Quantitative Integrative Survival Prediction in Multiple Myeloma Patients

Medicine Basic Medical Sciences Oncology Hematology
Multiple Myeloma Survival Prediction Quantitative Model Molecular Prognostic Factors Continuous Risk Assessment

Precision Medicine | Quantitative Comprehensive Survival Prediction for Patients with Multiple Myeloma: Based on Bortezomib Induction Therapy, High-Dose Treatment, and Autologous Stem Cell Transplantation

Introduction

Multiple myeloma is a malignant hematological disease characterized by the accumulation of clonal plasma cells within the bone marrow, associated with clinical manifestations related to impaired hematopoiesis and osteolytic bone disease. The prognosis of multiple myeloma patients is highly heterogeneous, with survival times ranging from several months to over 15 years. In clinical practice, risk stratification is usually conducted by combining high-risk chromosomal abnormalities detected by interphase fluorescence in situ hybridization (iFISH) and the International Staging System (ISS). The widely accepted standard currently is the revised ISS score (R-ISS), which includes serum B2-microglobulin, albumin, lactate dehydrogenase (LDH), and adverse prognostic chromosomal abnormalities. However, existing risk prediction models only categorize patients into two to four risk groups, such as high-risk, intermediate-high-risk, intermediate-low-risk, and low-risk groups. These groupings often fail to accurately reflect individualized survival predictions. Therefore, precisely predicting the survival probability of individual patients has significant potential clinical applications.

This study aims to develop a quantitative prediction tool to assess the 3-year and 5-year overall survival (OS) probability of individual multiple myeloma patients and validate its risk stratification capability.

Study Source

This study was a collaboration led by Manuela Hummel, Thomas Hielscher, Martina Emde-Rajaratnam, Hans Salwender, Susanne Beck, Christof Scheid, Uta Bertsch, Hartmut Goldschmidt, Anna Jauch, Jérôme Moreaux, Anja Seckinger, and Dirk Hose. It was published in JCO Precision Oncology on July 10, 2024.

Study Process

Objectives and Methods

The objective of the study was to develop a quantitative prediction tool to estimate the 3-year and 5-year OS probabilities of individual multiple myeloma patients. The study included the following steps:

Study Process

  1. Study Subjects:

    • Included 657 untreated multiple myeloma patients needing treatment.
    • All patients signed informed consent forms.
    • Patients underwent bortezomib-based induction followed by planned high-dose chemotherapy and autologous stem cell transplantation (ASCT).

  2. Sample Processing:

    • Plasma cells labeled with CD138 were isolated and purified from bone marrow aspirates.
    • Purified plasma cells underwent iFISH and nucleic acid extraction for gene expression profiling (GEP) analysis.

  3. iFISH Analysis:

    • Probes were used to detect changes in the number of multiple chromosomal regions and translocations.
    • Data were analyzed according to established methods.

  4. Gene Expression Analysis:

    • RNA was extracted using commercial kits, followed by quality control and quantification.
    • GEP analysis was performed using Affymetrix U133 2.0 plus arrays.
    • Expression data were stored in ArrayExpress.

  5. Statistical Analysis:

    • Patients were divided into a training group (n=536) and a validation group (n=121).
    • A Cox regression model was used to establish a prognostic model.
    • Variables with missing values were imputed.
    • A stepwise variable selection procedure optimizes the model.
    • A Cox model was used to construct a nomogram for estimating survival probabilities.

  6. Validation and Comparison:

    • The nomogram was validated in an external validation cohort to verify its discriminative ability and calibration.
    • The model’s predictive performance was compared with R-ISS, R2-ISS, and Mayo-2022 scores.

Experimental Methods

A quantitative comprehensive survival prediction tool was established using a Cox regression model combined with established risk factors. These risk factors included age, ISS stage, LDH, creatinine level, heavy chain type IgA, presence of del17p13, t(4;14), and 1q21 gain (copy number gain), and prognostic indicators based on GEP (e.g., UAMS GEP70-score and GPI50).

Research Results

Main Results

  1. Establishment of the Nomogram:

    • Based on the training data, a Cox model successfully constructed the nomogram for estimating 3-year and 5-year survival probabilities.
    • Each predictive factor was assigned an appropriate score; the total score could be converted into a continuous OS probability.

  2. Model Validation:

    • The nomogram showed good discriminatory ability in the validation dataset (C-index 0.76 in the training group, 0.75 in the validation group).
    • The nomogram’s 3-year survival prediction was significantly differentiated from those of the R-ISS (p < .001) and R2-ISS (p < .01).
    • The effectiveness of the staging was validated using time-dependent receiver operating characteristic (AUC) curves.

  3. Model Comparison:

    • The continuous risk assessment model demonstrated better discriminative ability compared to the R-ISS and R2-ISS models in both the training and validation groups.
    • The time-dependent AUC values of the model were also significantly better than those of the existing models.

  4. Calibration:

    • The model showed good calibration ability in both the training and validation groups, indicating accurate prediction of survival probabilities.

Conclusions and Value

  1. Conclusion:

    • The study developed and validated a nomogram-based quantitative individualized survival prediction tool.
    • The continuous risk assessment, which incorporated molecular prognostic factors, was superior to the R-ISS, R2-ISS, or Mayo-2022 scores alone.

  2. Significance of the Study:

    • The study achieved a more precise individualized survival prediction for multiple myeloma patients, usable as a routine clinical risk assessment tool.
    • By integrating serum and molecular prognostic factors, continuous risk assessment can finely and individually stratify risks.
    • The study promotes the clinical routine use based on molecular profiling, enhancing the accuracy and practicality of prognosis assessment.

  3. Highlights of the Study:

    • The study pioneered methods for quantitatively evaluating the survival probability of individual multiple myeloma patients, aiding personalized treatment decisions.
    • The measures support broader application development, promoting the use of molecular profiling methods in clinical practice to achieve precision medicine.