Reading the trend in old blood tests helped spot cancer after unexplained weight loss, for some tests

Losing weight without trying is one of the symptoms that makes doctors uneasy. It can mean cancer, and it can mean a dozen other things, which leaves a general practitioner deciding who needs urgent tests. A study of 275,205 English patients asked whether an overlooked piece of information could help: not the latest blood test, but the direction the earlier ones were heading.

The answer is a qualified yes. For about half the blood tests examined, the trend over the previous year separated cancer from non-cancer better than a single abnormal result did. All of the measures, though, were modest.

Why is unexplained weight loss such a hard symptom?

MedlinePlus lists cancer among the causes of unexplained weight loss and notes that cancer is not just one disease but many, each with its own symptoms. Weight loss can point to almost any of them, or to thyroid disease, diabetes, depression, celiac disease or simply a change in circumstances.

In this cohort, 5.0% of patients presenting with unexpected weight loss were diagnosed with cancer: 13,798 people out of 275,205. That number captures the clinical problem neatly. It is far too high to ignore and far too low to investigate everybody aggressively.

What does reading a trend add?

Blood tests are usually read against a reference range: a result is normal or it is not. That throws away history. A hemoglobin level at the bottom of normal means something different in someone whose readings have been falling steadily for two years than in someone who has always sat there.

Writing in PLOS Medicine, the researchers set out to test whether interpreting those historical changes, which they call trends, could enhance discrimination for undiagnosed cancer compared with a single abnormality.

How were the blood test records analyzed?

The team used the Clinical Practice Research Datalink, a large database of anonymized English primary care records, covering patients aged 18 and over who presented with unexpected weight loss between 2000 and 2018. Cancer diagnoses came from linked national cancer registry data.

For each patient they pulled every result from 26 blood tests taken before the weight loss was recorded, then modeled the trend over the previous 1, 3, 5 and 10 years. Each approach was scored using the area under the curve, a measure of how well a test separates two groups, where 0.5 is no better than chance and 1.0 is perfect.

Did the trends beat a single abnormal result?

Sometimes. Trends over the previous year had a higher area under the curve than the equivalent single abnormality for 12 of the 26 blood tests, or 46.2%. Stretching the window reduced the advantage: 7 tests at 3 years, 6 at 5 years, and 5 at 10 years. Recent history, in other words, carries more signal than distant history.

The absolute performance is the part to keep in view. Unadjusted trends scored between 0.50, which is a coin flip, and 0.66 for cancer overall. No single blood test, read either way, comes close to a diagnostic test.

The authors’ main practical conclusion is actually about something simpler: abnormalities should be interpreted after age and sex adjustment, because the same value means different things in a 30-year-old and a 75-year-old. Trend, they write, offers further discrimination for some test-cancer combinations.

What can’t this study of blood test trends show?

It is a look back through records, not a trial of a new way of working. Nobody was assigned to have their trends reviewed, so there is no evidence yet that doing so leads to earlier diagnoses or better survival.

There is also a bias built into the data that the authors name themselves: their modeling did not account for who gets tested in the first place. People with repeat blood tests are, by definition, people whose doctors were already concerned about something. The trend analysis was also restricted to patients with at least two tests in the window.

What does this mean if you have lost weight without trying?

The practical advice does not change: see a doctor, and the sooner the better. What may change, if this line of work continues, is what happens on the other side of the desk. Most practices now hold years of results for each patient, and this study suggests some of that history is being left unread.

For patients, that is an argument for continuity: results collected in one place, over time, are worth more than the same tests scattered across providers. And unexplained weight loss is worth mentioning explicitly, rather than waiting for it to come up.