Routine Patient Data Reveals Hidden Patterns of Cancer Cachexia in Landmark Study

Cancer cachexia has long stood as one of oncology’s most stubborn and least understood complications. This severe, uncontrollable wasting syndrome affects roughly half of all cancer patients, raiding the body’s own muscle tissue and fat for energy in a process that cannot be reversed simply by eating more. It makes treatments harder to tolerate, erodes quality of life, and is associated with worse outcomes. Yet despite its prevalence, there are currently no U.S. Food and Drug Administration-approved therapies to treat it. Now, a team of researchers from Memorial Sloan Kettering Cancer Center (MSK) and their collaborators has delivered the largest human study of the condition to date, mining routine clinical data collected during the treatment of nearly 60,000 cancer patients to reveal patterns that had remained invisible for decades.

The sheer scale of the dataset is what sets the work apart. The researchers analyzed more than 2 million body weight measurements and millions of blood test results, combined with detailed treatment histories and the genetic profiles of patients’ tumors. The findings, published October 12 in Nature Medicine, show that cachexia occurs across nearly all forms of cancer, not just those associated with eating and digestion; that it unfolds in distinct episodes that can potentially recur, many of which are linked to disease progression; that it leaves a consistent signature in routine blood tests; and that it is linked to specific tumor mutations. Together, these discoveries could help predict which patients are most at risk and enable timely support and interventions. Senior author Ed Reznik, PhD, whose lab applies data science to the study of cancer metabolism, noted that much of the information was routine clinical data collected to understand how patients were doing at different points during treatment. It turned out, he said, that the data had a lot to tell us.

The study was co-led by two students from the Reznik Lab: Venise Jan Castillon, a member of MSK’s Bridge program, which helps trainees transition into biomedical graduate programs, and Amy X. Xie, PhD, a recent graduate of the Biochemistry, Cell Biology, Molecular Biology Allied Program at Weill Cornell Medicine, where she focused on bioinformatics. Castillon, who studied physics and worked on solar panels before joining MSK as a step toward a career in biomedicine, explained why the team calls their work an atlas. Because the study spanned tens of thousands of patients and more than 650 cancer types, the researchers were able to look at broader patterns in cachexia that hold true across different cancer types and disease stages, filling critical information gaps left by earlier studies that were relatively small and often focused on a single cancer type.

That gap exists in part because much of the foundational science on cachexia has come from laboratory models, which are useful for understanding basic biology but limited in what they can reveal about how the condition unfolds in human patients over months and years. Studying cachexia in the clinic is also surprisingly difficult. The clinical definition is a loss of at least 5 percent of body weight over six months, which can confirm that someone is losing weight but not what kind of tissue is disappearing or what mechanism is driving the loss. Reznik emphasized the challenge of disentangling causes: the body’s energy needs can shift in response to changes caused by a tumor or cancer treatment, patients with stomach or esophageal cancer may have physical difficulty eating or digesting food, and cancer treatments themselves can suppress appetite. The hard question, he said, is how much of a person’s weight loss is really driven by something intrinsic to the cancer itself.

Reznik was inspired to pursue the project after attending a seminar at a cancer conference a few years ago. The speaker, Tobias Janowitz, MD, PhD, of Cold Spring Harbor Laboratory, opened his presentation with a photo of a patient who turned out to be his own mother, who died of cancer-related cachexia. Reznik recalled that roughly 1,000 people sat in silent attention, and that it dawned on him that investigating cachexia would be a good fit for his lab’s expertise in cancer metabolism. Janowitz, a coauthor of the MSK-led study, has argued that patients should be seen as more than passive recipients of new treatments, writing in a recent essay in the journal Neuron on neuroscience and cachexia that, in his mother’s words, whatever you do, make sure to listen to your patients. In a somewhat abstracted way, that is exactly what the Reznik Lab did, tuning in for the first time to what all those weight checks and blood tests had to say about one of the most challenging problems for people with cancer.

The technical approach centered on data already sitting in MSK’s medical records, subjected to rigorous statistical analysis. The work was enabled in part by MSK’s Cancer Data Science Initiative, which supports translational research by using artificial intelligence and other tools to remove identifying details and make anonymized clinical data available to clinicians, biologists, and computational scientists across the institution. The team began by searching each patient’s weight history for any six-month window that met the 5 percent weight loss criterion for cachexia. For a typical patient, the records contained about 40 weight measurements over a five-year period, a rich source for analysis. This granularity allowed the researchers to see that cachexia was not always just a long, steady decline, as Xie noted, but something far more dynamic.

One of the study’s most striking findings is that cachexia often occurs in distinct episodes that can recur multiple times over the course of a patient’s illness, something that had not been widely appreciated before. The data show that cachexia episodes are somewhat more likely to occur when a patient’s cancer is actively growing or when a treatment starts to fail, moments doctors call disease progression. But that does not tell the whole story: many episodes happen independent of progression, suggesting additional factors are at play. In colorectal cancer, for example, nearly a third of all disease progression events occurred while a patient was also experiencing cachexia, more than would be expected by chance, and similar patterns appeared in lung and pancreatic cancers. Overall, cachexia often emerges toward the end of life and hastens a patient’s final decline, but in patients with pancreatic and gastric cancers it frequently happens earlier and recurs. This variability suggests that new strategies against cachexia will need to be tailored to each patient’s disease and circumstances.

A second critical finding is that cachexia leaves a remarkably consistent signature in routine blood work across nearly all the cancer types studied. Wasting episodes were accompanied by lower albumin, low hemoglobin and red blood cell count indicating anemia, elevated liver enzymes, and higher white blood cell and neutrophil counts, which are markers of inflammation and nutritional depletion. Reznik observed that because cachexia affects the whole body, it makes sense that a variety of markers appear in patients’ blood, and that across almost all cancer types the team saw this same signature associated with cachectic weight loss. Notably, even though albumin, a vital protein produced by the liver, decreased during episodes of cachexia, it often remained within a normal range. Xie pointed out that these declines are real but may be subtle enough not to raise a red flag unless clinicians are actively looking for them, a finding with direct implications for earlier detection using tests already being performed.

The researchers also found that specific tumor mutations are significantly associated with the risk of developing cachexia, some increasing that risk while a few others appear to reduce it. Mutations in TP53, one of the most commonly altered genes in human cancer, were linked to higher cachexia risk across nine different cancer types, including lung, breast, and prostate cancers. In lung cancer, mutations in STK11 were also independently associated with elevated risk. In colorectal cancer, mutations in SMAD4 and KRAS were tied to higher risk, while mutations in ARID1A and KMT2B appeared protective. Crucially, these associations held even after accounting for disease stage and overall health, pointing to something in the tumor’s biology itself rather than simply how far the cancer had advanced. Reznik noted that scientists have not fully agreed on what actually causes cachexia, whether it is something the tumor releases into the bloodstream or a behavioral loss of energy and appetite that comes with being seriously ill, and that the genetic findings suggest tumor biology plays a significant role.

Building on these findings, the team developed predictive models for two common cancer types, lung adenocarcinoma and colorectal carcinoma, combining tumor genetic information, disease stage, ability to maintain daily activities, and baseline body weight to generate individualized risk scores for developing cachexia over one, two, and three years. The models were moderately accurate, which the researchers believe represents a meaningful step toward identifying high-risk patients early enough to intervene with additional nutritional and rehabilitation supports. The continuing work suggests there may be distinct subtypes of cachexia, and further research could identify which patients are at greatest risk and which therapies would be most effective for individual patients. Reznik described the ultimate goal as a precision trial for cachexia, one that he does not think is far off. The research was supported by the Society of MSK, a Pershing Square Sohn Cancer Prize, and Cancer Grand Challenges through the international team CANCAN, established in 2020 to tackle the cachexia challenge, along with funding from the National Cancer Institute, the National Institutes of Health, and the Department of Defense.

Subject of Research: A large-scale clinical data analysis of cancer cachexia episodes, blood signatures, and tumor mutation associations

Article Title: Insights about cachexia, one of cancer’s most common complications, hiding in routine patient data

Article References: Insights about cachexia, one of cancer’s most common complications, hiding in routine patient data. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: cancer cachexia, Memorial Sloan Kettering, Nature Medicine, tumor genetics, TP53, weight loss, blood biomarkers, predictive modeling, clinical data science, disease progression, oncology, precision medicine