Dr Yu discusses the ARCAD Nomogram in a Real-World CRC Population

In Partnership With:

Partner | Cancer Centers | <b>Moffitt Cancer Center</b>

James Yu, MD, discusses the validation of the ARCAD nomogram in the real-world setting for patients with stage IV colorectal cancer.

James Yu, MD, hematology/oncology fellow, Moffitt Cancer Center, discusses the external validation of the ARCAD nomogram in the real-world setting for patients with stage IV colorectal cancer (CRC) and highlights the outcomes derived from the evaluation.

The ARCAD CRC nomogram, developed for predicting 1-year survival in stage IV CRC, was validated using a large, real-world cohort of patients with CRC from the Flatiron database. The observed underestimation of survival in the Flatiron real-world cohort suggests advancements in CRC treatments, such as targeted therapy or immunotherapy. Notably, the ARCAD nomogram proves promising for clinicians when predicting 1-year overall survival in real-world scenarios.

In 2018, Australian investigators developed the ARCAD nomogram, specifically designed to predict the 1-year survival rate in stage IV CRC, Yu begins. This nomogram incorporates additional prognostic factors, such as age, ECOG performance status (PS), and mutational status. However, a notable limitation is that its construction is based on a clinical trial population. Clinical trials are known for their high selectivity, often comprising patients with better PS and fewer comorbidities than real-world populations, he explains. In collaboration with Moffitt Cancer Center, investigators conducted an external validation of the nomogram, Yu notes.

The study revealed 3 crucial findings. Firstly, the ARCAD nomogram demonstrated fair discrimination in predicting the 1-year survival rate in a real-world population, with an area under the ROC curve of 0.74, Yu expands. Secondly, when predicted survival rates were graphed against observed survival rates, the slope of the graph indicated effective stratification of groups for mortality, he explains. Lastly, the nomogram exhibited an underestimation of the survival rate, Yu notes. For example, in the cohort's overall population, the model predicted a 1-year survival rate of approximately 64% (95% CI, 63.7%-64.4%), whereas the actual survival rate was 72.7% (95% CI, 71.8%-73.6%), indicating a 9%underestimation, he continues. These findings are pivotal for understanding the nomogram's performance in diverse patient populations outside the confines of clinical trial enrollment criteria, Yu concludes.