Does Weight Loss Fake Kidney Numbers?

There is a mechanical reason to distrust kidney numbers taken during rapid weight loss. The standard estimate of kidney function is calculated from creatinine, and creatinine comes from muscle. Lose muscle, produce less creatinine, and the estimate improves while the kidney does exactly what it was doing before. A trial checked this against a directly measured filtration rate and found the worry does not materialize at around 10% weight loss.^[1]

Why the question exists

Estimated GFR is not a measurement. It is a formula fed a blood marker, and the usual marker is creatinine, a waste product of muscle metabolism. The formula assumes your creatinine production is roughly what a person of your age and sex produces. Someone shedding lean mass breaks that assumption in a specific direction: less muscle, less creatinine, an eGFR that reads higher than the truth.

Cystatin C is the usual alternative because it is not muscle-derived, but it has its own sensitivity to fat mass and inflammation. So both estimates could plausibly drift during weight loss, in different directions, for different reasons.

How the trial answered it

SMART randomized 101 adults with chronic kidney disease and overweight or obesity, without type 2 diabetes, to 24 weeks of semaglutide 2.4 mg weekly or placebo. The design point is what it measured alongside the estimates: iohexol clearance, a direct measurement of filtration rather than a calculation from it. Body composition came from bioimpedance spectroscopy.

Against every GFR measure — creatinine-based estimate, cystatin C-based estimate, and the measured value — changes in weight, lean mass and fat mass showed no correlation. All Spearman coefficients came in below 0.23, and multivariable adjustment did not change it. The authors conclude that weight reductions of 10% with semaglutide do not influence GFR estimates.

Read the confidence intervals on that table

That is a statement about statistical power rather than an error. A hundred-person trial is small for splitting a weight change into components with a technique as noisy as bioimpedance, the point estimates are the best available, and their direction agrees with the larger body of work we have covered in lean mass on a drug versus dieting. But a reader told flatly that lean mass fell is owed the interval that goes with it.

One more thing the table shows if you add it up: −2.5 kg of lean and −3.9 kg of fat comes to −6.4 kg, against a scale change of −9.1 kg. Bioimpedance is not a mass balance and its components are not obliged to sum to the scale. Some of the gap is the 0.9 L of extracellular water. The rest is measurement.

The blood pressure finding is separate and interesting

Systolic pressure fell 6.3 mm Hg more than placebo, and that fall correlated with the loss of extracellular water — Spearman 0.40, P = 0.005. It is the one correlation in the study that held.

The implication is that at least part of the blood pressure benefit here is fluid rather than weight. That is a mechanism worth knowing in a kidney population, where fluid status is already something clinicians manage, and it is a different pathway from the one people assume when they credit blood pressure improvement to having lost weight.

What this does not establish

• It is bounded at about 10%. The conclusion names that figure. Someone losing 20% or more is outside what this trial tested.
• It is 24 weeks. Whether estimates stay honest over years of treatment is not addressed.
• It is not a safety finding. Whether these drugs help or harm kidneys is a different question with its own evidence — see GLP-1s and your kidneys.
• Everyone in it already had CKD and none had type 2 diabetes. That is an unusual population, chosen deliberately, and not the average person on one of these drugs.