Cancer-Related Weight Loss Tied to Disease Progression

Episodes of cachexia-related weight loss affected 46.63% of more than 52,000 adults with cancer treated at Memorial Sloan Kettering Cancer Center (MSK) in New York, according to a study published in Nature Medicine. Cachexia is a wasting syndrome marked by unintended weight loss and muscle loss. Patients who had a weight loss episode had shorter survival. The episodes also tended to occur around the time their cancer grew or spread.

In the observational study, investigators analyzed health records from 52,236 patients whose tumors underwent genetic testing through MSK-IMPACT between 2010 and 2024. Several findings were then confirmed in 5,785 patients at Dana-Farber Cancer Institute (DFCI) in Boston.

How common was cancer cachexia-related weight loss in the MSK study?

Investigators identified 30,938 separate weight loss episodes. A weight loss episode was defined as losing 5% or more of body weight over six months, or 2% or more for patients who were already underweight. Among patients who had an episode, 22.56% had more than one. Patients spent an average of 13.7% of their follow-up time in an episode. The average weight loss during an episode was 10.45%.

Rates varied by cancer type. Gastrointestinal cancers were affected most often, including pancreatic cancer (72.7%), stomach cancer (71.7%) and colorectal cancer (49.6%). Rates were lower in prostate cancer (26.1%) and chronic lymphocytic leukemia/small lymphocytic lymphoma (19.9%). At DFCI, 55.2% of patients met the weight loss definition at least once.

Within 180 days of diagnosis, 23.23% of patients with stage 1 to 3 cancer who had not yet started treatment had an episode. The highest early rates were in pancreatic cancer (59.6%) and stomach cancer (57.6%). Some weight returned afterward. Among all episodes, 26.7% were followed by a gain of at least 5% in body mass index (BMI, a measure of weight relative to height).

Was cancer-related weight loss linked to survival and disease progression?

Across all cancer types combined, patients who ever had an episode had 3.02 times the risk of death of patients who never had one. Higher risk of death appeared in 17 of 18 cancer types studied. Glioblastoma, a type of brain cancer, was the exception. When the timing of each episode was taken into account, the start of an episode was tied to worse survival in all 18 cancer types. The strongest link was in prostate cancer.

Weight loss episodes also clustered around disease progression, meaning scans showed the cancer had grown or spread. In 17 of 18 cancer types, scans showing progression fell within a weight loss episode more often than expected by chance. Among 9,088 patients with weight data near the end of life, 53.85% died during an episode or within 30 days after one.

“Cachectic episodes clustered near periods of disease progression and were associated with [worse] outcomes in nearly all cancer types,” the researchers wrote in the study, “A health-system-scale, episode-resolved multimodal atlas of cancer cachexia.”

What blood test changes accompanied cancer cachexia in the study?

During episodes, blood tests showed lower albumin (a protein linked to nutrition) and lower hemoglobin (the oxygen-carrying protein in red blood cells). Both changes point to poorer nutrition and anemia. Markers of inflammation and organ stress were higher. Examples included neutrophils (a type of white blood cell), alkaline phosphatase (an enzyme tied to liver and bone stress) and C-reactive protein (a marker of inflammation).

Most values stayed within normal ranges. During episodes, 28.4% of albumin measurements fell below the normal range, compared with 12.0% of measurements outside episodes. Several of the blood markers could also reflect the effects of chemotherapy rather than cachexia alone, the researchers noted.

Which tumor mutations were linked to cancer cachexia risk?

Investigators found 34 links between 23 tumor gene mutations and weight loss risk across 14 cancer types. Mutations in the TP53 gene were tied to higher risk in nine cancer types, including breast cancer, prostate cancer and lung adenocarcinoma, a type of lung cancer.

Investigators then accounted for age, stage, starting weight and performance status (a measure of daily functioning). After those adjustments, patients with lung adenocarcinoma and a TP53 mutation had a 28% higher risk of a weight loss episode. Patients with an STK11 mutation had a 44% higher risk. In the DFCI group, TP53 mutations were tied to a 30% higher risk in patients with non-small cell lung cancer.

Some mutations were linked to lower risk. In the initial analysis, patients with colorectal cancer and a KMT2B mutation had a 66% lower risk.

Can cancer cachexia risk be predicted from medical records?

Investigators built models to estimate weight loss risk for patients with lung adenocarcinoma, prostate cancer and colorectal cancer. The models used stage, daily functioning, starting weight and tumor mutations. Accuracy was measured with a concordance index (C-index), where 0.5 equals a coin flip and 1.0 equals perfect prediction. Scores were 0.648 for lung cancer, 0.670 for prostate cancer and 0.639 for colorectal cancer. A reduced lung cancer model scored 0.625 when tested in patients at DFCI.

“This model, although moderate in predictive power, highlights that risk can be quantified early and suggests a path toward incorporating cachexia surveillance into oncology practice,” the researchers said.

What are the limitations of the MSK cancer cachexia study?

Body weight alone cannot show whether the loss came from muscle or fat. The full consensus definition of cachexia includes measures of muscle loss. Investigators could not apply those measures because imaging scans were not available for the entire group. In patients with obesity, muscle loss can occur while body weight stays the same.

Cancer treatment, surgery and nutrition support may influence weight loss. Nutrition counseling, supplements and exercise programs were not consistently recorded. All patients in the main analysis were treated at a single cancer center, which may limit how well the results apply to community hospitals.

“Finally, as with all observational analyses, our findings establish associations rather than causation,” the researchers said.

References

1. Castillon VJ, et al. “A Health-System-Scale, Episode-Resolved Multimodal Atlas of Cancer Cachexia.” Nature Medicine. Published October 12, 2026.

For more news on cancer updates, research and education, don't forget to subscribe to CURE®'s newsletters here.