The 250-cell rule does not transfer from cirrhotic to malignant ascites

Design

retrospective single-centre observational study with ROC analysis

Population

337 paracenteses in 261 patients with malignant ascites, 2020-2025

Primary outcome

diagnostic performance of ascitic fluid parameters against positive culture

Effect

250 cells/mm3: sensitivity 73.7%, specificity 72.4%; above 1318 cells/mm3: 68.4% and 93.0%

A single academic centre reviewed 337 paracenteses in 261 patients with malignant ascites between 2020 and 2025, of which 19 samples from 17 patients grew an organism.

Applied here, the familiar threshold of 250 polymorphonuclear cells/mm3 - which defines spontaneous bacterial peritonitis in cirrhosis - gave 73.7% sensitivity and 72.4% specificity. That specificity is the problem: in a population where malignancy itself raises cell counts, a quarter of uninfected patients cross the line. Higher thresholds traded differently. A polymorphonuclear percentage of 63% or more gave 68.4% sensitivity and 90.8% specificity, and a count above 1318 cells/mm3 gave 68.4% and 93.0%.

None of these is ready to use. Nineteen positive cultures is a very small number on which to fit thresholds, the study is retrospective and single-centre, and the authors call the figures hypothetical. What the study does establish is the negative: the number you carry in your head from cirrhosis was derived in a different population and does not behave the same way here. Treat the decision in malignant ascites as a clinical one, informed by the count rather than dictated by it, and send the culture.

• Do not apply the 250 cells/mm3 cirrhosis threshold unchanged to malignant ascites.
• Malignancy raises ascitic cell counts on its own, so specificity falls at that cut-off.
• Always send ascitic fluid for culture in bedside bottles, whatever the count shows.
• Treat empirically on clinical grounds in an unwell patient rather than waiting for a threshold.
• The proposed higher thresholds are hypothetical and not validated for use.

Why it matters

A number most physicians apply automatically was derived in cirrhosis, and this is a reminder that it travelled without being checked.

Don't overread it

Retrospective, single-centre, with 19 positive cultures - the alternative thresholds are not validated and should not be used.

The statistics, in plain English

Nineteen events is too few to fit a reliable cut-off: any threshold derived from them will look better in this dataset than it will anywhere else, which is why the authors label them hypothetical. The useful part of a small study like this is usually the negative finding - that an established threshold performs poorly outside the population it came from - because that does not depend on estimating a new number precisely.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.