Immune checkpoint inhibitors (ICIs) have ushered in a paradigm shift in oncology, delivering durable remission for patients with advanced malignancies by antagonizing the CTLA-4, PD-1, or PD-L1 pathways (1, 2). By reinvigorating T-cell surveillance against tumor neoantigens, these agents have fundamentally altered the prognostic landscape for melanoma, non-small cell lung cancer, and renal cell carcinoma (3). However, this systemic immune activation inherently disrupts self-tolerance, precipitating a spectrum of immune-related adverse events (irAEs) that can affect organ system (4). Among these, nephrotoxicity—predominantly manifesting as acute tubulointerstitial nephritis (ATIN)—has emerged as a critical complication that threatens both renal survival and oncologic outcomes (5). While early clinical trials reported a relatively low incidence of ICI-associated acute kidney injury (ICI-AKI) ranging from 2% to 3%, subsequent real-world cohorts and biopsy series have unmasked a more pervasive burden, with incidence rates climbing significantly higher in unselected populations, particularly those receiving combination immunotherapy (6–8). The clinical conundrum of ICI-AKI is profound: it necessitates the interruption of potentially life-saving cancer therapy and the initiation of immunosuppression, typically corticosteroids, thereby placing the patient at a precarious intersection of tumor progression risk and renal failure (9). Furthermore, sustained AKI has been independently associated with increased mortality and major adverse cardiovascular events in this fragile population, underscoring that kidney injury is not merely a transient bystander but a determinant of overall survival (10).

Despite the accumulating decade of clinical experience, the management of ICI-AKI remains fraught with uncertainty and relies heavily on expert consensus rather than high-level evidence. Current guidelines universally recommend withholding ICIs and administering corticosteroids for moderate-to-severe AKI; however, the precise efficacy of steroids remains debated, with recent data suggesting heterogeneous recovery rates and questioning the necessity of aggressive immunosuppression for all grades of injury (11, 12). Perhaps the most agonizing dilemma facing the onco-nephrologist is the decision to rechallenge with ICIs after renal recovery. Resuming therapy offers the best chance for tumor control but carries the theoretically elevated risk of recurrent AKI, which could lead to permanent dialysis dependence (13). Historically, smaller cohorts have painted a discouraging picture, reporting recurrence rates as high as 20–30%, leading to a pervasive reluctance to rechallenge (14). However, the landscape of evidence is shifting with unprecedented velocity. A 2025 meta-analysis by Ho et al. provided a foundational assessment of these risks up to mid-2024 (15). Yet, since that cutoff, a wave of large-scale, real-world data—including massive registry studies published in late 2024 and 2025 involving tens of thousands of patients—has fundamentally challenged our prior assumptions (16). These emerging datasets suggest that with improved recognition and management, the “real-world” recurrence risk may be substantially lower than previously feared. Consequently, existing syntheses may already be outdated, potentially leading to overly conservative practices that deny patients vital cancer treatments based on obsolete risk estimates.

Building upon our group’s long-standing dedication to characterizing the pathophysiology and epidemiology of immune-mediated organ damage, this study represents both a logical extension and a critical update to the existing canon of knowledge. We hypothesize that the integration of recent, high-volume real-world evidence will reveal a “safety signal evolution,” characterized by stabilizing incidence rates and, crucially, a more favorable safety profile for ICI rechallenge than historical data suggested. Furthermore, we posit that specific host factors, particularly age, may differentially modulate the risk of initial injury versus recurrence, a nuance previously obscured in smaller aggregate analyses. To test these hypotheses, we conducted this updated systematic review and meta-analysis. Our objective is to provide the compelling, risk-stratified evidence base required to transition the field from standardized caution to precision onco-nephrology, ultimately empowering clinicians to maximize the therapeutic window of immunotherapy while safeguarding renal health.

Our cumulative meta-analysis, synthesizing data from over 60,000 patients through late 2025, represents the most comprehensive evaluation of ICI-AKI to date. The findings challenge the prevailing therapeutic conservatism that has historically governed the management of nephrotoxicity in immuno-oncology. We demonstrate that while the incidence of ICI-AKI has stabilized at approximately 2.6% in the modern era, the risk of recurrence upon rechallenge—previously the most feared barrier to resuming life-prolonging therapy—has declined significantly to 14.1%. Furthermore, our identification of an age-dependent dissociation between initial risk (higher in younger patients) and recurrence risk (higher in the elderly) introduces a novel framework for personalized risk stratification. These data collectively advocate for a paradigm shift from a “one-size-fits-all” discontinuation policy to a precision onco-nephrology approach (Figure 6), where rechallenge is considered a viable standard of care for the majority of patients, particularly the non-elderly.

The stabilization of the ICI-AKI incidence rate at 2.6% in our cumulative cohort reflects the maturation of clinical practice. Early registration trials and initial case series often reported widely divergent rates, ranging from <1% to >5%, largely driven by inconsistent definitions of AKI and variable surveillance intensity (13). Our inclusion of large-scale, real-world registry data (e.g., Chuang et al.) likely captures the “true” burden of disease more accurately than highly selected clinical trial populations. Notably, this 2% figure is not trivial; given the exponential growth in ICI prescriptions worldwide, it translates to thousands of patients annually facing the dual threat of renal failure and cancer progression.

A critical finding in our updated analysis is the attenuation of the steroid efficacy signal (OR 0.55, 95% CI 0.06–1.04; P = 0.03). The wide confidence interval (95% CI 0.06–1.04) suggests substantial heterogeneity. This supports the emerging view that not all ‘clinical ICI-AKI’ is immune-mediated (some may be ATN), and thus, not all cases will respond to corticosteroids. Clinicians should weigh this uncertainty against the risks of high-dose immunosuppression. While corticosteroids remain the cornerstone of management for biopsy-proven acute tubulointerstitial nephritis (ATIN), the broadening of the confidence interval to cross unity suggests that not all ICI-AKI events are steroid-dependent. This aligns with the emerging concept of “clinically diagnosed ICI-AKI” being a heterogeneous entity, comprising not only ATIN but also hemodynamic injury and acute tubular necrosis (ATN), which do not respond to immunosuppression (30). The indiscriminate use of high-dose steroids in such cases may needlessly expose patients to infectious risks and metabolic derangements without conferring renal benefit (31).

The pooled incidence of 2.6% reported here sits at a crucial intersection between early retrospective cohorts and recent prospective vigilance. In the seminal multicenter study by Cortazar et al. (2016), the incidence was estimated at 2-3%, but this was derived from academic centers with high indices of suspicion (5). Conversely, a large pharmacovigilance analysis of the FDA FAERS database by Magee et al. suggested a reporting rate of less than 1%, a figure heavily criticized for under-reporting bias inherent to spontaneous reporting systems (32). Our analysis, by integrating comprehensive electronic health record (EHR) data from TriNetX and other registries, bridges this gap. It suggests that while severe, Stage 3 AKI remains rare (<0.5%), milder forms of immune-mediated kidney injury are a pervasive reality of modern oncology practice, occurring at rates comparable to other severe irAEs like pneumonitis. Furthermore, our finding that combination therapy (CTLA-4 + PD-1 blockade) more than doubles the risk (4.5% vs. 1.8%) reinforces and quantifies observations from smaller cohorts (33). This magnitude of risk elevation is consistent with the “dosage” hypothesis of autoimmunity, where broader uncoupling of T-cell tolerance leads to indiscriminate tissue attack. Comparing our data with the meta-analysis by Gupta et al. (2021), which reported a similar trend, our updated estimates provide narrower confidence intervals, offering clinicians a precise risk metric to quote when consenting patients for dual immunotherapy (34).

The most striking divergence between our results and historical data lies in the safety of ICI rechallenge. Previous influential studies, such as the multicenter series by Cortazar et al. (2020) and the systematic review by Oleas et al., reported recurrence rates ranging from 23% to as high as 40% (35, 36). These figures fostered a widespread “therapeutic nihilism,” where any grade of ICI-AKI was viewed as a permanent contraindication to future immunotherapy. In sharp contrast, our cumulative analysis, heavily weighted by 2024–2025 data, indicates a recurrence rate of only 14.8%, with some individual high-quality studies reporting rates as low as 7% (17). This precipitous drop is unlikely to be biological; rather, it reflects a profound evolution in patient selection and management. In the “early era” (2015-2020), clinicians may have rechallenged patients prematurely, before full renal recovery, or without adequate secondary prophylaxis. In the “modern era” (2021-2025), rechallenge is likely restricted to patients who have achieved complete renal recovery (Grade 0-1) and are monitored with heightened vigilance (e.g., weekly creatinine/biomarker surveillance) (37). Furthermore, the increasing use of concomitant low-dose maintenance steroids or alternative immunosuppressants (e.g., mycophenolate, infliximab) during rechallenge in refractory cases may be dampening recurrence risks, a practice pattern that was virtually non-existent in earlier cohorts (38). This discrepancy highlights the limitations of comparing “raw” rates across decades. The “Cortazar cohort” represented the pioneers navigating uncharted territory; the “Chuang and Zhou cohorts” represent a mature field applying learned lessons. Our meta-analysis confirms that the modern risk of recurrence is manageable and should not preclude the resumption of potentially curative cancer therapy.

Our finding that the odds ratio for steroid efficacy has attenuated over time warrants careful comparison with the landmark study by Manohar et al., which reported near-universal recovery with early steroid initiation (39). The varying effect sizes likely stem from the definition of the “control” arm. In early studies, patients not receiving steroids were often those with unrecognized AKI or those deemed too frail for treatment, creating a bias against the non-steroid group. In recent real-world datasets, the non-steroid group increasingly includes patients with “presumed ATN” or mild AKI managed with hydration alone—many of whom recover spontaneously. This mirrors the trajectory seen in other irAEs, such as ICI-hepatitis, where guidelines have shifted from mandatory steroids to a “watch-and-wait” approach for lower-grade toxicity (40). Our data supports the position of the revised 2025 ASCO/NCCN guidelines, suggesting that a trial of conservative management is appropriate for Grade 1 and selected Grade 2 ICI-AKI cases, reserving steroids for progressive or biopsy-proven disease (41).

The shift in recurrence rates also invites comparison with the fundamental biology of T-cell memory. Research by Rauber et al. suggests that tissue-resident memory T cells (TRM) drive recurrent irAEs (42). The lower recurrence rates observed in our updated analysis might imply that prolonged intervals between the initial event and rechallenge (a common feature in recent studies) allow for the attrition of these pathogenic kidney-resident clones. Alternatively, the “successful” rechallenge patients might represent a distinct immunological phenotype—those whose initial AKI was driven by transient, circulating effectors rather than entrenched tissue-resident memory (43).

A novel and hypothesis-generating finding of our study is the “Age Paradox,” where younger patients face a higher incidence of initial injury, while older patients face a higher risk of recurrence. This dichotomy can be understood through the lens of immunosenescence and nephron reserve. Younger patients typically possess a more robust and responsive immune system. Upon checkpoint blockade, the magnitude of T-cell disinhibition and the subsequent “cytokine storm” may be more intense, leading to a higher likelihood of off-target inflammation (higher incidence) (44). This aligns with data from melanoma trials showing that younger age is a predictor for high-grade irAEs (45). However, younger patients also possess a robust “renal functional reserve”—a high density of healthy nephrons capable of repair and compensatory hypertrophy (46). Consequently, even if they experience a second inflammatory hit upon rechallenge, their kidneys can withstand the insult without manifesting as clinical AKI (lower recurrence). Conversely, elderly patients (affected by “inflammaging”) may mount a more sluggish initial autoimmune response. However, their kidneys are characterized by glomerulosclerosis, tubulointerstitial fibrosis, and a critical lack of functional reserve (47). When these patients are rechallenged, even a minor, sub-clinical inflammatory insult—which a younger kidney would absorb—may precipitate a decline in GFR that crosses the threshold of clinical AKI (higher recurrence). This hypothesis, while requiring mechanistic validation, suggests that age should be a primary factor in the risk-benefit calculation for rechallenge: younger patients can be rechallenged with confidence, whereas older patients require stringent monitoring of renal function, potentially utilizing novel biomarkers like urine retinol-binding protein or kidney injury molecule-1 (KIM-1) to detect sub-clinical injury before overt failure (48).

Our study has several limitations inherent to meta-analyses of observational data. First, the high heterogeneity observed in the incidence analysis persists despite subgroup stratification, reflecting the irreducible variability in clinical practice and patient populations across the globe. Second, the included studies utilized varying definitions of AKI, ranging from creatinine-based KDIGO criteria to symptom-based CTCAE grading. This heterogeneity, combined with the lack of universal biopsy confirmation in registry data, may result in the misclassification of hemodynamic AKI (ATN) as ICI-AKI, potentially diluting incidence estimates and confounding steroid efficacy analyses. While we prioritized studies with adjudicated outcomes, some misclassification of ATN as ICI-AKI is inevitable in real-world data, which may dilute the apparent efficacy of steroids. Third, while we identified an age-related signal, we lacked granular data to control for specific comorbidities (e.g., pre-existing CKD stages) in the meta-regression, which could confound the relationship between age and recurrence. Additionally, the ‘rechallenge’ population is subject to significant survivorship bias. This cohort represents a highly selected subgroup of patients who achieved renal recovery and maintained a performance status sufficient for further therapy. Therefore, our pooled recurrence rate of ~14% should be interpreted as the risk for eligible, recovered candidates, and cannot be generalized to patients with residual renal dysfunction or severe initial injury. Finally, our funnel plot analysis reveals an asymmetry consistent with publication bias or ascertainment bias in smaller, early cohorts (Seethapathy et al., Meraz-Munoz et al.), which reported rates >5%. Conversely, recent large-scale registries (Chuang et al.) stabilize the incidence at ~1.9-2.6%. This confirms that ‘real-world’ incidence is lower than initially feared in academic centers.

This updated meta-analysis, anchored by the most recent large-scale evidence, redefines the clinical framework for ICI-associated AKI. We provide compelling evidence that the risk of recurrent AKI upon immunotherapy rechallenge is substantially lower than historically reported, overturning a decade of therapeutic nihilism. Furthermore, we unveil a critical divergence in risk profiles based on age, identifying younger patients as a high-incidence but high-resilience subgroup, and the elderly as a vulnerable population during rechallenge. These findings should empower clinicians to abandon the practice of permanent ICI discontinuation for all AKI cases. Instead, we advocate for a confident, active management strategy: optimizing renal recovery with judicious steroid use and aggressively pursuing rechallenge in eligible patients, thereby ensuring that the kidneys do not become a barrier to cancer survival.