Several pathogen-induced innate and adaptive immune responses can foster allosensitization and trigger an immune cascade that may lead to allograft rejection (2, 9). Direct graft inflammation, ischemia-reperfusion injury (IRI) following septic shock, priming of the adaptive immune response, T-cell phenotype switching from regulatory to inflammatory T-cells, heterologous immunity/cross-reactivity of virus-specific memory T-cells, and modulation of surface proteins (ICAM, VCAM) to facilitate immune cell infiltration or altering antigen (MHC-Class II) expression are postulated pathogen-induced injurious mechanisms ( Table 1 ) (2, 9, 20). While the relationship between CMV infection and allograft rejection has been well described, chemoprophylaxis against CMV was demonstrated to reduce rejection risk (21–24). Heterologous immunity describes the cross-reactivity of virus-specific memory T-cells to alloantigens and is a potential pathogen-driven mechanism of allosensitization and rejection (9, 25). Recurrent CMV infections and accumulation of cross-reactive T-cells through heterologous immunity may be a potential barrier to achieving graft tolerance (26–28).

Direct graft inflammation and immune cell recruitment following infections with viral (CMV, BK) and bacterial pathogens (pyelonephritis) have also been linked to allograft rejection. However, the mechanisms by which rejection occurs remains poorly understood (2, 3, 9, 29). Lymphopenia is a common complication of CMV disease and/or its treatment. Immunosuppression modulation in response to lymphopenia, particularly mycophenolate dose reduction, may increase the risk of immune activation and rejection (10, 11).

Several studies demonstrate that immunosuppression modulation/interruption in the context of infection, particularly following BK viremia, increases the risk of rejection (4, 14). However, most studies on rejection risk following immunosuppression reduction or interruption are mixed (30–33). There may be an increased risk of denovo DSA formation and/or rejection in transplant recipients who experience more severe infection and require intensive and/or prolonged immunosuppression reduction and in those who are at high immunological risk (31, 34).

Table 1 outlines the various mechanisms by which infection may elicit immunological changes that lead to rejection.

Treatment of allograft rejection with lymphocyte-depleting therapies increases the risk of opportunistic infections that carry significant morbidity and mortality risk (14, 16). The infection risk is often prolonged with therapies such as anti-thymocyte globulin, where T-cell depletion can persist beyond one year. Immunosuppression-induced leukopenia/lymphopenia increases the risk of infections, particularly CMV (57–59). Furthermore, CMV and PJP chemoprophylaxis with agents such as valganciclovir and trimethoprim/sulfamethoxazole can also cause lymphopenia/neutropenia, and interruption of these agents to reverse the leukopenia can increase the susceptibility to these opportunistic infections (10, 11).

Transplant recipients with neutropenia experienced more bacterial infections and the degree of neutropenia correlated with infection risk (60). Several international guidelines recommend initiating CMV prophylaxis for at least three months following anti-rejection therapy with lymphocyte-depleting agents (61, 62). Intentional or accidental omission of chemoprophylaxis following potent anti-rejection immunosuppression may pave the way for viral reactivation and disease.

Chronic kidney disease, particularly late-stage kidney dysfunction, is associated with adaptive and innate immune system dysregulation and accelerated immune aging (63, 64). Shift towards exhausted and immunosenescent lymphocyte phenotype, CD4⁺ T-cell lymphopenia, reduced CD4⁺/CD8⁺ ratio, increased terminally differentiated T-cells, and chronic systemic inflammation are all described to be associated with uremia and progressive CKD (63, 64). Profound allograft dysfunction in the context of rejection, coupled with anti-rejection therapy, can compound the state of immunocompromise and hinder infection and vaccine-induced immune responses. Transplant recipients with lower eGFR (progressive dysfunction beyond <30mL/min/1.73m²) were less likely to achieve positive vaccine sero-response and require extra booster doses to achieve seroprotection (65).

The decision for immunosuppression reduction should consider infection severity, immunological risk of rejection, availability of targeted anti-microbial therapies, and the need for immune reconstitution for infection clearance ( Figure 3 ) (14). Invasive fungal infections (e.g., Cryptococcus spp) and other severe opportunistic infections (e.g., norcadia spp) have considerable 1-year mortality rates and often require immune reconstitution for pathogen control.

While immunosuppression reduction following severe infection was associated with improved patient survival, most infections, particularly mild infections, did not require immunosuppression reduction for infection resolution (14). Additionally, there are no clear guidelines on the immunosuppression agents to reduce. Higher immunosuppression levels are required in the early post-transplantation period. Therefore, immunosuppression reduction in the early phase may be associated with a higher risk of rejection (4).

Most clinicians reduce anti-metabolites first, particularly in the setting of leukopenia, followed by calcineurin inhibitor (CNI) reduction. However, there are no evidence-based protocols on the best immunosuppression reduction strategy (66, 67). Individual immunosuppression agents have different impacts on the innate and adaptive immune systems and, as such, may have different risk profiles for infections with specific microbial organisms (14, 68). Understanding this may assist with clinical decision-making around immunosuppression modulation (14). Adopting a mammalian target of rapamycin (mTOR) inhibitor-based immunosuppression regimen is a common strategy, particularly following viremia. The mTOR inhibitors suppress viral replication and are associated with a reduced risk of viral respiratory infections, increased viral clearance following BK and CMV infections, and milder SARs-CoV-2 infections (69–72).

Several immune scoring systems have been evaluated to improve the infection-risk prediction. Dendle et al. (2018) describe a 4-point composite scoring system predicting severe infection risk in kidney transplant recipients (KTRs). Lower NK cell or CD4⁺ T-cell count, mycophenolate use, and lower eGFR positively correlated with the risk of severe infection (116). Crepin et al. (2016) examined an “immune risk profile (IRP)” defined by positive CMV status, CD4/CD8 ratio <1 and/or CD8 T-cell count >90^th percentile, which was predictive of opportunistic and spontaneous bacterial infections (117). The incidence of acute allograft rejection was lower in the IRP-positive group.

In kidney transplant recipients, the “simplicity score” calculated one month post-transplant was able to predict future risk of infections with good discrimination capacity (118). The score incorporates immune (C3 level, IgG level, CD4+ T-cell count, CD8+ T-cell count), and clinical (recipient age, glomerular filtration rate, recipient age, and infection within the first month) variables (118).

Sarmiento et al. (2014) described a composite “immunological score” using immunoglobulins (IgG, IgM, IgA), complements (C3, C4), and lymphocyte subsets (CD3+, CD4+, CD8+ T cells, NK cells, and B-cells) correlating with severe infections in a cohort of heart transplant recipients (119). IgG <600 mg/dL, C3 <80 mg/dL, C4 <18 mg/dL, NK count <30 cells/µL, and CD4 count <350 cells/µL were all associated with a significantly higher risk of infection, and these five parameters were used to derive the immunological score. Assigned points for each parameter totaled a maximum score of 16. An immune score ≥13 was associated with the highest risk of infection.

Several viruses, including CMV, EBV, and BK, are highly seroprevalent and remain latent until waning immune surveillance and clearance allows for viral replication and reactivation. Higher quantitative viral loads positively correlate with the intensity of immunosuppression (14, 120). Identifying and quantifying viral replication through nucleic acid amplification helps guide prophylactic, pre-emptive, and therapeutic treatment strategies. The emergence of viremia and/or rising viral loads suggests overimmunosuppression and waning viral surveillance (14). Viral infections lacking targeted therapies rely on robust immune responses. T-cell function is particularly important for viral suppression and clearance (14).

BK viral loads over 10,000 copies/mL increase the risk of BK nephropathy. Clinicians commonly use this threshold for immunosuppression reduction (121). Calcineurin level reduction >20%, mycophenolate discontinuation, and high peak BK viral loads correlate with an increased risk of allograft rejection (5). Higher EBV load is linked to PTLD development and the need for immunosuppression reduction (56).

Torque tenovirus (TTV) is a ubiquitous human virus of unclear pathogenic significance. It is unaffected by currently available antiviral therapies and, as such, is a valuable marker of immune competence (14, 122, 123). The kinetics of TTV can provide a measure of the integrated innate and adaptive immune responses. High TTV titres correlate with a reduced risk of allograft rejection and an increased risk of infection (124, 125). A prospective observational study by Doberer et al. (2020) demonstrated a 22% reduced risk of kidney allograft rejection and an 11% increased risk of infection with each log increase in TTV copies/mL (122). TTV counts were significantly higher in KTRs with bacterial, viral, and fungal infections, and elevated counts were detected up to 3 months before the infection (125, 126). A TTV viral load of >3.45 log DNA copies/ml within the first ten days post-transplantation positively predicted the risk of CMV reactivation (127). A TTV threshold level >1x10⁶ copies/ml could exclude rejection with a sensitivity of 94% (124).

KTRs with histopathological lesions of active rejection had lower TTV loads (128). Furthermore, the risk of developing histological features of chronic rejection was associated with the number of days with a TTV viral load <1x 10⁶ copies/ml between 3 to 12 months post-transplant, suggesting suboptimal immunosuppression (128).

Measurement of virus-specific cell-mediated immunity (CMI) can provide insights into cellular immunocompetence and the ability to suppress viral replication. Understanding the strength of virus-specific responses can risk stratify transplant recipients and personalize chemoprophylaxis duration and monitoring (129–132). Immune assays commonly stimulate and measure T-cell functions, including activation, cytokine expression/production (IFN-γ), proliferation, and cytotoxicity (133).

Enzyme-linked immunospot, enzyme-linked immunosorbent assay, flow cytometry and intracellular cytokine staining (ICS), and MHC multimer staining (CMV CD8⁺ Immune Competence) are currently available techniques to assess viral-CMI (133). The flow cytometry ICS also allows for the co-staining of other cytokines and T-cell surface markers to characterize immune cell phenotypes.

CMV-CMI (QuantiFERON-CMV, ELISpot and ICS assays) correlated with functional T-cell responses and viral control (129–132). A positive QuantiFERON-CMV assay at the end of chemoprophylaxis in high-risk recipients (donor +ve/recipient -ve) yielded a positive predictive value of 90% for immune protection (132). Whereas indeterminant or negative results yielded the highest risk of CMV disease, likely correlating with blunted cellular responses (129–132). Quantiferon-CMV status was helpful for risk stratification of individuals with asymptomatic, low-level viremia following prophylaxis cessation (134). Those who were Quantiferon-CMV positive had a higher likelihood of spontaneous clearance, whilst those who were negative had a higher risk of CMV disease (134). Quantiferon-CMV guided pre-emptive therapy was more cost-effective than pre-emptive therapy alone (135). CMV-CMI responses were suppressed for up to 3 months post-ATG therapy. High dose prednisolone and elevated tacrolimus levels particularly impaired CMV-specific functional T-cell responses (129).

T-cell alloreactive CMI assays may be helpful in predicting immune activation and rejection risk, but studies to date are heterogeneous and quote different performance characteristics. Positive pre-transplant donor-reactive IFN-γ responses (ELISpot), particularly at higher levels, were associated with a greater risk of post-transplant rejection (136, 137). The meta-analysis by Montero et al. (2019) found that KTRs with positive pre-transplant donor-specific IFN-γ ELISPOT results had a 3.3-fold higher risk of acute rejection (138). A high number of donor-reactive memory T-cells, as measured by IFN-γ and interleukin-21 ELISPOT assays, was significantly associated with the risk of kidney allograft rejection (139).

CMI is an evolving additive tool for infection risk quantification, particularly for CMV infection. However, further research is needed to establish clear test thresholds relevant to different risk groups (recipient CMV positive vs. negative).

Drug levels are often used to assess the overall immune state. However, the impact of the individual immunosuppressants on the overall immune state is difficult to quantify due to heterogeneity in individual drug pharmacokinetics and pharmacodynamics coupled with concomitant drug dosage adjustments. Therapeutic drug monitoring and target concentration intervention aims to optimize immunosuppression while minimizing toxicity. Calcineurin inhibitor and mTOR inhibitor levels are routinely measured for dose titration.

The proposed optimal target trough tacrolimus level for kidney allografts is 5-8ng/ml beyond the first few months post-transplant[128]. Lower levels are associated with rejection, and higher levels with infection and toxicity (140). CNI level variability relating to non-adherence or underdosing increased the risk of intragraft interstitial fibrosis/tubular atrophy, allograft rejection, and failure (121). Time in therapeutic range (TTR) >78% in the first year post-transplantation was associated with reduced rates of rejection and infection (141). Tacrolimus trough levels >10ng/ml were associated with an increased risk of BKVN, while lower levels were associated with rejection (142, 143). Every 1ng/mL increase in Tacrolimus trough levels beyond 5.35ng/mL at one month post-transplant was associated with an 11% higher rate of infections (144). Genetic polymorphisms related to drug clearance can play a critical role in rejection. Friebus-Kardash et al. (2022) demonstrated that CYP3A5 expressers achieved lower tacrolimus trough levels and were at greater risk of de novo DSA formation (145).

In contrast to CNI and mTOR level-based dose titration, mycophenolate dosing based on regular area under the concentration-time curve (MMF.AUC) measurements or Cmin (minimum concentration) levels have yet to be universally implemented. A fixed dosage mycophenolate regimen is standard practice in many transplant centers worldwide (146). Mycophenolate AUC measurements (targeting 30-60mg/L.h) are often used on a case-by-case basis, such as in the context of infection or malignancy, to guide dose adjustments. Mycophenolate AUC-based dose adjustment reduced infection rates in the 12 months post-kidney transplantation relative to a fixed-dose regimen (147). Individuals with high mycophenolate exposure (AUC 60-100mg/L.h) may safely and cautiously have their mycophenolate dose reduced (e.g. following leukopenia) while maintaining a level required for rejection prophylaxis. A mycophenolate AUC >50mg/L.h at three months post-transplant was associated with sustained BK viremia and BKVN in the subsequent two years (142). Conversely, mycophenolate AUC levels <30mg/L.h have been strongly correlated with rejection, particularly with other risk factors such as high immunological mismatch, delayed graft function, and low levels of concomitant immunosuppressants (148, 149). Despite this, most studies suggest an increased risk of rejection below a threshold of 40mg/L.h, particularly beyond six months post-transplantation (146). Reduction or discontinuation of mycophenolate, irrespective of tacrolimus levels, was associated with adverse graft outcomes, including rejection (146, 150). Hence, the impact of individual drug levels on the overall immune state warrants consideration.

Anti-HLA donor-specific antibodies may be pre-formed or de novo (develop post-transplantation). The detection of donor-specific antibodies (against HLA antigens) has long been the hallmark of the diagnosis of antibody-mediated rejection (AMR) according to the BANFF criteria (151). Recipients with pre-formed antibodies at the time of transplant are at a higher risk of AMR. Within five years post-transplant, 15-25% of transplant recipients develop de novo DSA, with an incidence of 2% per year in immunosuppression adherent transplant recipients (152, 153). The DSA titer, as measured by the mean fluorescence intensity (MFI), correlated strongly with the risk of antibody-mediated rejection and graft failure (154).

This correlation is especially true of complement activating (C1q-positive) and anti-HLA class II DSA (155, 156). There is emerging evidence on the significance of several non-HLA antibodies, including those against major histocompatibility complex class 1-related chain A (MICA), type I angiotensin II receptor, endothelin A receptor, and collagen, in the development of AMR (157).

Under-immunosuppression, owing to medication dosing, genetic variability in drug metabolism, or non-adherence strongly favors DSA formation (152, 158, 159). A high degree of HLA mismatches (especially DQ) and events that cause graft inflammation and increased immunogenicity, including ischemic injury, infections, and cellular rejection, can also trigger de novo DSA formation (154, 160–162).

During the first-year post-transplantation, de novo DSA formation was more common in transplant recipients with subtherapeutic tacrolimus levels (<5ng/ml), particularly in the preceding 6 months (163, 164). Additionally, achieving less than 60% of the time within the therapeutic range (5-10ng/mL) in the first year post-transplant was associated with an increased risk of de novo DSA formation, rejection at 12 months, and graft loss by five years (143).

With the advent of molecular diagnostics, DSA-negative AMR was revealed to be a more common AMR phenotype than previously recognized (165). Several minimally invasive biomarkers may detect graft injury and assist in diagnosing AMR without clear indicators, such as DSA.

Donor-derived cell-free DNA (dd-cfDNA) is one of the most promising and extensively studied biomarkers. Also termed the “liquid biopsy,” dd-cfDNA is a useful screening tool to identify individuals at risk of rejection, who would benefit from a definitive histological diagnosis. A threshold of 1% dd-cfDNA distinguished allograft injury and rejection (166–169). Higher levels of dd-cfDNA correlated with a greater degree of histological injury (170).

Following cell apoptosis and necrosis, a small proportion of DNA, termed cell-free DNA, enters the circulation. Graft tissue necrosis was associated with larger fragment cfDNA (10,000 base pairs), while apoptosis with smaller fragments (60-500bp) (168). In transplant recipients, a small fraction of dd-cfDNA may enter the recipient’s circulation following graft injury (171). Urinary dd-cfDNA can arise from glomerular filtration of circulating dd-cfDNA or donor DNA released from the donor urinary tract (169, 172). Dd-cfDNA can be quantified as relative (proportion (%) of total cf-DNA) or absolute quantitative dd-cfDNA (cp/ml). The diagnostic performance of dd-cfDNA varies depending on the rejection phenotypes, clinical context, immunological risk, and assay type.

Several local and systemic causes affecting graft integrity can elevate circulating dd-cfDNA levels, including infections, calcineurin inhibitor-induced nephropathy, disease recurrence, and ischemia/acute tubular necrosis (167). As such, all potential causes of allograft injury must be considered when assessing dd-cfDNA elevation.

Urinary chemokines are gaining significant ground as a valuable biomarker in assessing kidney allograft health. CXC-motif chemokine ligands 9 and 10 (CXCL9 and CXCL10) are interferon-γ induced chemokines that promote leukocyte migration and infiltration during allograft rejection (195, 196).

Molecular diagnostics is a rapidly evolving field that could change the landscape of precision medicine by providing mechanistic insights into immune cell phenotypes and molecular pathways involved in allograft disease states. Gene expression changes in blood, peripheral blood mononuclear cells (PBMC), urine and tissue have been described in relation to various microbial infections and rejection phenotypes.

Integrating several markers, including molecular, immune, serological, and clinical parameters, may strengthen the overall diagnostic and predictive accuracy for rejection compared to a single method used in isolation. The combination of dd-cfDNA with gene expression profiling increased the predictive power for detecting allograft rejection (262). Additionally, in the study by Park et al. (2021) GEP detected more TCMR whilst dd-cfDNA detected more AMR (262). In a heart transplant cohort, compared to GEP alone for rejection surveillance, GEP with dd-cfDNA produced similar one year rejection-free survival but did so with significantly fewer biopsies (263). Integrating a pre-transplant functional immune assay (donor-alloreactive IFN-γ release assay/ELIspot) with the six-month post-transplant blood gene-expression kSORT assay was shown to predict subclinical rejection (264).

An integrated model combining urinary chemokines, CXCL9, and CXCL10 with six other clinical parameters (recipient age, gender, eGFR, DSA, signs of urinary tract infection, blood BK viral load) showed moderate diagnostic accuracy (AUROC, 0.85) for the detection of kidney allograft rejection (198). Similarly, an integrated model combining CXCL9, CXCL10 with clinical markers (eGFR, DSA, and BK viremia) also showed similar diagnostic performance for detecting acute rejection (AUROC-0.81) (265). Moreover, using the latter composite score, when the predicted rejection risk was <10%, 59 out of 100 protocol biopsies were avoided (265).

The combination of chemokines and miRNA profiling has also been assessed. Millan et al. (2023) described an integrated plasmatic model including three miRNAs (miR-155, miR-181a-5p, miR-122-5p) and CXCL10, which predicted TCMR in liver transplant recipients with good diagnostic performance, AUROC 0.98 and high PPV-97.7% and NPV-97.1% (266). In another study, an integrated model of miR-155-5p, miR-615-3p, and CXCL-9 had strong diagnostic accuracy for predicting kidney transplant allograft rejection (AUROC- 0.92) (267).