our centers, has impeded to infer the potential value of measuring antitumor effector T cell responses to stratify the risk of tumor recurrence, also in the general population.

In summary, the findings of our study add additional knowledge in the field of nonmelanoma skin cancer in kidney transplant patients. The assessment of tumor-specific T cell responses could therefore help transplant physicians to stratify those patients on CNI drugs that could safely benefit of switching to mTOR-i–based immunosuppression, thus selectively improving their antitumor immunity.

ACKNOWLEDGMENTS

The authors acknowledge the expert assistance of our transplant physicians from each hospital for the careful follow-up and care of all our transplant patients. The authors are grateful to Dr. María J. Fuente and Dr. María T. Fernandez-Figueras from the Dermatology and Pathology departments at Germans Trias i Pujol University Hospital for assisting in the biopsy samples collection as well as to Dr. Octavi Servitge from the Dermatology department at Bellvitge University Hospital. Also, the authors are thankful to our laboratory technicians Nuria Bolaños and Cristian Varela for their efficient work performing all the immune staining experiments, and Gema Cerezo for carefully managing all biological samples.