The net state of immunosuppression refers to all factors that contribute to a patient's risk of infection and cancer, the most important being the immunosuppressive therapy used.1,2 Ideally, immunosuppression is maintained within a therapeutic window, such that the patient is free of adverse clinical events while maintaining adequate immune depression.3 Overimmunosuppression (OIS) occurs when there is excessive therapy-related adverse event relative to the benefits.4 Although achieving the correct level of immunosuppression is a compromise between the risk of rejection and the risk of adverse effects, there seems to be a shift toward an excess of adverse events in the recent era, with no improvement in the prevention of rejection or long-term graft survival.5,6 Meanwhile, the prevalence of serious opportunistic infections, such as BK viremia (BKV), is approximately 20% during the first year posttransplant, nearly double that of the current rejection rate during the same period.7 Posttransplant infections now exceed acute rejection as the main cause of hospitalization in kidney pediatric recipients and are the main cause of mortality in the first year posttransplant.8,9 Death with a functioning graft remains one of the main causes of graft loss.10-13 The majority of these deaths are caused by infections and cancer.14,15 Thus, achieving a safe balance of immunosuppression is a subtle task.
There is currently no tool to identify patients with levels of immunosuppression that could be suggestive of OIS when balancing the benefits versus the risks of the immunotherapy. Thus far, the only guides to adjust immunosuppressive drugs have been the patient's anthropomorphic data, medical history, blood levels, and genetic testing to a very limited extent.16 In current clinical practice, it is only after the occurrence of clinical episodes such as opportunistic infection or the recurrence of infections without a predisposing factor that OIS is suspected. As stated by Thaunat,4 complications are indeed the current marker of OIS.
So far, available biomarkers only provide information on the pharmacokinetics of immunosuppressive drugs.3 Variability in interindividual and intraindividual responses to immunosuppressive treatment makes it impossible to translate these pharmacokinetic data into immune potency in vivo.17 Although some pharmacodynamic biomarkers of calcineurin inhibitors or mycophenolate mofetil go further in analyzing immunosuppressive effects on the expression of genes that are regulated by the nuclear factor of activated T cells and the activity of inosine-5′-monophosphate dehydrogenase, respectively,17,18 these biomarkers do not determine the combined effect of these drugs on the immune response. This is a critical limitation, given that transplant recipients most often receive a combination of several drugs as part of their maintenance regimen.
The ex vivo examination of cellular activation is a promising avenue to detect OIS. From a biological perspective, the cells collected from a patient's venous puncture have been subjected in vivo to the full immunosuppressive regimen received in real time by the patient. Thus, their activation capacity reflects the net state of the immunosuppression of a patient at a given time. From a research perspective, the main challenge is to derive a single or a simple set of in vivo assays that quantifies this net state of immunosuppression ex vivo. Recent data suggest that Epstein-Barr virus (EBV) viral load posttransplant is positively associated with the occurrence of cancer and opportunistic infections, and the cumulative incidence of cancer is more than doubled in patients with persistent high EBV replication.19,20
We hypothesized that incubating whole peripheral blood mononuclear cells (PBMCs) with EBV-derived peptides and measuring the immune response might provide useful information. Circulating innate immune cells are of particular interest because they act as both antigen-presenting cells in the early phase of the immune response and as effector cells after T-cell activation.21 Here, we derive and validate the notion that a low TNF-α response by pro-inflammatory CD14+CD16+ monocytes is predictive of OIS events in a manner that is independent of renal function and immunosuppressive therapy.
MATERIALS AND METHODS
Study Design and Population
This single-center prospective, observational study involved the longitudinal collection of biological samples and clinical data. The cohort comprised 45 adult kidney graft recipients who were transplanted between May 2012 and February 2014. During this period, any de novo kidney transplant recipient aged at least 18 years was eligible for enrolment in the study. This study is in compliance with the principles of the Istanbul Declaration and received approval from the local ethics committee (protocol H11-05-027). Written informed consent was obtained from all participants at the time of enrolment. Additional details are available in the SDC, Materials and Methods (http://links.lww.com/TP/B536).
Prospective Monitoring and Sample Collection
Each patient was followed over 2 years posttransplant, during which time, 9 protocol visits were scheduled in addition to visits triggered by clinical events. On each visit, patients were asked to undergo a blood test in respect of their medical condition. A total of 319 PBMC samples were analyzed over the study period. Additional details are available in the SDC, Materials and Methods (http://links.lww.com/TP/B536).
In accordance with previous reports,3 we established operational criteria for OIS according to the occurrence of the following events: an opportunistic infection, recurring bacterial infections (≥3 episodes in 12 months), or de novo neoplasia. Primary infection to cytomegalovirus and EBV were not included in the definition. Patients with no event assigned to OIS were assigned to the control group. Monitoring for poliomavirus was conducted every month during the first year and every 2 months in the second year. None of the patients was lost to follow-up for viral monitoring.
Immediately after venous puncture, PBMCs were isolated from heparinized blood by density gradient centrifugation (Ficoll-Paque PLUS; Thermo Fisher Scientific, Mississauga, ON). Peripheral blood mononuclear cells were aliquoted and frozen at −80°C overnight in serum-free medium (Cellular Technology Limited CTL, Cleveland, OH) and then stored in a liquid nitrogen tank. Flow cytometry data analyses were performed with the FlowJo vX software (FlowJo LLC). Additional details are available in the SDC, Materials and Methods. The gating strategies used are shown in Figure S1, SDC (http://links.lww.com/TP/B536).
The training and validation sets were constructed randomly from the complete cohort. The training cohort was designed with an equal number of OIS patients and controls. Clinical variables were examined using Mann-Whitney test or Fisher exact test. Longitudinal data were analyzed using linear mixed models. We used the coordinates of the receiver operating curve to determine the threshold of TNF-α–positive cells within the CD14+CD16+ monocytes. Additional details are available in the SDC, Materials and Methods (http://links.lww.com/TP/B536).
Study Population, Baseline Characteristics and Clinical Events
Between May 2012 and February 2014, 45 de novo kidney transplant recipients were enrolled at the time of transplant and participated in prospective PBMC collection at 9 time points during the first 24 months posttransplant: 0, 1, 3, 6, 9, 12, 16, 20, and 24 months. The flow chart of the study population is displayed in Figure 1. One patient lost the allograft 2 weeks after transplant and was eliminated from the analysis. An additional patient was discarded for reasons of biohazard safety because he was positive for hepatitis C virus. Cells from 2 patients were discarded because of a technical issue with a biofreezer.
The patients were classified according to their clinical phenotype. To study the OIS as the primary outcome, we established an operational definition of OIS as the occurrence of at least one of the following: opportunistic infection, recurrent infections (≥3 episodes in 12 months) in the absence of a predisposing factor or de novo cancer.22 Overall, there were 16 OIS patients and 25 controls. Patients were randomly distributed into a training cohort (n = 12) and a validation cohort (n = 29). The sole distribution rule was the need for the training cohort to be balanced between OIS and controls (n = 6 for each group). A summary of the OIS (including BK viral loads) and rejection events is presented in Table S1 SDC (http://links.lww.com/TP/B536). The OIS events that occurred were mostly BKV, BK viral nephropathy, secondary cytomegalovirus viremia, and secondary EBV viremia. Baseline characteristics and the early immunosuppression regimen are presented in Table 1. The groups were similar, with the exception that OIS patients were older (55 ± 7 versus 46 ± 12 years, P = 0.01).
Monocyte Activation After Stimulation With Synthetic EBV Peptides
Exploratory analyses were conducted on 86 PBMC samples with viable cells collected from the training cohort. For technical purposes, we first studied T cell and monocyte subsets. Monocytes are distinguished by their surface phenotype as the CD14+CD16− classical subset or as the CD14+CD16+ intermediate subset that secrete proinflammatory cytokines, such as TNF-α.23,24 We initially examined classical monocytes, which represent the main subset (90% of the cells). We found that the percentage of TNF-α positive cells was very low under resting conditions and after incubation with EBV peptides (0.1 ± 0.1% and 1.0 ± 0.5%, respectively, Figure 2A left and middle panels). Cell responsiveness was confirmed upon stimulation with lipopolysaccharide (Figure 2A). Overall, no difference between OIS patients and controls was observed in this subset.
We next examined the CD14+CD16+ monocytes. Unstimulated, nearly 15% of the cells in this subset secreted TNF-α, without a notable difference between the groups (Figure 2B-C). In contrast to classical monocytes, CD14+CD16+ monocytes responded strongly to EBV peptides (Figure 2B). Analysis of the individual patient data showed that the variability of response was greater in the OIS patients than in the controls. The within-patient variability (individual SD) of TNF-α–positive cells was 10% for the control group and 17% for the OIS patients (P = 0.013 according to a t test comparing the mean of individual SDs).
Overall, the mean percentage of TNF-α–positive cells was lower in the OIS patients than in the controls (mean of all time points, 65 ± 17 vs 79 ± 10% respectively, P = 0.023 based on the linear mixed model, Figure 2C). Visual analysis over time shows that although both groups had a similar response at the time of transplant, there was a decrease in monocyte activation in the OIS patients after transplant (Figure 2C). This attenuation in the response progressed until month 9 and remained relatively stable afterward. This suggests that upon stimulation with EBV peptides, the controls were able to generate a proinflammatory response at a constant level during follow-up. In contrast, the response of OIS patients posttransplant was lower and more variable, not only as a group but also at the individual patient level.
T-Cell Activation After Stimulation With Synthetic EBV Peptides
One can expect that a difference in the effector response of peripheral monocytes between these 2 groups might merely reflect a difference in upstream T cell activation. To examine this possibility, we carried out the same cell incubation protocol as described above and studied the secretion of IFN-γ for CD4+ and CD8+ T cells subsets. Anti-CD3/CD28 beads were used as a positive control.
Figure S2, SDC (http://links.lww.com/TP/B536) illustrates the secretion levels of IFN-γ by memory CD4+CD45RO+ T cells. Upon stimulation with EBV peptides, a similar profile was observed between OIS patients and the controls (Figure S2A and B, SDC for the individual patient and group analysis, respectively, http://links.lww.com/TP/B536). The activation capacity of the recovered cells was confirmed in the wells that had been stimulated with CD3/CD28 beads. Overall, the percentage of positive cells under the unstimulated condition was low in both groups of patients (Figure S2B, SDC, http://links.lww.com/TP/B536). The activation of memory CD8+ T cells revealed a similar profile to that observed with memory CD4+ T cells (Figure S3, SDC, http://links.lww.com/TP/B536). These data show that there was no difference in memory T-cell activation profiles between OIS patients and controls under any of the stimulations. Similar responses were observed in the naive (CD45RO−CD62L+), effector memory (CD45RO+CD62L−), and central memory (CD45RO+CD62L+) subsets (not shown).
To further investigate whether the monocyte response observed above was dependent on T-cell recognition of EBV peptides, and whether the antigen-presenting cells, such as DCs and B cells, were essential to this response ex vivo, we conducted a series of experiments in which we depleted different leucocyte subsets in healthy volunteers (Figure S4, SDC, http://links.lww.com/TP/B536). We found that incubation of monocytes alone resulted in a significant reduction, although not an abrogation, in the proportion of TNF-α-positive monocytes (80.2 ± 1.7% vs 52.5 ± 2.4%, P < 0.01). CD4+-depleted PBMCs led to a substantial reduction in the proportion of CD14+CD16+TNF-α+ monocytes (80.2 ± 1.7% vs 67.3 ± 6.4%, P = 0.06). Depletion of CD8+ T cells led to a lower reduction, whereas depleting both CD4+ and CD8+ simultaneously led to a significant decrease (80.2 ± 1.7% vs 67.9 ± 3.0%, P = 0.03). We observed little effect of depleting DCs or B cells. In all, these mechanistic data suggest that T-cell depletion did not abrogate the monocyte response but that, along with the other cell population, they are necessary to trigger the full response noted with PBMCs. However, monocytes can respond to these peptides, as has been shown in the past for antigliadin peptides in patients with celiac disease.25
Development of a Prediction Test for OIS
The exploratory analyses above led us to build a prediction tool for immunosuppression status according to the activation profile of the CD14+CD16+ monocytes after stimulation with EBV peptides. From a clinical standpoint, we aimed to develop a rule based on the serial measurement of monocyte response rather than on a single measurement. We empirically tested the diagnostic accuracy of the following prediction rule: at least 2 consecutive percentages of TNF-α–positive cells fall below a discrimination threshold. To do so, we conducted a receiver-operating curve analysis using this criterion and analyzed the coordinates of the curve in the training set (Figure 3). The data indicated that an optimal value of 73% of TNF-α–positive cells resulted in the best threshold, providing a balanced sensitivity and specificity of 83% each (Figure 3A-B). Considering that the prevalence of OIS was set by design at 50% in this training cohort, the positive predictive value (PPV) and the negative predictive value (NPV) were also both 83%.
Validation of the Assay
External validation was carried out in an independent set of 233 PBMC samples from 29 patients, including 10 OIS and 19 control patients. Patients were classified based on the rule established above. The OIS patients tested positive in all cases but one, whereas the controls tested negative in 12 of 19 cases (Figure 4A; for each group, patients with positive tests were split into 2 subfigures, middle and right, for clarity purposes). Overall, this validation cohort reported the following diagnostic accuracy: sensitivity 90% (9/10), specificity 63% (12/19), PPV 56% (9/16), and NPV 92% (12/13) (Figure 4B). The data from this validation cohort demonstrate a better sensitivity but a lower specificity than those found in the training cohort as well as a better NPV but a lower PPV; these latter findings are expected because of the lower prevalence of OIS patients, which reflects the real-life setting.
In addition to the validation above, we assessed the reproducibility of the assay across experiments, wells, operators, and cell handling (frozen vs fresh cells). Overall, the coefficient of variability (CV) obtained were all below 5% (intra-assay CV, 3.9% between duplicates; inter-assay CV, 2.6% between experiments; interoperator CV, 4.3%; frozen vs fresh cells CV, 4.0% (Table S2A-D, SDC, http://links.lww.com/TP/B536).
Impact of Renal Function, Age and Immunosuppressive Regimen on the Assay
In the clinical setting, the selection of immunosuppressive agents and dose adjustments is an iterative process and is guided in part by the occurrence of OIS events. Therefore, we sought to determine whether the association between these OIS events and the CD14+CD16+ monocyte response to EBV peptides was confounded by the immunosuppressive regimen received by the patients. This analysis was performed on the complete cohort of patients using clinical data that were prospectively collected during the study follow-up. The routine maintenance regimen comprised a combination of prednisone, tacrolimus (TAC), and mycophenolate.26 Tacrolimus exposure was measured based on trough (T0) levels. Because there is no reliable blood level monitoring method for prednisone and mycophenolate, we used the prescribed dose to assess exposure to these drugs.
We first conducted a longitudinal descriptive analysis of drug exposure over time (Figure 5A-C). Prednisone doses were very similar between OIS and controls for the entire study duration (Figure 5A), whereas exposure to TAC was nonsignificantly lower (Figure 5B, P = 0.167 by linear mixed model). The doses of mycophenolate were lower in OIS patients (Figure 5C, P = 0.002 by linear mixed model). This difference was expected because mycophenolate dose is routinely reduced in our center after the occurrence of a viral infection or neoplasia. Overall, these data suggest that OIS received less maintenance immunosuppression than controls.
We used data on induction regimen and maintenance immunosuppression to investigate whether the association between OIS and the TNF-α secretion by monocytes was confounded by the immunosuppressive regimen received by the patients. In the unadjusted model, there was a significant association between the OIS phenotype and the percentage of TNF-α-positive cells (−8.2 ± 4.1% vs controls, P = 0.033 according to the mixed model, Figure 5D). This effect remained similar following adjustment (−9.2 ± 4.1% vs controls, P = 0.027, Figure 5E). Illustrative examples taken from the analysis of 3 patients show the lack of correlation between TAC trough levels and the monocyte response over time (Figure S5, SDC, http://links.lww.com/TP/B536). Next, we tested the robustness of the association after additional adjustment for recipient age and renal function, which are known to dampen the immune response.27,28 The coefficient and P value were similar to those reported above, showing no signal of potential residual confounding (−9.0 ± 4.3% vs controls, P = 0.040, Figure 5F).
The modulation of immunotherapy has been suggested as a key element of precision medicine for the 21st century.29,30 In this report, we investigated whether measuring the response of transplant recipients' PBMCs to EBV peptides in vitro could be informative regarding the clinical status of OIS. We used a training cohort to explore several T cell and monocyte subsets. From this exploratory analysis, we developed a test that identified OIS based on the CD14+CD16+ monocyte response to EBV peptides as measured by TNF-α secretion. We then validated this assay in an independent, external validation cohort. The association between OIS status and the monocyte response was independent of the immunosuppressants received. In all, the current data indicate that impaired TNF-α secretion by monocytes retrospectively identify patients at increased risk of infectious complications.
These results are clinically relevant for many reasons. First, there is an unmet need for an “OIS barometer,” not only in transplantation but also in all fields of medicine using immunosuppressive drugs. OIS is a serious problem that currently leads to morbidity and mortality at levels that now surpass those due to rejection in solid organ transplantation.1,5,31 Second, our results show that using this assay, patients who are not OIS were reliably identified with a high validated sensitivity (90%) and NPV (92%). Of note, these findings were obtained in a cohort of patients with a substantial prevalence of OIS. This test could be helpful in adapting the immunosuppression for patients at high immunological risk. The high NPV could guide a safe increase in the immunosuppressive load or aid in safely maintaining high doses of immunosuppressants.
The ex vivo approach used here is among the first attempts at using the innate immune response to conduct precise characterization of the in vivo immune response in immunosuppressed patients. In recent years, clinical models have been developed to stratify the risk of rejection and infection.13 Despite the commercialization of assays such as the ImmuKnow from Cylex, as yet, no single test has been widely translated to the clinic.32 Meta-analyses concluded that the predictive power of these tests remained too limited for clinical translation.33,34 Recent reports on the profile of circulating immune cells have shown an association between the number of CD4+ T cells and cancer incidence and mortality in renal transplant recipients, but it remains to be validated.35,36
Importantly, the focus on CD14+CD16+ cells is the result of data obtained in the discovery phase, during which we explored the responses of many PBMC subsets. These cells are found in large numbers in the peripheral blood during infection and inflammation.23 Therefore, it is biologically plausible that they play an essential role in patients receiving the current immunosuppression regimen, which predominantly targets T cells.
The main reported disadvantages of cell-based assays are their coefficient of variation and the fact that they are laborious and time-consuming.4 However, such assays are now used in clinical practice for the diagnosis of specific infections.37,38 One main advantage of the test developed here is that it is not prone to variations depending on cell concentrations, a caveat of enzyme-linked immunosorbent assay-based or enzyme-linked immunospot-based assays. Another advantage is that it does not require the knowledge of an individual HLA-type, as the tetramer-based assay does.9,32 The equipment required for the readout of the assay, a flow cytometer, is currently available in any transplant center and the procedure described here could be standardized by an organization such as the American Society for Histocompatibility and Immunogenetics. Recently, a combination of cellular assays measuring natural killer cell function and T-cell alloresponse produced promising results to identify cancer and its prognosis in 56 kidney recipients.39 Therefore, cellular assays are amenable to routine clinical testing, and their use is likely to expand in the future.
There are some limitations to this study. First, the assay focuses of OIS. Given the promising new urinary assays to identify rejection, for instance, those based on urinary chemokine measurements for which interventional trials are ongoing,40,41 we believe that the TNF-α assay could be validated and used in combination with such biomarkers of rejection.
One can argue that the specificity and PPV in the validation cohort were rather low. Indeed, several patients who falsely tested positive did so in the early period (the first 3-6 months posttransplant, Figures 3C and 4A), the time window during which the immunosuppression is maximal. Although we have no data available beyond 24 months posttransplant, one can speculate that the assay could be most useful after this early posttransplant period. Second, OIS is not a definite medical diagnosis, which makes its definition arbitrary. We feel that, although the definition per se is debatable, there is a consensus that OIS is a medical reality with serious consequences from both clinical and economical perspectives.3,42,43 Here, we developed an operational definition encompassing relevant clinical endpoints that are routinely attributed to the use of immunosuppressants. Most of the OIS events were related to specific viremias that are recognized as signs of OIS.42 Third, at this stage of the assay development, the study design did not allow to identify a timely correlation between TNF-α responses and the occurrence of clinical events. A prospective study, in which the results of the assay will be confirmed in a second sample within a short timeframe, will be needed to rigorously test the predictive value of the assay. Finally, some cell subsets, in particular B and NK cells, were not studied here. Measuring their response could both improve the accuracy of this cellular test and potentially help determine more precisely the type of OIS event (eg, viremia vs recurring bacterial infection) that the patient is at risk of developing. Additional experiments are underway to achieve this goal.
In summary, the data presented here support the idea that the serial examination of the CD14+CD16+ monocyte response to EBV peptides is informative about the occurrence of OIS events in de novo renal transplant recipients. These results provide the impetus to expand this approach to different populations of immunosuppressed patients and to investigate additional cell subsets to refine ex vivo cellular assays.
The authors thank all the patients who participated in the study. We thank Dr. Eric Wagner for his technical assistance. The authors also thank Mrs. France Samson and Mrs. Danielle Villeneuve for patient follow-up and data collection.
1. Fishman JA. Infection in solid-organ transplant recipients. N Engl J Med
2. Fishman JA. Infection in Organ Transplantation. Am J Transplant
3. Budde K, Matz M, Durr M, et al. Biomarkers of over-immunosuppression. Clin Pharmacol Ther
4. Thaunat O. Finding the safe place between the hammer and the anvil: sounding the depth of therapeutic immunosuppression. Kidney Int
5. Halloran PF. Immunosuppressive drugs for kidney transplantation. N Engl J Med
6. Meier-Kriesche HU, Schold JD, Srinivas TR, et al. Lack of improvement in renal allograft survival despite a marked decrease in acute rejection rates over the most recent era. Am J Transplant
7. Dall A, Hariharan S. BK virus nephritis after renal transplantation. Clin J Am Soc Nephrol
. 2008;3(Suppl 2):S68–S75.
8. Dharnidharka VR, Stablein DM, Harmon WE. Post-transplant infections now exceed acute rejection as cause for hospitalization: a report of the NAPRTCS. Am J Transplant
9. Fernandez-Ruiz M, Kumar D, Humar A. Clinical immune-monitoring strategies for predicting infection risk in solid organ transplantation. Clin Transl Immunology
10. Ojo AO, Hanson JA, Wolfe RA, et al. Long-term survival in renal transplant recipients with graft function. Kidney Int
11. Matas AJ, Gillingham KJ, Sutherland DE. Half-life and risk factors for kidney transplant outcome—importance of death with function. Transplantation
12. West M, Sutherland DE, Matas AJ. Kidney transplant recipients who die with functioning grafts: serum creatinine level and cause of death. Transplantation
13. Cippa PE, Schiesser M, Ekberg H, et al. Risk stratification for rejection and infection after kidney transplantation. Clin J Am Soc Nephrol
14. Pilmore H, Dent H, Chang S, et al. Reduction in cardiovascular death after kidney transplantation. Transplantation
15. Sanders-Pinheiro H, da Silveira ST, Carminatti M, et al. Excessive immunosuppression in kidney transplant patients: prevalence and outcomes. Transplant Proc
16. Kidney Disease: Improving Global Outcomes (KDIGO) Transplant Work Group. KDIGO clinical practice guideline for the care of kidney transplant recipients. Am J Transplant
. 2009;9(Suppl 3):S1–S155.
17. Babel N, Reinke P, Volk HD. Lymphocyte markers and prediction of long-term renal allograft acceptance. Curr Opin Nephrol Hypertens
18. van Rossum HH, de Fijter JW, van Pelt J. Pharmacodynamic monitoring of calcineurin inhibition therapy: principles, performance, and perspectives. Ther Drug Monit
19. Bamoulid J, Courivaud C, Coaquette A, et al. Late healthistent positive EBV viral load and risk of solid cancer in kidney transplant patients. Transplantation
20. San-Juan R, De Dios B, Navarro D, et al. Epstein-Barr virus DNAemia is an early surrogate marker of the net state of immunosuppression in solid organ transplant recipients. Transplantation
21. Serbina NV, Jia T, Hohl TM, et al. Monocyte-mediated defense against microbial pathogens. Annu Rev Immunol
22. Vajdic CM, van Leeuwen MT. Cancer incidence and risk factors after solid organ transplantation. Int J Cancer
23. Auffray C, Sieweke MH, Geissmann F. Blood monocytes: development, heterogeneity, and relationship with dendritic cells. Annu Rev Immunol
24. Guilliams M, Ginhoux F, Jakubzick C, et al. Dendritic cells, monocytes and macrophages: a unified nomenclature based on ontogeny. Nat Rev Immunol
25. Cinova J, Palova-Jelinkova L, Smythies LE, et al. Gliadin peptides activate blood monocytes from patients with celiac disease. J Clin Immunol
26. Ekberg H, Tedesco-Silva H, Demirbas A, et al. Reduced exposure to calcineurin inhibitors in renal transplantation. N Engl J Med
27. Heinbokel T, Elkhal A, Liu G, et al. Immunosenescence and organ transplantation. Transplant Rev (Orlando)
28. Kato S, Chmielewski M, Honda H, et al. Aspects of immune dysfunction in end-stage renal disease. Clin J Am Soc Nephrol
29. Bluestone JA, Tang Q. Immunotherapy: making the case for precision medicine. Sci Transl Med
30. McDonald-Hyman C, Turka LA, Blazar BR. Advances and challenges in immunotherapy for solid organ and hematopoietic stem cell transplantation. Sci Transl Med
31. Dharnidharka VR, Storch GA, Brennan DC. Urinary-cell mRNA and acute kidney-transplant rejection. N Engl J Med
32. Brunet M, Shipkova M, van Gelder T, et al. Barcelona consensus on biomarker-based immunosuppressive drugs management in solid organ transplantation. Ther Drug Monit
. 2016;38(Suppl 1):S1.
33. Rodrigo E, Lopez-Hoyos M, Corral M, et al. ImmuKnow as a diagnostic tool for predicting infection and acute rejection in adult liver transplant recipients: a systematic review and meta-analysis. Liver Transpl
34. Ling X, Xiong J, Liang W, et al. Can immune cell function assay identify patients at risk of infection or rejection? A meta-analysis. Transplantation
35. Carroll RP, Segundo DS, Hollowood K, et al. Immune phenotype predicts risk for posttransplantation squamous cell carcinoma. J Am Soc Nephrol
36. Sommerer C, Giese T, Meuer S, et al. Pharmacodynamic monitoring of calcineurin inhibitor therapy: is there a clinical benefit? Nephrol Dial Transplant
37. Prevention CfDCa. Updated guidelines for using interferon gamma release assays to detect Mycobacterium tuberculosis
38. Ferrara G, Losi M, D'Amico R, et al. Use in routine clinical practice of two commercial blood tests for diagnosis of infection with Mycobacterium tuberculosis
: a prospective study. Lancet
39. Hope CM, Troelnikov A, Hanf W, et al. Peripheral natural killer cell and allo-stimulated T-cell function in kidney transplant recipients associate with cancer risk and immunosuppression-related complications. Kidney Int
40. Hricik DE, Nickerson P, Formica RN, et al. Multicenter validation of urinary CXCL9 as a risk-stratifying biomarker for kidney transplant injury. Am J Transplant
41. Hirt-Minkowski P, Amico P, Ho J, et al. Detection of clinical and subclinical tubulo-interstitial inflammation by the urinary CXCL10 chemokine in a real-life setting. Am J Transplant
42. Fishman JA. Opportunistic infections—coming to the limits of immunosuppression? Cold Spring Harb Perspect Med
43. Borni-Duval C, Caillard S, Olagne J, et al. Risk factors for BK virus infection in the era of therapeutic drug monitoring. Transplantation