Finding Flaws in Using Current Clinical Markers to Define HR SMM
By Sarah LaCorte
ORLANDO— Results of an independent sequential patient cohort that indicated current clinical variables used to determine high risk (HR) in smoldering multiple myeloma (SMM) are not reliable were presented at the 2019 ASH Annual Meeting (Abstract 1794).
The study aimed to define concordance among patients defined as HR SMM by two commonly used classification systems developed to identify patients with a higher rate of progression: the 2008 Mayo Clinic model and the PETHEMA (Programa de Estudio y Tratamiento de las Hemopatias Malignas) models.
“We have noted for years in our clinic, that the current clinical scores developed to assign smoldering myeloma patients a risk score have significant limitations; these scores are all indirect measures of disease burden. When applying different proposed scores in parallel—on any given single patient—consistently result in conflicting results. Almost every patient is high risk with one risk score and low risk with another score. This motivated us to conduct a large study designed to quantify the degree of discordance versus concordance,” said study author Ola Landgren, MD, PhD, Chief of the Myeloma Service in the Department of Medicine at Memorial Sloan Kettering Cancer Center.
To define the concordance, researchers reviewed the medical records of patients sequentially assigned a diagnosis of SMM by the myeloma program at the NIH Clinical Center between April 2010 to July 2019, excluding patients’ myeloma-defining events.
Using the 2008 Mayo Clinic model, the 2018 Mayo Clinic model, and the PETHEMA model, each patient was assigned a risk score. The distribution of patients in the low-risk, intermediate (IR), and HR groups were compared between the models, and concordance ratios were calculated between the three models.
A total of 236 patient records were reviewed and 136 patients identified as having SMM were stratified by risk based on all three models. According to the study, “the rate of concordance between the 2008 Mayo Clinic model and the PETHEMA model was 31.6 percent (95% CI: 24.4-39.8%), similar to previously published results. The concordance between the 2018 Mayo Clinic model and the PETHEMA model was slightly higher at 44.8 percent (95% CI: 36.7-53.2%; P=0.0337). There was significant discordance between the models in classifying patients as HR versus non-HR. However, the 2018 Mayo Clinic model had a higher concordance with the PETHEMA model (27.2%; 95% CI: 20.4-35.3%) than the 2008 Mayo Clinic model (4.4%; 95% CI:1.8-9.5%).”
In effect, the study demonstrated that no one model has been found to be superior than the other, and that some patients may be defined as high risk by one model, and low risk by another.
These results impact treatment of HR SMM as it is currently being investigated in multiple clinical trials. The authors point out that “as the results from these trials are published, the data will need to be scrutinized as to how patients were defined as high risk in order to compare results.”
“The logical step forward is the development of genomic signatures to better predict risk,” said Landgren. “Current models are outdated and based on old technologies. We need move from tumor burden to cancer genomics to better define risk.”