Многовариативная модель прогнозирования ранней смертности у пациентов с множественными миеломами с дефектами рената.

Introduction

Multiple myeloma (MM), a malignant plasma cell disorder, remains incurable despite transformative therapeutic advances over the past decades [ 1 ]. A critical complication driving early mortality is renal impairment (RI), affecting 20%–50% of newly diagnosed MM (NDMM) patients [ 2 ]. MM patients with renal impairment (RI) have a poor clinical prognosis, with shorter median overall survival and increased risk of early death than patients without renal impairment [ 3 , 4 ]. MM patients with RI achieving kidney function restoration demonstrated superior overall survival (median 56 months) compared to those with persistent renal impairment (median 33 months; p < 0.001) [ 5 ]. These patients demonstrate a pronounced temporal concentration of mortality within the first year from diagnosis, characterized by a strikingly rapid rise to 12% within just 2 months [ 6 ].

Current prognostic tools, such as the revised international staging system (R‐ISS), inadequately address early mortality in MM‐RI cohorts [ 2 ]. Recent efforts have focused on renal recovery prediction. Retrospective analyses demonstrate a significant correlation between serum involved free light chain (iFLC) levels and renal impairment, with iFLC reduction predictive of renal recovery [ 7 ]. Regarding overall survival (OS), dialysis‐dependent patients demonstrated the shortest median OS of 25 months, which was significantly inferior to both dialysis‐independent patients (45 months) and those achieving renal recovery (69 months) [ 8 ]. High bone marrow plasma cell percentage is recognized as an independent adverse prognostic factor, which is typically associated with elevated free light chain production, thereby increasing the risk of severe cast nephropathy and irreversible tubular injury [ 9 ]. Advanced age and hypercalcemia were established as independent adverse prognostic factors in MM patients with renal impairment. They contribute to poorer outcomes by reflecting increased comorbidity, diminished physiological reserve, and direct renal toxicity, which collectively exacerbate kidney injury and hinder recovery [ 6 , 10 ]. Additional prognostic determinants—including renal recovery status, hematologic response, and transplantation eligibility—demonstrate heterogeneous impacts on survival outcomes in myeloma‐associated renal impairment, as evidenced by diverse clinical studies [ 5 , 11 , 12 ].

Although diverse prognostic factors have been characterized in MM‐RI cohorts, no validated model currently exists to predict early mortality (< 12 months) in this population, underscoring an urgent need for dedicated risk assessment tools to inform clinical management. This study aims to develop and validate a multivariable early mortality prediction model for MM‐RI patients, addressing an unmet need for risk stratification to guide intensive interventions or palliative care.

Methods

Study Design and Patient Cohorts

This multicenter retrospective study analyzed newly diagnosed adult MM patients with RI treated between 2007 and 2020. From an initial pool of 334 previously reported cases, we excluded patients lacking baseline bone marrow plasma cell (BMPC) percentages or follow‐up hematologic/renal response data. The training cohort comprised 222 patients treated at Chaoyang Hospital and external validation utilized 63 patients from two independent centers: Chinese PLA General Hospital ( n = 40) and Peking University People's Hospital ( n = 23) (Figure 1 ). The median follow‐up time was 34.0 months (range 0.5–127.0) for the training cohort and 47.5 months (range 1.0–132.0) for the validation cohort. The major diagnostic criteria for MM include confirmation by bone marrow biopsy of plasmacytoma or bone marrow plasma cells (BMPC) > 10% met the International Myeloma Working Group (IMWG) criteria. RI was defined by International Myeloma Working Group (IMWG) criteria as estimated glomerular filtration rate (eGFR) < 40 mL/min/1.73 m 2 or sCr > 2 mg/dL (177 umol/L) [ 13 ]. Patients met IMWG diagnostic criteria for MM and exhibited RI at diagnosis. Exclusion criteria encompassed: non‐myeloma‐related renal pathology; monoclonal gammopathy of renal significance (MGRS) variants; severe comorbidities (uncontrolled infections, advanced heart failure or hepatic insufficiency). The study was approved by the Beijing Chaoyang Hospital Institutional Review Board.

Assessment of Hematological Response and Renal Response to Therapy

The hematological response was assessed on the first day of each treatment cycle and within 30 days following the last therapy cycle. Hematological response to therapy was defined according to the standard IMWG response criteria for MM [ 14 ]. Renal function was assessed by deriving eGFR, based on the MDRD (Modification of Diet in Renal Disease) formula [ 13 ]. Renal response was evaluated according to the IMWG criteria for the definition of the renal response to anti‐myeloma therapy [ 15 ]. The renal recovery was defined according to the IMWG criteria: eGFR from < 15 mL/min/1.73 m 2 to ≥ 15 mL/min/1.73 m 2 , 15–29 mL/min/1.73 m 2 to ≥ 30 mL/min/1.73 m 2 , or 30–59 mL/min/1.73 m 2 to ≥ 60 mL/min/1.73 m 2 .

Data Collection and Variables

All data were retrospectively extracted through comprehensive manual review of medical records across participating institutions. Trained hematologists abstracted baseline and follow‐up variables using a standardized electronic case report form (eCRF). Source documents included: (1) inpatient/outpatient electronic medical records (EMRs), (2) laboratory information systems, (3) imaging archives, and (4) scanned paper charts for historical cases. Verified parameters comprised: demographics, disease characteristics, renal parameters, treatment variables, response outcomes, and survival data. Ethical approval granted waiver of informed consent under institutional review board protocols.

Statistical Analysis

Variables with > 10% missing data, including baseline cytogenetic profiles and serum free light chain measurements, were excluded from the prognostic analysis to ensure analytic robustness. Missing data were addressed using multiple imputation by chained equations (MICE) with sensitivity analyses. Continuous variables are reported as mean ± SD (normal distribution) or median [IQR] (non‐normal distribution), categorical variables as counts (%), with baseline differences between derivation and validation cohorts assessed using t ‐tests, Mann–Whitney U tests, or χ 2 /Fisher's exact tests, supplemented by standardized mean differences (SMD < 0.10 indicating balance). Prognostic factors for 12‐month mortality were identified through univariable logistic regression ( p < 0.10 inclusion threshold), followed by multivariable analysis using backward stepwise selection with the Akaike Information Criterion ( p < 0.05 retention threshold). The resulting model was converted to a clinical nomogram using the RMS package in R (v4.2.0), internally validated via 200 bootstrap replicates to generate optimism‐corrected calibration curves (Lowess smoothing) and discrimination metrics (AUC with 95% CI), and externally validated in an independent cohort through receiver operating characteristic analysis and Hosmer‐Lemeshow goodness‐of‐fit testing. Based on the established predictive model, we scored the risk factors and divided them into high, medium, and low risk groups according to the score size. The Kaplan–Meier was used for survival analysis (IBM SPS 26.0).

Results

Patient Characteristics

A total of 285 newly diagnosed multiple myeloma (MM) patients with renal impairment (RI) were retrospectively recruited between January 2016 and December 2020, comprising a primary training cohort ( n = 222) and an independent temporal validation cohort ( n = 63) (Table 1 ). Baseline characteristics across both cohorts reflected the high‐risk nature of the population, with the overall group demonstrating a median age of 61 years (range: 37–85), male predominance (62.8%), and near‐universal advanced disease (97.5% ISS Stage III). The cohorts were well‐balanced in most parameters: sex distribution (training cohort: 62.2% male vs. validation: 65.1%, p = 0.179), hemoglobin levels (median 77 vs. 81 g/L, p = 0.235), platelet counts (median 152 × 10 9 /L vs. 183 × 10 9 /L, p = 0.212), lactate dehydrogenase (median 179 vs. 168 U/L, p = 0.257), and bone marrow plasma cell (BMPC) infiltration (median 40.0% vs. 32.5%, p = 0.166). However, significant inter‐cohort differences were noted in M‐protein isotypes ( p = 0.039) with higher IgG prevalence in the training cohort (39.6% vs. 31.7%) but greater light‐chain disease in the validation group (39.7% vs. 31.5%), along with expected variations in renal parameters including higher median creatinine (306.2 vs. 348 μmol/L, p = 0.042) and lower eGFR (17.8 vs. 15.0 mL/min/1.73 m 2 , p = 0.043) in the validation cohort. Treatment patterns were comparable with protease inhibitor (PI)‐based induction predominating (83.7% vs. 90.5%), similar autologous stem cell transplantation (ASCT) rates (21.6% vs. 19.0%), and dialysis utilization (25.2% vs. 27.0%). Depth of response was marginally superior in the training cohort for hematologic complete response rates (42.8% vs. 28.5%, p = 0.120) and renal response (CR + PR: 56.3% vs. 55.6%, p = 0.940), though not statistically significant. These minor variations in disease burden and renal severity between cohorts provided a robust foundation for validating the prognostic model's generalizability across heterogeneous clinical scenarios.

Establishment and Validation of Early Mortality Prediction Model in MM‐RI Patients

In the training cohort, univariate analysis was initially performed to identify prognostic factors potentially associated with early mortality (Figure S1 ). Subsequently, significant variables were selectively incorporated into multivariate analysis based on sample size adequacy and inter‐factor correlations. Multivariate logistic regression analysis employing backward stepwise elimination (variable retention criterion: p < 0.05) applied to the training cohort of 222 MM‐RI patients definitively identified five independent predictors of 12‐month all‐cause mortality: age ≥ 55 years (adjusted odds ratio [OR] = 3.476, 95% confidence interval [CI]: 1.097–11.007, p = 0.034), serum calcium level ≥ 2.5 mmol/L (OR = 3.285, 95% CI: 1.302–8.285, p = 0.012), BMPC percentage > 40% (OR = 2.899, 95% CI: 1.241–6.769, p = 0.014), failure to achieve at least complete hematologic response (OR = 20.934, 95% CI: 4.610–95.061, p < 0.001), and critically achieving partial or complete renal response (PR + CR) which conferred substantial protection against early mortality (OR = 0.269, 95% CI: 0.113–0.645, p = 0.003), with all covariates maintaining significance after adjusting for International Staging System classification, induction therapies, and dialysis requirement (Table 2 ). The final model demonstrated robust discriminative performance in the derivation cohort with an AUC of 0.858 (95% CI 0.807–0.909; p < 0.001), which was subsequently validated in an independent temporal cohort ( n = 63) that exhibited comparable baseline characteristics but poorer renal function, where it maintained clinically meaningful discrimination (AUC 0.747, 95% CI 0.594–0.901; p = 0.010), thereby confirming its clinical applicability and robustness. Calibration accuracy was rigorously assessed through bootstrap‐corrected curves (1000 resamples), revealing near‐perfect agreement between predicted and observed mortality probabilities in the primary cohort (χ 2 = 6.32, p = 0.612), while validation cohort calibration demonstrated acceptable performance (calibration intercept = −0.07, slope = 0.94) despite mild underestimation in the highest risk decile (predicted probability 0.60 vs. observed 0.70) (Figure 2 ). Critically, the protective association between renal response attainment and survival reduction (OR = 0.269) underscores the paramount importance of nephroprotective interventions in this vulnerable population, while the profound hazard associated with suboptimal hematologic response (OR = 20.934) highlights the necessity of treatment intensification strategies.

Risk Nomogram for Predicting Early Mortality for MM‐RI Patients

The prognostic nomogram presented in Figure 3 incorporates five clinically significant predictors (age, serum calcium, bone marrow plasma cell percentage, hematologic response, and renal function) and provides both visual representation of their relative contributions and a quantitative scoring system (range: 0–300 points) that stratifies patients according to their probability of early mortality (estimated range: 10%–80%). The model clearly demonstrates the substantial heterogeneity in prognostic impact among the variables. Age ≥ 55 years carries influence on adverse outcomes in this cohort. High tumor burden, reflected by a BMPC percentage > 40%, also confers a considerably strong negative impact, contributing less weight than serum calcium levels ≥ 2.5 mmol/L. Importantly, while both are modifiable indicators of treatment efficacy, hematological response status exerts a demonstrably greater influence on the calculated risk than renal response status; achieving a less than complete hematological response (< CR) adds more substantially to the risk score than achieving only a minimal renal response (< PR) or less. The nomogram efficiently combines these variably weighted factors. This visual model provides immediate clinical utility, allowing for rapid stratification of patients into distinct risk categories solely by summing the points derived from their specific characteristics. It starkly illustrates the potent adverse prognostic effect of age and high tumor burden while simultaneously highlighting that the depth of hematological and renal response is a critical modifiable factor impacting survival probability. Moreover, it quantifies how improvements in treatment response, particularly hematological response, can tangibly lower a patient's overall risk score, thus facilitating personalized prognostication and underscoring the potential survival benefit of striving for deeper disease control, especially in patients already possessing non‐modifiable high‐risk features. Validation confirmed the model's discriminatory ability and calibration accuracy.

A Prognostic Scoring System Effectively Stratifies Early Mortality Risk in MM‐RI Patients

Based on the comprehensive risk stratification model developed for early mortality (≤ 12 months) in MM‐RI patients, we established a prognostic scoring system incorporating five key clinical variables: age (< 55 years = 1 point, ≥ 55 years = 2 points), serum calcium (< 2.5 mmol/L = 1 point, ≥ 2.5 mmol/L = 2 points), bone marrow plasma cell percentage (≤ 40% = 1 point, > 40% = 2 points), hematologic response (≥ CR = 1 point, suboptimal response = 2 points), and renal response (≥ PR = 1 point, < PR = 2 points). The risk score categorized 285 patients into three prognostically distinct groups: Low Risk (5–6 points, n = 61), Mid Risk (7–8 points, n = 158), and High Risk (9–10 points, n = 66) (Figure 4 ). The early mortality rates demonstrated profoundly significant intergroup differences: Low Risk patients exhibited only 1.6% mortality (1/61), Mid Risk had 13.9% mortality (22/158), and High Risk showed dramatically elevated 43.9% mortality (29/66), with chi‐square testing confirming extreme statistical significance (χ 2 = 42.465, p < 0.001; Fisher test = 40.777, p < 0.001). The OS analysis of the total study population ( N = 285) demonstrated a median OS of 45.0 months (95% CI: 39.5–50.5). When stratified by risk groups, significant survival disparities were observed ( χ 2 = 40.31, p < 0.001) (Figure 5 ). Low‐risk patients demonstrated superior outcomes with a median OS of 74.0 months (95% CI: 63.4–84.6), while intermediate‐risk patients showed significantly shorter median OS of 46.0 months (95% CI: 40.6–51.4) ( p < 0.001). High‐risk patients had markedly poorer prognosis, with median OS of only 14.0 months (95% CI: 2.1–25.9), representing a 5.3‐fold difference in median survival compared to the low‐risk group. The Kaplan–Meier curves (depicting months on x ‐axis and cumulative survival probability on y ‐axis) showed complete separation between risk tiers, with Log‐Rank tests confirming overwhelmingly significant differences ( χ 2 = 45.731, p < 0.001) and Breslow tests indicating early discrimination power ( χ 2 = 45.176, p < 0.001). The risk function visualization demonstrated three non‐overlapping survival curves: a near‐horizontal trajectory in Low Risk (98.4% survival), moderate early decline in Mid Risk (86.1% survival), and precipitous drop in High Risk (56.1% survival), with the most dramatic mortality occurring within the initial 3–6 months in the High Risk cohort. Given the incomplete baseline cytogenetic data, cytogenetic factors were not incorporated into the primary prognostic model. However, for exploratory subgroup analysis, we implemented a pragmatic classification system: patients harboring ≥ 1 identified cytogenetic high‐risk feature were categorized as high‐risk ( n = 121), while those without detectable high‐risk abnormalities comprised the standard‐risk group ( n = 73). This approach enabled survival analysis in 194 evaluable patients with available cytogenetic profiles (Figure S2 ). The Kaplan–Meier analysis revealed comparable survival trajectories between groups (high‐risk median OS: 48.0 months [95% CI 32.3–63.7] vs. standard‐risk: 46.0 months [95% CI 34.5–57.5]; log‐rank p = 0.987), suggesting further investigation is warranted to elucidate the impact of cytogenetic abnormalities on OS even early mortality.

Discussion

This study successfully developed and validated a clinically applicable prognostic nomogram for MM patients with RI. The model identified five adverse prognostic factors: age ≥ 55 years, serum calcium ≥ 2.5 mmol/L, failure to achieve partial renal remission, failure to achieve complete hematologic remission, and BMPC percentage > 40%. This nomogram demonstrates strong discriminatory power (AUC: 0.858 in the training cohort; 0.747 in the validation cohort) and excellent calibration, representing a significant improvement in risk stratification for this vulnerable population. The identified prognostic factors reflect the complex pathophysiology of MM‐related renal injury and are consistent with contemporary clinical evidence and expert consensus guidelines.

The inclusion of advanced age (≥ 55 years) as an independent negative prognostic factor is consistent with extensive epidemiological data. Older patients with MM frequently present with greater comorbidity burdens, diminished renal functional reserve, and increased vulnerability to the toxicities of necessary anti‐myeloma therapies. Multiple studies, including large registry analyses, confirm that older age is consistently associated with inferior overall survival and reduced rates of renal recovery in MM, even in the era of novel agents [ 15 , 16 ]. The International Myeloma Working Group (IMWG) recognizes age as a key component in patient fitness assessment, influencing treatment goals and intensity [ 17 ]. Hypercalcemia is a well‐established adverse prognostic feature in MM, integrated into both the International Staging System (ISS) and its revised version (R‐ISS) [ 18 ]. Its strong association with poor outcomes in our renal impairment model underscores its multifaceted pathological role. Hypercalcemia directly contributes to renal dysfunction by inducing vasoconstriction, reducing glomerular filtration rate (GFR), promoting volume depletion through nephrogenic diabetes insipidus, and causing tubular damage via calcification and crystal formation [ 19 ]. Furthermore, it serves as a potent marker of high tumor burden, osteolytic activity, and aggressive disease biology [ 20 ]. Its persistence or severity often correlates with a greater load of nephrotoxic free light chains (FLCs), exacerbating cast nephropathy and directly impeding renal recovery [ 21 ]. Our analysis substantiates that relative hypercalcemia (serum calcium > 2.5 mmol/L) and advanced age (≥ 55 years) jointly function as critical adverse prognostic determinants in MM patients with RI, aligning with established pathophysiological mechanisms of age‐related renal vulnerability and calcium‐induced nephrotoxicity.

Achieving hematologic complete response (CR) emerged as the dominant modifiable predictor of early mortality in our model (adjusted OR 20.93, 95% CI 4.61–95.06). This strong association reflects a clear mechanistic link: CR rapidly and substantially reduces circulating monoclonal free light chains (FLCs), the key nephrotoxins that drive cast nephropathy and light‐chain proximal tubulopathy in MM‐related renal injury. Lower FLC levels relieve intratubular obstruction, decrease proximal tubular endocytosis‐mediated apoptosis, dampen interstitial inflammation, and interrupt progression to irreversible fibrosis, thereby promoting renal recovery and lowering the risk of fatal uremic complications, fluid overload, and infection. At the same time, CR signals effective suppression of the malignant plasma‐cell clone, which reduces overall tumor burden and mitigates systemic contributors to early death such as persistent hypercalcemia and immunosuppression. These mechanisms are supported by prior studies. Leung et al. reported significantly higher renal response rates in patients achieving ≥ VGPR, an effect driven primarily by greater FLC reduction [ 22 ]. Likewise, Rodríguez‐Otero et al. identified CR attainment as an independent predictor of renal recovery in their prognostic model for early myeloma‐related mortality [ 23 ]. Our center's evaluation of bortezomib–pomalidomide–dexamethasone induction confirmed markedly superior renal recovery among patients reaching ≥ CR compared with those achieving only PR [ 8 , 24 ]. Collectively, these data reinforce the clinical imperative for deploying the most potent regimens from diagnosis in MM patients presenting with renal impairment, a principle now quantified by our nomogram.

Crucially, the model identifies the achievement of a renal response (PR or CR) itself as a strong independent positive prognostic factor. IMWG criteria define renal response based on sustained improvement in estimated GFR (eGFR) [ 2 ]. Attaining renal response signifies successful reversal of the functional impairment caused by the underlying renal lesion (e.g., cast nephropathy, light chain proximal tubulopathy, amyloidosis [ 25 ]). It is not merely a consequence of hematological response but an active process involving resolution of tubular obstruction, reduction of interstitial inflammation, and potential intrinsic renal repair mechanisms. Studies have consistently shown that achieving a renal response is associated with significantly improved overall survival compared to patients who do not recover renal function [ 24 , 26 ]. Antlanger et al. demonstrated that patients achieving renal response after autologous stem cell transplant had survival outcomes comparable to MM patients without baseline renal impairment [ 27 ]. Therefore, its inclusion in our model (OR = 0.269 for not achieving PR/CR) powerfully captures the direct impact of functional renal recovery on patient prognosis, acting as a critical endpoint in its own right. It validates the clinical focus on renal response as a key therapeutic goal.

A high bone marrow plasma cell percentage (> 40%) emerged as another independent negative prognostic factor. This reflects a higher burden of malignant plasma cells, typically correlating with elevated FLC production and, consequently, a greater risk of severe cast formation and irreversible tubular injury [ 28 ]. High plasma cell infiltration is often associated with more aggressive disease biology and cytogenetic abnormalities, potentially explaining its association with poorer renal recovery and survival [ 9 ]. While the R‐ISS incorporates cytogenetics rather than plasma cell percentage directly for defining high‐risk disease [ 2 ], baseline plasma cell burden remains a practical and accessible indicator of disease intensity. Its significance in our model suggests that patients presenting with very high tumor loads, even if achieving good hematological responses, may still face challenges in renal recovery, possibly due to pre‐existing irreversible structural damage or a higher intrinsic resistance to therapy.

This study's primary contribution lies in synthesizing these interrelated factors—encompassing patient factors (age), disease burden or aggressiveness markers (hypercalcemia and plasma cell percentage), treatment efficacy endpoints (hematologic and renal response ≥ CR, Renal Response PR/CR)—into a single, clinically practical prognostic tool. While individual factors are known, the development of a validated multivariable model that quantifies their combined prognostic weight specifically for patients with renal impairment is novel and addresses a significant unmet need. We also stratified patients into three distinct risk tiers, revealing drastically different early mortality rates and survival trajectories, enabling clinical risk‐tailored interventions. Although significant inter‐cohort differences were observed in M‐protein isotype distribution ( p = 0.039, with a higher proportion of light‐chain disease in the validation cohort) and ISS stage ( p = 0.03), together with modestly worse baseline renal function in the validation set (median creatinine 348 vs. 306.2 μmol/L, p = 0.042; eGFR 15.0 vs. 17.8 mL/min/1.73 m 2 , p = 0.043), these imbalances did not compromise the performance or generalizability of the early‐mortality prediction model. As emphasized in the 2023 International Myeloma Working Group (IMWG) recommendations, light‐chain myeloma is a well‐established driver of cast nephropathy and severe renal impairment in MM patients. Critically, the five independent predictors retained after multivariable backward selection—age ≥ 55 years, serum calcium ≥ 2.5 mmol/L, BMPC percentage > 40%, hematologic response < CR, and renal response ≥ PR—exhibited consistent distributions and effect directions across cohorts. In accordance with TRIPOD guidelines, which stress the importance of external validation in heterogeneous case mixes to assess model transportability, internal bootstrap validation (200 replicates) confirmed excellent discrimination and calibration in the training cohort (AUC 0.858), while the model retained clinically meaningful performance in the independent validation cohort despite greater renal severity (AUC 0.747). These results affirm the nomogram's robustness across real‐world variations in disease presentation.

Despite the multicenter design and external validation that strengthened this study, several limitations should be acknowledged. First, its retrospective nature and reliance on historical records meant that high‐risk cytogenetic data were incomplete, while minimal residual disease (MRD) status and serial serum free light chain measurements were not systematically captured. Consequently, we could not directly assess the kinetic relationship between free light chain reduction and renal recovery or explore potential interactions with cytogenetic risk. Second, although the multivariable model retained strong performance in both training and validation cohorts, the absence of these contemporary biomarkers limits deeper mechanistic insight into why hematologic complete response exerted such a pronounced protective effect. Finally, the relatively modest size of the validation cohort and the era of treatment (2007–2020) may influence applicability in settings where newer agents or quadruplet regimens are used. Prospective studies with standardized longitudinal biomarker monitoring, comprehensive cytogenetic profiling, and larger, more contemporary cohorts will therefore be required to confirm the model's generalizability and further refine its clinical utility.

In conclusion, this study establishes a validated prognostic nomogram integrating three adverse predictors (age ≥ 55 years, serum calcium ≥ 2.5 mmol/L, BMPC > 40%) and two treatment‐responsive protective factors (≥ CR hematologic response, ≥ PR renal response). This model provides clinicians with a practical, evidence‐based tool for individualized risk assessment in MM patients presenting with renal impairment. By stratifying early mortality risk using on‐treatment biomarker dynamics, this tool directs nephroprotective interventions and precision counseling to improve renal recovery and survival. Its implementation, following further external validation, represents a significant step towards precision medicine in the management of myeloma‐related kidney disease.

Author Contributions

Yuan Jian: methodology. Jing Jia: data curation. Menghan Liu: writing – original draft. Huixing Zhou: software. Xinan Cen: resources. Wenming Chen: visualization. Jiachen Li: formal analysis. Honghu Zhu: supervision. Guangzhong Yang: investigation. Chuanying Geng: supervision. Wenrong Huang: resources. Wen Gao: conceptualization, writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

This retrospective study was approved by the institutional ethics committee of the Department of Hematology, Beijing Chaoyang Hospital, Capital medical University.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Forest plot of OR: Univariate analysis.

Figure S2: Survival function.

Acknowledgments

The authors thank the patients who make the study possible, the clinical study teams, and all the investigators.

Data Availability Statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

References

1. Kumar S. K. and Rajkumar S. V., “The Multiple Myelomas—Current Concepts in Cytogenetic Classification and Therapy,” Nature Reviews. Clinical Oncology 15, no. 7 (2018): 409–421.

2. Dimopoulos M. A., Merlini G., Bridoux F., et al., “Management of Multiple Myeloma‐Related Renal Impairment: Recommendations From the International Myeloma Working Group,” Lancet Oncology 24, no. 7 (2023): e293–e311.

3. Kleber M., Ihorst G., Deschler B., et al., “Detection of Renal Impairment as One Specific Comorbidity Factor in Multiple Myeloma: Multicenter Study in 198 Consecutive Patients,” European Journal of Haematology 83 (2009): 519–527.

4. Eleutherakis‐Papaiakovou V., Bamias A., Gika D., et al., “Renal Failure in Multiple Myeloma: Incidence, Correlations, and Prognostic Significance,” Leukemia & Lymphoma 48 (2007): 337–341.

5. Gonsalves W. I., Leung N., Rajkumar S. V., et al., “Improvement in Renal Function and Its Impact on Survival in Patients With Newly Diagnosed Multiple Myeloma,” Blood Cancer Journal 5 (2015): e296.

6. Courant M., Orazio S., Monnereau A., Preterre J., Combe C., and Rigothier C., “Incidence, Prognostic Impact and Clinical Outcomes of Renal Impairment in Patients With Multiple Myeloma: A Population‐Based Registry,” Nephrology, Dialysis, Transplantation 36, no. 3 (2021): 482–490.

7. Yadav P., Cockwell P., Cook M., et al., “Serum Free Light Chain Levels and Renal Function at Diagnosis in Patients With Multiple Myeloma,” BMC Nephrology 19, no. 1 (2018): 178.

8. Jian Y., Zhou H. X., Xie W. W., et al., “Impact of Dialysis Dependence on Survival for Multiple Myeloma With Renal Impairment: A Multicenter Study in China,” American Journal of Cancer Research 13, no. 4 (2023): 1571–1581.

9. Vignon M., Cohen C., Faguer S., et al., “The Clinicopathologic Characteristics of Kidney Diseases Related to Monotypic IgA Deposits,” Kidney International 91, no. 3 (2017): 720–728.

10. Wang L., Liu C., Song H., Yuan J., Zha Y., and Deng Y., “Update on Kidney Injury Caused by Multiple Myeloma,” Annals of Hematology 103, no. 12 (2024): 5007–5018.

11. Lin Z. S., Qin A. B., Wang S. X., et al., “A Non‐Invasive Differential Diagnostic Model for Light Chain Cast Nephropathy in Newly Diagnosed Multiple Myeloma Patients With Renal Involvement: A Multicenter Study,” Journal of Nephrology 34, no. 4 (2021): 1169–1177.

12. Mohyuddin G. R., Abbasi S., Okoniewski M., et al., “Inpatient Mortality of Patients With Multiple Myeloma and Renal Impairment Undergoing Autologous Stem Cell Transplantation,” European Journal of Haematology 105, no. 5 (2020): 571–577.

13. Rajkumar S. V., “Updated Diagnostic Criteria and Staging System for Multiple Myeloma,” American Society of Clinical Oncology Educational Book 35 (2016): e418–e423.

14. Kumar S., Paiva B., Anderson K. C., et al., “International Myeloma Working Group Consensus Criteria for Response and Minimal Residual Disease Assessment in Multiple Myeloma,” Lancet Oncology 17 (2016): e328–e346, 10.1016/s1470-2045(16)30206-6.

15. Zhao Y., Niu D. D., Ye E. L., et al., “Secular Trends in the Burden of Multiple Myeloma From 1990 to 2019 and Its Projection Until 2044 in China,” Frontiers in Public Health 8, no. 10 (2022): 938770.

16. Arfani C. E., Veirman K. D., Maes K., et al., “Metabolic Features of Multiple Myeloma,” International Journal of Molecular Sciences 19, no. 4 (2018): 1200.

17. Dimopoulos M. A., Sonneveld P., Leung N., et al., “International Myeloma Working Group Recommendations for the Diagnosis and Management of Myeloma‐Related Renal Impairment,” Journal of Clinical Oncology 34, no. 13 (2016): 1544–1557.

18. Palumbo A., Loiseau H. A., Oliva S., et al., “Revised International Staging System for Multiple Myeloma: A Report From International Myeloma Working Group,” Journal of Clinical Oncology 33, no. 26 (2015): 2863–2869.

19. Ralston S. H., “Pathogenesis and Management of Cancer Associated Hypercalcaemia,” Cancer Surveys 21 (1994): 179–196.

20. Roodman G. D., “Pathogenesis of Myeloma Bone Disease,” Leukemia 23, no. 3 (2009): 435–441.

21. Hutchison C. A., Batuman V., Behrens J., et al., “The Pathogenesis and Diagnosis of Acute Kidney Injury in Multiple Myeloma,” Nature Reviews. Nephrology 8, no. 1 (2011): 43–51.

22. Leung N., Gertz M. A., Zeldenrust S. R., et al., “Improvement of Cast Nephropathy With Plasma Exchange Depends on the Diagnosis and on Reduction of Serum Free Light Chains,” Kidney International 73, no. 11 (2008): 1282–1288.

23. Rodríguez‐O P., Mateos M. V., López J. M., et al., “Early Myeloma‐Related Death in Elderly Patients: Development of a Clinical Prognostic Score and Evaluation of Response Sustainability Role,” Leukemia 32, no. 11 (2018): 2427–2434.

24. Jian Y., Chang L., Shi M. X., et al., “Pomalidomide, Bortezomib, and Dexamethasone for Newly Diagnosed Multiple Myeloma Patients With Renal Impairment,” Blood Advances 7, no. 24 (2023): 7581–7584.

25. Mitchell H., Rosner M. D., Mark A., et al., “Acute Kidney Injury in Patients With Cancer,” New England Journal of Medicine 376 (2017): 1770–1781.

26. Dimopoulos M. A., Roussou M., Gavriatopoulou M., et al., “Reversibility of Renal Impairment in Patients With Multiple Myeloma Treated With Bortezomib‐Based Regimens: Identification of Predictive Factors,” Clinical Lymphoma & Myeloma 9, no. 4 (2009): 302–306.

27. Antlanger M., Dust T., Reiter T., et al., “Impact of Renal Impairment on Outcomes After Autologous Stem Cell Transplantation in Multiple Myeloma: A Multi‐Center, Retrospective Cohort Study,” BMC Cancer 18, no. 1 (2018): 1008.

28. Karam S., Haidous M., Dalle I. A., et al., “Monoclonal Gammopathy of Renal Significance: Multidisciplinary Approach to Diagnosis and Treatment,” Critical Reviews in Oncology/Hematology 183 (2023): 103926.

Associated Data

Supplementary Materials

Figure S1: Forest plot of OR: Univariate analysis.

Figure S2: Survival function.

Data Availability Statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.