IDENTIFICATION OF PREDICTIVE FACTORS FOR SEVERE INFECTION IN MULTIPLE MYELOMA: A RISK MODEL DEVELOPMENT AT TERTIARY CARE CENTER

M.F. MONTES-RODRIGUEZ¹, A.OROZCO-COLLAZO¹, S.BURGOS-CANALES¹ and A.SANCHEZ-RODRIGUEZ²

1. Department of Hematology, Hospital Juarez de Mexico, Mexico City, Mexico

2. Department of Internal Medicine, Centro Medico ABC, Mexico City, Mexico

 

INTRODUCTION

Despite remarkable therapeutic advances, infectious complications represent a leading cause of morbidity and non-relapse mortality in Multiple Myeloma (MM). This vulnerability stems from disease-related inmune dysfunction, cumulative immunosuppression from sequential therapies and host factors.

As survival improves and treatment strategies become increasingly complex, identifying high risk patients for severe infections remains a critical challenge, limited by a lack of objective, biomarker – based predictive tools beyond mere clinical judgment.

AIM

To identify predictive factors of severe infection by assessing the baseline clinical and microbiological characteristics of Multiple Myeloma patients treated at a tertiary care hospital.

METHODS

Study design: Observational, retrospective and cross-sectional study.
Setting: Tertiary care referral center, Centro Medico ABC, Mexico City, Mexico.
Patients: 75 patients with confirmed MM hospitalized between January 2023 and December 2024.
Primary outcome: Severe infection (CTCAE grade >4, version 5.0).
Statistical analysis: Associations were evaluated via uni/multivariate logistic regression (ORs). CRP discrimination and model performance were assessed using ROC-AUC curves, sensitivity, specificity, accuracy, and Nagelkerke R2.

RESULTS

Cohort: A total of 75 patients were included, 17.3% (n=13) developed sever infection. The mean age 70.0 + 11.3 years; 73.3% male. At infection onset, 50.7% were receiving antimicrobial prophylaxis, and pneumonia (40.5%) along with bacteremia (31.0%)were the primary manifestations. D-VRd was the most frequent regimen (33.3%), notably, lenalidomide treatment show a trend toward severe infection (100% vs 70%, p=0.066), while CRd correlated with increased severe episodes (p=0.028).

Univariate Analysis: In the comparative analysis, severe infection cases presented significantly higher mean creatinine levels (1.74 vs 1.13 mg/dl, p < 0.001) as well as markedly elevated CRP (21.86 vs 5.51 mg/L, p=0.0035).

Multivariable Model: After adjustments, febrile neutropenia (OR 8.04, 95% CI 1.76-36.64; p=0.007) and CRP > 12.45 mg/L (OR 9.77, 95% CI 2.19 – 43.70;-p=0.003) independently predicted severe infection. Additionally, CRP discrimination yielded an AUC of 0.762. The final predictive model demonstrated 30.8% sensitivity, 96.8% specificity, and 85.0% overall accuracy (Nagelkerke R² 0.290).

 

CONCLUSIONS

In this cohort, elevated CRP (>12.45 mg/L) combined with febrile neutropenia at presentation were independent predictors of severe infection. This highly specific (96.8%) and accurate (85%) clinical-biochemical combination offers an accessible framework to address current gaps in Latin American practice, refine clinical decision-making and optimizing high-risk patient care.