The Antibiotic That Wins the Lab Test Doesn't Clear the Infection

Deciding which antibiotic to give for a urinary tract infection relies on a susceptibility test: grow the patient's bacteria in a standard broth, add the drug, see what stops it. The test is fast, standardised and worldwide. It is also, in this disease, a poor guide to what actually happens. Around 30 percent of women who get a UTI have another within the year, often despite having received a drug the laboratory called effective.

A team spanning UCL, Oxford and Leicester rebuilt the test to resemble a bladder, and watched the standard answer come apart in stages.

Why it matters: Resistance is usually framed as bacteria acquiring defences. Here the drug works perfectly well against the bacteria and still fails, because of where the bacteria are and what is flowing past them.

Three steps away from the broth

The first step was the fluid. Testing in pooled human urine rather than laboratory broth left most antibiotics unchanged, though it pushed three isolates over the resistance threshold for one drug. Nitrofurantoin, the front-line choice, still looked the strongest.

The second step was to watch over time rather than score a single endpoint. A standard test asks whether growth is stopped at a fixed concentration. Killing curves ask how fast, and by that measure nitrofurantoin was already weaker in urine, in a strain the endpoint test had scored as fully sensitive.

The third step was tissue: a three-dimensional model of human bladder lining, in which the bacteria can do what they do in people. They form clusters on the surface, and they get inside the cells, forming intracellular communities that are a well-documented feature of real infections.

Here nitrofurantoin reduced the infection substantially and cleared none of it, at a concentration matching normal dosing. Surface bacteria fell; the intracellular communities persisted. The drug is not failing to kill bacteria. It is failing to reach some of them.

Adding phages

Bacteriophages, viruses that infect bacteria, are the main alternative under development for resistant infections. A cocktail was tested alongside the antibiotic.

Phages were weaker in urine than in rich media, which is itself worth knowing and matches earlier reports. In the tissue model they did something the antibiotic could not: they suppressed the intracellular communities and kept the cells healthier.

Combining the two eliminated free-swimming bacteria entirely, which neither did alone. But it did not clear more of the tissue-associated population than the antibiotic managed by itself. As Garcia Maset and colleagues write in Nature Communications, the therapeutic benefit of the combination may derive from complementary targeting of extracellular and intracellular niches, rather than from increased overall bactericidal activity. Two treatments covering different ground, not one stronger treatment.

The phages also provoked a response. They raised inflammatory signals and the chemical signals that recruit neutrophils, which the authors suggest comes from bacterial fragments released by lysis being detected by the immune system. Whether that helps clear the infection or merely adds inflammation is not established, and they say the idea needs testing in more complex systems.

The part that reverses the answer

Bladders are not still. Urine flows, and the lining experiences shear. Almost all tissue-model infection work is done in static conditions because flow systems are difficult and expensive, so the team built one.

Under flow, the bacteria elongated, attached more readily, and formed more intracellular communities. Both nitrofurantoin and the combination performed worse. Not because less drug arrived, but despite more of it arriving.

So the ranking inverts across the sequence. In broth, nitrofurantoin is the clear winner. In urine it is weaker than it looks. In tissue it cannot finish the job. Under flow it does worse still. Each added piece of realism costs the drug something, and none of it involves resistance in the usual sense.

What the study can't say yet

No patients were treated. This is a laboratory model that resembles a bladder more closely than a test tube does, which is not the same as resembling a person. There is no immune system beyond the urothelial cells themselves, no bladder emptying, no repeat dosing over days.

The phage result is a single cocktail against particular strains. Phages are narrow by nature, and the authors say plainly that how effective they will prove for UTI in the clinic remains to be determined.

The mechanism behind the flow effect is also open. Shear changed how the bacteria behaved and drug performance fell, but whether the cause is the physical force, the altered bacterial state it produces, or the way flow moves drug past rather than into a surface, is not separated here.

Quick questions

What is an intracellular bacterial community? A cluster of bacteria living inside the cells lining the bladder. Being inside shelters them from drugs in the urine and is thought to contribute to infections coming back.

Does this mean susceptibility testing is useless? No. It reliably detects genuine resistance. What it does not capture is whether a drug can reach bacteria hiding in tissue under flowing conditions, which is a different question it was never designed to answer.

What's the one-line takeaway? The antibiotic that scores best in standard testing does not clear infection from a bladder-like tissue model, intracellular bacteria survive it, and adding physiological flow makes matters worse rather than better.