In cancer patients, a popular “biological age” blood score may mostly measure the cancer

Biological age scores promise to say how old the body really is. One of the most widely used is PhenoAge. It doesn't need DNA tests: it is calculated from ordinary blood tests that hospitals run all the time, such as albumin (a blood protein that drops with poor nutrition and illness), the share of lymphocytes (a type of white blood cell), red-cell size variation, creatinine (a kidney marker) and glucose. Each test gets a weight, chosen so that the total best predicted death in a large US health survey, and the result is expressed in years. PhenoAge acceleration is how many years someone's PhenoAge is above or below their actual age. This blog has covered studies that used it, for example on vaping.

Earlier studies found that cancer patients with a higher PhenoAge do worse, and read this as a sign that faster aging shapes how cancer goes. A team at Memorial Sloan Kettering Cancer Center in New York asked a simpler question: in someone who has a tumour, is the score measuring aging, or the illness itself?

The study is a preprint: it has not yet been peer-reviewed.

What they did

The team went back through the records of 3,067 patients who had surgery to remove a tumour with the aim of curing it, and who had not received chemotherapy or other cancer drugs beforehand:

• 577 with colon cancer, 1,724 with kidney cancer, 241 with a common type of lung cancer, and 525 with triple-negative breast cancer.
• PhenoAge was calculated from a blood panel taken in the year before surgery, typically about a week before. Because one of its usual tests (C-reactive protein, an inflammation marker) is rarely measured before surgery, they used a version with the eight remaining tests.
• They then checked how PhenoAge related to tumour stage and to survival, and compared it with each of its eight blood tests on their own and with the predictors doctors already use, such as age and stage.

What they found

• The score did predict survival. In all four cancers, a higher PhenoAge acceleration went with shorter survival: for each step up of one standard deviation, the risk of death was 1.66 times higher in colon cancer, 1.57 in kidney cancer, 1.28 in lung cancer and 1.17 in breast cancer.
• But it rose with tumour size. Patients whose tumours were more advanced at surgery had higher PhenoAge acceleration in colon, kidney and lung cancer. In breast cancer almost all tumours were small, so there was little to compare. Taking tumour stage into account shrank the link with survival a little.
• It added little to what doctors already know. On top of the usual clinical predictors, the score improved survival prediction only slightly, and clearly so only in kidney cancer.
• One or two plain blood tests did as well. In colon cancer, albumin alone predicted survival about as well as the full score; in kidney cancer, lymphocyte share nearly did. Each cancer had a pair of tests, different from cancer to cancer, that came within 0.01 of PhenoAge on the measure of prediction used.
• The weights didn't matter. When the team threw away PhenoAge's carefully fitted weights and simply added up the eight standardized tests with equal weight, the result predicted survival at least as well as PhenoAge in every cancer.

These blood tests are things cancer doctors already read as signs of poor nutrition, a weakened immune system, stress on the bone marrow or kidney trouble. The authors conclude that in cancer, PhenoAge seems to pick up the body-wide effects of the disease rather than aging as such.

What to keep in mind

• Not yet peer-reviewed, and the results have not been checked in patients from another hospital.
• The authors' own limits: the study looked back at existing records, at one centre, and only included surgical patients with a complete blood panel. It used a modified, eight-test PhenoAge (though the usual nine-test version gave similar results). Tumour stage and lab results are imperfect measures of how much disease someone has, and there was no separate measure of biological aging.
• It doesn't say PhenoAge is useless. The team writes that their findings can't show that PhenoAge holds no aging signal in cancer patients, only that much of what it predicts here is explained by cancer-related lab changes. They tested only one clock; clocks based on DNA methylation were not studied, and the authors stress that different clocks measure different things.
• Observational. This shows associations between blood values and survival, not causes.

Why it matters

Biological age scores are increasingly used in research to argue that aging drives disease. This study is a reminder that a score built from routine blood tests will also move when someone is ill, for reasons that have nothing to do with aging. Before reading a high "biological age" in a sick person as faster aging, researchers need to check whether the score's parts are simply tracking the illness. The team argues that studying aging in cancer will need measures that can be told apart from the effects of the cancer and its treatment.