Design: systematic review and random-effects meta-analysis of randomised trials, Cochrane RoB 2.0 and TESTEX assessed
Population: 8 trials, 504 patients with breast, oesophageal, gastric or rectal cancer on neoadjuvant chemotherapy
Primary outcome: pathological complete response
Effect: risk ratio 1.08 (95% CI 0.82-1.43, p=0.58); breast cancer meta-regression coefficient 0.83 (-0.00 to 1.67, p=0.05) for HR-positive HER2-negative disease

Eight randomised trials with 504 patients receiving neoadjuvant chemotherapy for breast, oesophageal, gastric or rectal cancer were pooled to test whether an aerobic and resistance exercise programme during treatment improves pathological complete response. It did not: pooled risk ratio 1.08 (95% CI 0.82 to 1.43, p=0.58).

A meta-regression restricted to the four breast cancer trials (367 participants) produced a coefficient of 0.83 (95% CI -0.00 to 1.67, p=0.05) suggesting possible benefit in hormone-receptor-positive, HER2-negative disease. The authors label this exploratory and the certainty low, and they are right to: a meta-regression across four trials, with an interval whose lower bound is exactly zero, is a hypothesis.

The reason to report a null result here is that exercise during chemotherapy is already recommended, and correctly so - for fatigue, function, cardiorespiratory fitness, treatment tolerance and quality of life, where the evidence is much better than this. What this paper removes is a claim that was starting to attach itself to exercise programmes: that they improve the tumour's response to the drugs. They should be offered for what they demonstrably do.

Keep recommending exercise during neoadjuvant chemotherapy - for function, fatigue and tolerance, not tumour response
Do not tell patients that exercise will improve their chance of a complete response
Prescribe the dose in specifics - aerobic plus resistance, with a supervised or checked component
Address the barriers that actually stop it: neuropathy, anaemia, port sites, transport to supervised sessions
Read the breast cancer subtype finding as a hypothesis, not as a reason to differentiate advice by subtype

Why it matters

It separates what exercise during chemotherapy genuinely delivers from a tumour-response claim the evidence does not support.

The statistics, in plain English

A risk ratio of 1.08 with an interval from 0.82 to 1.43 is a null result in a small evidence base - 504 patients across eight trials cannot exclude a moderate effect in either direction, so this is best read as no signal rather than as proof of no effect. The breast cancer meta-regression is weaker still: meta-regression uses trial-level averages to explain differences between trials, which with four trials means four data points, and a coefficient whose interval reaches exactly zero at p=0.05 is the definition of a finding that needs its own trial.