19 SEP 2026 — In 48 of 49 patients with one of the deadliest kidney cancers, a standard radiotracer lit the disease up clearly. In 31 of them it showed tumours that conventional scans had not.
For 10 patients, that changed the treatment they were given. The study reports no specificity figure at all.
What renal medullary carcinoma is
Renal medullary carcinoma is rare, aggressive and defined by the loss of a gene called SMARCB1. It appears mainly in adolescents and young adults, more often in males, and patients frequently already have metastatic disease when they are diagnosed. Median overall survival is around 14.5 months.
Those characteristics give staging unusual weight here. When a cancer is this fast and this often already spread, the question of whether disease exists beyond the kidney is not a refinement of the diagnosis. It decides whether an operation is worth performing.
What the scans found
The study, published in the Journal of Nuclear Medicine, reviewed 49 patients who had 18F-FDG PET/CT as part of their care at the University of Texas MD Anderson Cancer Center between 2016 and 2025.
Forty-eight of the 49 had clearly FDG-avid disease. Forty-three showed avid nodal disease and 36 had disease outside the lymph nodes, most often in the lungs, then the bones. In 31 patients the scan identified lesions that anatomic imaging had not detected.
Among the 48 patients with active disease, 10 had their management altered on the strength of the PET findings. The authors put that at 21 per cent.
Why staging changes the surgery
Simone Krebs of MD Anderson, who led the work, said that in this cancer "the value of accurate staging cannot be overstated." Her reasoning is specific: moving a patient from localised to metastatic disease spares them surgery that cannot cure them and carries substantial harm.
That is an unusual way to frame an imaging benefit. The gain is not that more patients get treated. It is that fewer patients undergo a major operation whose premise the scan has just removed.
It also explains why FDG suits this tumour specifically. Kidney cancers vary in how avidly they take up the tracer, which limits its usefulness across the disease group as a whole. Renal medullary carcinoma takes it up strongly, so a technique that is unreliable in renal cancer broadly is well matched to this subtype.
What 49 patients cannot settle
This is a retrospective, single-centre review. Nobody was randomised, nothing was compared against a control arm, and the scans were ordered as part of clinical care rather than to a protocol.
Retrospective single-centre work at a major cancer centre also reflects its own referral patterns. Patients who reach MD Anderson with a rare tumour are not a random sample of everyone who has it. The reported rates describe this cohort.
Forty-nine patients is nevertheless a substantial series for this disease. That tension runs through rare-cancer evidence: the sample is small because the cancer is rare, and waiting for a larger one means waiting years while treating patients on less.
What the release leaves out
The abstract reports no specificity and no false-positive count. Detection figures alone cannot establish how often the scan indicated disease that was not there. In a study whose stated value is sparing patients from surgery, a false positive has a direct cost: an operation withheld on the strength of a lesion that is not cancer.
Nor does the published summary report how the additional lesions were confirmed, or what happened to the 10 patients whose management changed.
The gap resembles one we reported this week, where a pancreatic cancer blood test performed worst in the group it was designed for. A diagnostic is only as good as the population it is measured in and the errors its evaluation counts.
View author profile → · Editorial policy
