Multiple Myeloma Cancer
Bladder models
- Kystis (Brown) Brown
- COBRAS (Ottawa) Ottawa
- SCOUT (NYU) NYU
Bladder Model Comparison Grid (PDF, 145 KB)
See all Comparison Grids & Profiles (Includes historical versions)
Lung models
- BCCRI-LunCan (BCCRI)
- BCCRI-Smoking (BCCRI)
- LCOS (Stanford)
- LCPM (MGH)
- MISCAN-Lung (Erasmus)
- SimSmoke (Georgetown)
- Smoking-Lung Cancer (Georgetown)
- MULU (Mount Sinai)
- ENGAGE (MDACC)
- YLCM (Yale)
- OncoSim-Lung (CPAC-StatCan)
- LMO (FHCC) (Historical)
Lung Model Comparison Grid (PDF, 161 KB)
See all Comparison Grids & Profiles (Includes historical versions)
Background
The Cancer Intervention and Surveillance Modeling Network (CISNET) Multiple Myeloma Incubator Program has built, calibrated and validated evidence-based multiple myeloma (MM) model across the MM care continuum. Three natural history models have been developed independently by investigators at two institutions, Washington University and Yale University:
- Washington University Natural History of Multiple Myeloma Model—Discrete Event Simulation (WUMM-DES)
- Washington University Natural History of Multiple Myeloma Model—Compartmental Model (WUMM-CM)1
- Yale University Natural History of Multiple Myeloma Model (YUMM)
Common structures and inputs
All three models share the same disease stage progression and risk factors. The current versions model disease stages from no disease (healthy) to a premalignant stage, monoclonal gammopathy of undetermined significance (MGUS), followed by MM and death (see Figure 1). Smoldering multiple myeloma (sMM), as a more advanced disease stage than MGUS is not currently included in the models, due to its asymptomatic nature, no administrative codes, and insufficient data for this disease stage. However, more data are being collected to build the evidence to add this disease stage to the models.
The current models include age, sex, and race as risk factors, which determine disease progression. Obesity, as another established risk factor, will be added using an obesity generator to the next version of the three models.
Figure 1. Model Scheme

To ensure that our models are evidence-based and generalizable to the studied subpopulations, we populated the models using data from published studies and estimates derived via data from several databases representing different U.S. populations. These included:
- Three national databases that are nationally representative samples of the U.S. general population: the National Health and Nutrition Examination Study (NHANES), 1971-, the National Health Interview Survey (NHIS), 1963-, and its linked mortality data files (with mortality follow-up until end of 2019, the Medical Expenditure Panel Survey (MEPS), 1996-.
- Surveillance, Epidemiology, and End Results (SEER) and related software, 1975-: When combined with the use of the Complete Prevalence (ComPrev) and Projected Prevalence (ProjPrev) Software, we will be able to obtain annual prevalence of MM based on the limited-duration prevalence obtained from SEER*Stat.
- Veteran Health Administration (VHA), 1998-: VHA contains electronic medical records for veterans utilizing the VA healthcare system in the entire nation and are linked to Medicare and Medicaid.
- Centers for Medicare & Medicaid Services Chronic Condition Warehouse (CCW), 1999-: CCW consists of the longitudinal 5% random sample of Medicare beneficiaries, including the Inpatient, Carrier Claims, Outpatient, Beneficiary Summary, Home Health Agencies, Skilled Nursing Facility, Durable Medical Equipment, and Part D Drug Event files.
- SEER-Medicare linked database (SEER-Medicare), 2000-: SEER-Medicare combines clinical information from population-based cancer registries with claims information from the Medicare program.
- Truven Health Analytics MarketScan Commercial Claims and Encounters (MarketScan), 2006-: The MarketScan is a claims database, which contains >160 million privately insured individuals and their medical and pharmacy claims.
- Blue Cross Blue Shield Axis, the largest source of commercial insurance claims in the US.
- Flatiron Health’s nationwide electronic health record oncology database, which includes >265 cancer clinics across the United States, representing more than 2 million patients.
Model calibration and validation were conducted by comparing model outputs with national data. For MGUS prevalence, data from the NHANES III and the continuous NHANES, 1999-2004, were used, because serum samples for participants aged ≥50 years screened for MGUS by immunofixation electrophoresis were only available in the NHANES III and the continuous NHANES, 1999-2004.2 For MM incidence, the SEER data were used (see details in the model profile of each of the three models).
Impact
The three new CISNET population-based models of MM will be the first to conduct comparative modeling analyses designed to address the critical knowledge gaps in MM prevention and control. All three models are evidence-based using both published evidence and real-world data to inform the models. They are all calibrated and validated against the real-world data lending credibility. All three models are designed to allow for evaluations of novel prevention strategies or new treatment regimens for premalignant stages (MGUS or sMM) and MM to guide public health endeavors and cancer control and prevention. The natural history of MM modeling and intervention strategies are flexible and can be extended to incorporate and evaluate other novel interventions throughout various disease stages of MM. Further, the models can be adapted to study this disease in different populations and can be used to guide the prioritization of public health endeavors and cancer control strategies under limited resources and budget constraints. These models allow for assessment of current clinical guidelines to promote value-based therapies, advising guideline recommended therapies. In addition, the MM natural history modeling will be extended to evaluate guideline recommended therapies and impact of policy changes or new policies, such as the Inflation Reduction Act.
References
- Huber JH, Ji M, Shih YH, Wang M, Colditz G, Chang SH. Disentangling age, gender, and racial/ethnic disparities in multiple myeloma burden: a modeling study. Nat Commun. 2023;14(1):5768. Epub 20230920. doi: 10.1038/s41467-023-41223-8. PubMed PMID: 37730703; PubMed Central PMCID: PMC10511740.
- Ji M, Huber JH, Schoen MW, Sanfilippo KM, Colditz GA, Wang SY, Chang SH. Mortality in the US Populations With Monoclonal Gammopathy of Undetermined Significance. JAMA Oncol. 2023;9(9):1293-5. doi: 10.1001/jamaoncol.2023.2278. PubMed PMID: 37498610; PubMed Central PMCID: PMC10375386.