Integrating the cardiovascular-renal-metabolic syndrome（ckm）: Construction and validation of a comprehensive model for early identification of high-risk multiple myeloma

Abstract BackgroundHigh-risk multiple myeloma (HRMM) is a subtype that accounts for approximately 15-20% of MM patients and is characterized by early relapse or treatment refractoriness, even in the context of autologous stem cell transplantation and novel pharmacological agents. Currently, the risk stratification of HRMM predominantly relies on genetic markers; however, this framework exhibits limitations in facilitating early detection and accurate prognostication, thereby inadequately addressing the clinical requirements for precision management. The recently articulated concept of Cardiovascular-Kidney-Metabolic (CKM) syndrome, particularly in its stages 3-4,which correlate with increased all-cause mortality, underscores the complex interrelationships among metabolic, cardiac, and renal factors and presents a promising avenue for a novel prognostic indicator in MM. The development of a composite prognostic model that integrates tumor burden, genetic risk, and CKM stage stratification has the potential to significantly improve outcome prediction for HRMM and inform individualized therapeutic strategies.MethodsThis study examined clinical data from 301 patients treated at Henan Provincial People's Hospital from January 2017 to December 2024, all of whom received ≥4 chemotherapy cycles and first-line therapy. Patients with ASCT, unclear diagnoses, or incomplete genetic data were excluded. Ultimately, 241 patients were split into experimental (192) and validation groups (49) in an 8:2 ratio. Logistic and LASSO regressions identified model predictors, evaluated with ROC, calibration, and decision curves, and Bootstrap resampling. Statistical analysis used SPSS version 22.0.ResultsWe analyzed 241 non-transplanted NDMM patients with a median age of 56.9, categorizing them into 54 (22.4%) high-risk and 187 (77.6%) non-high-risk groups based on overall survival of less than 3 years. The high-risk group exhibited more aggressive disease, a higher tumor load, and 14.8% had soft tissue-related extramedullary disease (EMD-S), compared to 4.3% in the non-high-risk group (P=0.011). They also had a higher median bone marrow plasma cell percentage (32.5% vs. 24.8%, P < 0.05) and significant differences in CKM (P = 0.023) and high-risk FISH genetics (HRCA > 2) (P = 0.011).This study used logistic regression to analyze variables like age, gender, height, weight, hemoglobin, PLT, LDH, creatinine, BMPC%, peripheral blood plasma cell ratio, ISS, R-ISS, EMD-S, HRCA, CKM, chemotherapy regimen, and date of death/follow-up, collected after diagnosis but before treatment. It found age (>65), PLT, LDH, ISS, EMD-S, BMPC%, CKM, and HRCA as significant risk factors (P < 0.1). These were included in a LASSO regression model, showing age, platelets, LDH, ISS, EMD-S, CKM, and HRCA significantly influenced predictions. The predictive model is: logist(p) = –2.23 + 0.98 × age + 0.50 × ISS2 + 0.90 × ISS3 + 2.06 × EMD-S – 0.003 × PLT + 0.49 × LDH + 0.35 × CKM1 – 0.005 × CKM2 + 1.18 × CKM3-4. The main risk factors identified were EMD-S (OR = 7.85), age (OR = 2.66), and CKM 3-4 (OR = 3.25).Kaplan-Meier analysis stratified by CKM stage demonstrated statistically significant differences (p =0.011).The model evaluation reveals an ROC curve area of 0.731 (0.640-0.822) for the experimental group, while the validation group's ROC area is currently unavailable due to data limitations. The calibration curve shows a Brier Score of 0.145 and a Hosmer-Lemeshow Test result of P=0.3. Internal validation using bootstrap resampling (500 iterations) yielded a corrected C-statistic of 0.73 (0.630-0.81), confirming model stability. Decision curve analysis (DCA) indicated a net benefit exceeding both the all and none lines between thresholds of 0.2 and 0.6, highlighting good clinical utility. Various validation methods confirmed the model's high clinical utility, offering robust decision support for physicians.ConclusionThis study developed a predictive model to identify HRMM in non-transplanted NDMM patients using accessible clinical data, enhancing early detection and prognosis. Our model incorporates novel metrics like CKM, EMD-S, and HRCA to better assess disease aggressiveness and quantify the combined impact on the cardio-renal-metabolic system.