Nucleic acid based risk assessment and staging for clinical practice in multiple myeloma

Does Ethnicity Matter in Multiple Myeloma Risk Prediction in the Era of Genomics and Novel Agents? Evidence From Real-World Data

Introduction

Current risk predictors of multiple myeloma do not integrate ethnicity-specific information. However, the impact of ethnicity on disease biology cannot be overlooked. In this study, we have investigated the impact of ethnicity in multiple myeloma risk prediction. In addition, an efficient and robust artificial intelligence (AI)-enabled risk-stratification system is developed for newly diagnosed multiple myeloma (NDMM) patients that utilizes ethnicity-specific cutoffs of key prognostic parameters.

Methods

K-adaptive partitioning is used to propose new cutoffs of parameters for two different datasets—the MMIn (MM Indian dataset) dataset and the MMRF (Multiple Myeloma Research Foundation) dataset belonging to two different ethnicities. The Consensus-based Risk-Stratification System (CRSS) is designed using the Gaussian mixture model (GMM) and agglomerative clustering. CRSS is validated via Cox hazard proportional methods, Kaplan–Meier analysis, and log-rank tests on progression-free survival (PFS) and overall survival (OS). SHAP (SHapley Additive exPlanations) is utilized to establish the biological relevance of the risk prediction by CRSS.

Results

There is a significant variation in the key prognostic parameters of the two datasets belonging to two different ethnicities. CRSS demonstrates superior performance as compared with the R-ISS in terms of C-index and hazard ratios on both the MMIn and MMRF datasets. An online calculator has been built that can predict the risk stage of a multiple myeloma (MM) patient based on the values of parameters and ethnicity.

Conclusion

Our methodology discovers changes in the cutoffs with ethnicities from the established cutoffs of prognostic features. The best predictor model for both cohorts was obtained with the new ethnicity-specific cutoffs of clinical parameters. Our study also revealed the efficacy of AI in building a deployable risk prediction system for MM. In the future, it is suggested to use the CRSS risk calculator on a large dataset as the cohort size of the present study is 25% of the cohort used in the R-ISS reported in 2015.

Survival Outcomes of Newly Diagnosed Multiple Myeloma at a Tertiary Care Center in North India (IMAGe: 001A Study)

PURPOSE

The outcomes of patients with myeloma from developing countries are often lacking because of poor record maintenance. Publications from such settings are also limited because of the retrospective nature of the data collection. Information technology can bridge these gaps in developing countries with real-time data maintenance. We present the real-time survival data of the patients with myeloma from a tertiary care center in North India using one such indigenously built software.

PATIENTS AND METHODS

These are real-time data of all patients with myeloma presenting to a tertiary care center from North India. The patient characteristics (demographics, baseline disease characteristics, risk stratification, and outcomes) were recorded contemporaneously. The survival of the study population was analyzed and grouped based on various disease characteristics at diagnosis.

RESULTS

The median age of the study population (N = 696) was 65.9 (34.9-94.9) years with male predominance (65%). The median follow-up was 3.7 years (0-18.6 years) with the median overall survival (OS) not achieved. The OS of the study population at 1, 3, and 5 years was 94% (n = 558), 87.5% (n = 394), and 83.1% (n = 267), respectively. Most of the patients presented in advanced stages based on International Staging System (III:70%). On Kaplan-Meier analysis, the presence of weight loss (P = .01), renal dysfunction (P = .047), and anemia at diagnosis (P = .004) had a significant impact on survival. On Cox proportional model univariate analysis, the presence of renal dysfunction, anemia, and weight loss had the significant hazard ratio of 1.68 (1-2.82, P = .049), 3.18 (1.39-7.29, P = .0063), and 2.81 (1.22-6.42, P = .014), respectively, whereas on multivariate analysis of hypercalcemia, renal disease, anemia, and bone disease (CRAB) features, only anemia was found to have a significant hazard ratio of 2.56 (1.01-6.47, P = .046).

CONCLUSION

The real-world data show OS comparable with the published western literature. Only anemia was found to have significant impact on survival. The use of such software can aid in better data-keeping in resource-constrained settings.

AI-supported modified risk staging for multiple myeloma cancer useful in real-world scenario

•

An AI-enabled risk staging method, MRS, is developed using easy-to-acquire parameters.

•

Genomic tests cannot be performed owing to economical or geographical constraints.

•

MRS does not use cytogenetic abnormalities for risk stage prediction unlike RISS.

•

K-adaptive partitioning (KAP) used to find new thresholds for the parameters.

Introduction

: An efficient readily employable risk prognostication method is desirable for MM in settings where genomics tests cannot be performed owing to geographical/economical constraints. In this work, a new Modified Risk Staging (MRS) has been proposed for newly diagnosed Multiple Myeloma (NDMM) that exploits six easy-to-acquire clinical parameters i.e. age, albumin, β2-microglobulin (β2M), calcium, estimated glomerular filtration rate (eGFR) and hemoglobin.

Materials and Methods

: MRS was designed using a training cohort of 716 NDMM patients of our inhouse MM Indian (MMIn) cohort and validated on MMIn (n=354) cohort and MMRF (n=900) cohort. K-adaptive partitioning (KAP) was used to find new thresholds for the parameters. Risk staging rules, obtained via training a J48 classifier, were used to build MRS.

Results

: New thresholds were identified for albumin (3.6 g/dL), β2M (4.8 mg/L), calcium (11.13 mg/dL), eGFR (48.1 mL/min), and hemoglobin (12.3 g/dL) using KAP on the MMIn dataset. On the MMIn dataset, MRS outperformed ISS for OS prediction in terms of C-index, hazard ratios, and its corresponding p-values, but performs comparable in prediction of PFS. On both MMIn and MMRF datasets, MRS performed better than RISS in terms of C-index and p-values. A simple online tool was also designed to allow automated calculation of MRS based on the values of the parameters.

Discussion

: Our proposed ML-derived yet simple staging system, MRS, although does not employ genetic features, outperforms RISS as confirmed by better separability in KM survival curves and higher values of C-index on both MMIn and MMRF datasets.

Funding

: Grant: BT/MED/30/SP11006/2015 (Department of Biotechnology, Govt. of India), Grant: DST/ICPS/CPS-Individual/2018/279(G) (Department of Science and Technology, Govt. of India), UGC-Senior Research Fellowship.