Design: retrospective single-centre comparison of photon-counting against energy-integrating detector CT, with invasive angiography as reference
Population: 141 patients with coronary stents — 176 stents on photon-counting CT, 126 on conventional CT
Primary outcome: diagnostic performance for in-stent restenosis of 50% or more, and in-stent lumen image quality
Effect: non-diagnostic stents 1.1% versus 46.8%; per-stent accuracy 88.5%, negative predictive value 100%, positive predictive value 56.3%

One hundred and forty-one consecutive patients with coronary stents referred for CT angiography were studied retrospectively: 70 patients with 176 stents on photon-counting CT, and 71 patients with 126 stents on conventional energy-integrating detector CT, with invasive angiography as the reference for in-stent restenosis of 50% or more.

The image quality difference is the headline. Non-diagnostic stents fell from 46.8% on conventional CT to 1.1% on photon-counting CT (P<0.001), and the proportion graded excellent rose from 23% to 70.5% (P<0.001). Nearly half of stents being unreadable is the reason CT has never been a usable follow-up test after stenting, and that is the number that changed.

For detecting restenosis of 50% or more, per-stent accuracy was 88.5% (78.2 to 94.3) with sensitivity 100% (70.1 to 100), specificity 86.5% (74.7 to 93.3), negative predictive value 100% (92.1 to 100) and positive predictive value 56.3% (33.2 to 76.9). Per-patient figures were similar. So the test is good at excluding restenosis and poor at confirming it: a negative scan can reasonably stop the pathway, a positive one still needs the catheter.

That asymmetry is exactly how it should be used. In a patient with recurrent chest pain after percutaneous intervention, a non-invasive test that reliably rules restenosis out has real value even when its positives need confirming. The limits are that this was retrospective, single-centre, with few restenosis events — a sensitivity of 100% whose confidence interval starts at 70% rests on very small numbers — and that photon-counting scanners are available in only a handful of Indian centres.

Consider photon-counting coronary CT to exclude in-stent restenosis, not to confirm it
Expect about half of stents to be non-diagnostic on conventional energy-integrating CT; do not read that as a negative study
Send a positive result to invasive angiography — the positive predictive value here was 56.3%
Record stent diameter and material; narrow and radiopaque stents remain the hardest on any detector
Be explicit in the report about which stents were diagnostic and which were not

Don't overread it

Retrospective, single-centre, with two non-randomised scanner groups and few restenosis events — this establishes readability, not diagnostic equivalence to angiography.

The statistics, in plain English

Sensitivity of 100% with a confidence interval from 70.1% to 100% means only a handful of true restenoses were present — the point estimate is perfect and the uncertainty is large. Negative predictive value of 100% (92.1 to 100) is the usable figure, but it is inflated by the low prevalence of restenosis in this referral group and would fall in a higher-risk population. A positive predictive value of 56.3% means close to half of positive calls were wrong, which is the honest limit of the test.