A New Prognostic Scoring Model in Myelodysplastic Neoplasms

Objectives:

Myelodysplastic neoplasms (MDS) exhibit variable clinical courses including the risk of progression to acute myeloid leukemia (AML), highlighting the need for precise prognostic score system to guide stratified therapy. Until 2022, the IPSS-M prognostic score system was introduced by the International Prognostic Working Group for MDS. However, the IPSS-M requires more extensive sequencing and may encounter challenges in its generalization within China. In this study, we conducted a comprehensive analysis of MDS data from a cohort of Chinese patients, leading to the development of an innovative prognostic stratification model that incorporates mutation information, named the Gene-Incorporated Prognostic Scoring System for MDS (GIPSS-R).

Methods

We conducted a retrospective cohort study involving 111 primary MDS patients to identify prognostic factors, including gene mutations. Our analysis employed both univariate and multivariate Cox regression models to pinpoint significant predictors. The GIPSS-R model was developed through Stepwise Cox regression multivariate analysis. To evaluate the predictive accuracy of the new model, we employed decision analysis curves and calibration curves. Validation was performed using two independent cohorts: an internal validation cohort of 49 primary MDS patients from Tongji Hospital and an external validation cohort of 141 patients from the dataset GSE129828. We compared the accuracy of the GIPSS-R model with the IPSS-R model using the Concordance Index (C-index) and Area Under the ROC Curve (AUC).

Results

Analysis of the Chinese cohort (training and validation) versus the Western validation cohort revealed a median age of 67 years for Chinese patients and 72 years for Western patients, suggesting an earlier disease onset in the Chinese population (P<0.01). In peripheral blood, Chinese individuals exhibited lower levels of hemoglobin, platelets, leukocytes, and neutrophils compared to Western individuals (P<0.001). Karyotype abnormalities (P<0.001), particularly complex karyotypes (P=0.021) were more frequent in Chinese individuals.

We identified several independent prognostic factors significantly associated with overall survival: the number of mutated genes, specific mutations (TP53, EZH2, U2AF1, RUNX1, ASXL1, SRSF2), IPSS-R score, and higher age. We conducted a stepwise multivariate Cox regression analysis, which led to the development of a new prognostic scoring model called GIPSS-R. This model stratifies patients into distinct risk groups: very low risk (≤1.0), low risk (>1.0-1.5), intermediate risk (>1.5-2.5), high risk (>2.5-3.5), and very high risk (>3.5). The corresponding median survival times for these groups are >60, 37.93, 27.90, 17.13, and 4.27 months, respectively (P=7.5154E-16). Calibration and decision curves demonstrated excellent goodness of fit and clinical predictive accuracy. Additionally, we constructed a nomogram based on these independent prognostic factors, offering a simple and clear calculation of survival probability for patients. Furthermore, the new scoring model underwent validation in both an independent internal validation cohort and an external validation cohort. The median survival, Kaplan-Meier curve, and area under the ROC curve were all superior to those of the conventional model. The C-indexes of the new model in the three cohorts (0.777, 0.706, and 0.644) exceeded those of IPSS-R (0.697, 0.638, and 0.604), respectively.

We also employed the training cohort to assess the effectiveness of the IPSS-M model. When patients in this cohort were classified according to the IPSS-M for risk assessment, the survival curves exhibited recurrent overlaps among the distinct risk groups. Notably, the median survival time for the very-low-risk group was shorter than that of both the low-risk and intermediate-risk groups. Furthermore, the survival of the moderate-low-risk group was inferior to that of the moderate-high-risk group. Consequently, the IPSS-M may not be as effective as the IPSS-R in prognostic stratification for the Chinese population.

Conclusions

We have developed a new prognostic scoring model (GIPSS-R) based on gene mutation variables, which can assist clinicians in more effectively guiding the prognosis and treatment of MDS patients.

No relevant conflicts of interest to declare.