MoCaPS: A Machine Learning Model for Stratification of Cancer-Associated Cachexia Based on Blood Biomarkers

Cancer-associated cachexia (CC) is a multifactorial syndrome observed in up to 80% of PDAC patients, characterized by unintentional weight loss, muscle wasting in the presence or absence of fat loss, and fatigue [1], which can lead to a reduction in quality of life and poor clinical outcomes. [2, 3].

To distinguish between cachexia stages, we followed the Florida Pancreas Collaborative and used criteria described in [4], which are based on the Vigano classification [5]. This classification system is comprised of the following types of data: (a) biochemistry (level of C-reactive protein (CRP) or albumin, or hemoglobin, or white blood cell count), (b) changes in food intake, (c) minimal or significant weight loss (WL), and (d) changes in daily activities based on the Patient-Generated Subjective Global Assessment (PG-SGA) performance status [6]. The recognized 4 cancer cachexia stages are: noncachexia (NCa), precachexia (PCa)—an early stage of the syndrome characterized by abnormal food intake or blood chemistry but no significant weight loss, cachexia (Ca), and refractory cachexia (RCa)—a stage that is largely irreversible [7].

We provide three tools to distinguish between NCa and Ca stages, PCa vs. Ca stages, and PCa vs. NCa stages.

These classifiers need the following libraries

numpy
sklearn
matplotlib
seaborn
pandas
tensorflow
statsmodels
scipy

Authors

Kayode Olumoyin kayode.olumoyin@moffitt.org, Magaret Park, Evan W. Davis, Jennifer B. Permuth, Katarzyna Rejniak

Source Code

https://github.com/okayode/MoCaPS

License

This project is licensed under the GNU General Public License v3.0.

References

1. Baracos, V.E., et al., Cancer-associated cachexia. Nat Rev Dis Primers, 2018. 4: p. 17105.
2. Sun, L., X.Q. Quan, and S. Yu, An Epidemiological Survey of Cachexia in Advanced Cancer Patients and Analysis on Its Diagnostic and Treatment Status. Nutr Cancer, 2015. 67(7): p. 1056-62.
3. Yu, Y.C., et al., Review of the endocrine organ-like tumor hypothesis of cancer cachexia in pancreatic ductal adenocarcinoma. Front Oncol, 2022. 12: p. 1057930.
4. Permuth, J.B., et al., Leveraging real-world data to predict cancer cachexia stage, quality of life, and survival in a racially and ethnically diverse multi-institutional cohort of treatment-naive patients with pancreatic ductal adenocarcinoma. Front Oncol, 2024. 14: p. 1362244.
5. Vigano, A.A.L., et al., Use of routinely available clinical, nutritional, and functional criteria to classify cachexia in advanced cancer patients. Clin Nutr, 2017. 36(5): p. 1378-1390.
6. Gabrielson, D.K., et al., Use of an abridged scored Patient-Generated Subjective Global Assessment (abPG-SGA) as a nutritional screening tool for cancer patients in an outpatient setting. Nutr Cancer, 2013. 65(2): p. 234-9.
7. Yue, M., et al., Understanding cachexia and its impact on lung cancer and beyond. Chin Med J Pulm Crit Care Med, 2024. 2(2): p. 95-105.