Patient-Specific IMRT QA Fails at 87%: What Do You Do Before Treating?

Not until the failure is explained. The discordance is informative: a point-dose agreement within 1.1% in the high-dose region is reassuring for the primary fluence, so an error in MU or output is unlikely. The failure is therefore likely in the spatial/dose-distribution comparison - low-dose regions, penumbra, or a field-specific effect. Note that the gamma threshold is 10%, so failures in the low-dose area are counted.

Before treating, determine where the failures are: map the gamma map to the field and to the patient anatomy. If failures cluster in the posterior fields or in a penumbra band, the suspect list includes couch-top/rail attenuation (not modeled or modeled incorrectly in the TPS), detector angular response for the posterior oblique beams, the 2.5 mm calculation grid in a steep gradient, and the detector's calibration for 10 MV. "It failed" is not a root cause; "the posterior 240-degree field fails because the couch is modeled without rails" is.

Work from cheapest/fastest to most disruptive:

1. Re-analyze the existing measurement - field-by-field gamma, absolute dose per field, and location of failures. Identify whether one field or all fields drive the failure.
2. Repeat the measurement with a different detector or geometry (film, EPID dosimetry, a second array orientation, or an ion chamber in a second point in a gradient region). This tests whether the result is detector-related.
3. Check the plan export and calculation - recalculate with a finer grid (1–1.5 mm) and confirm the dose grid, density overrides, and couch model match what was measured. Compare the independent MU/secondary calculation; it was not reported, so obtain it.
4. Measure with the couch in versus out if posterior fields are implicated - this is a classic cause of a gamma failure that point dosimetry misses.

Clearing criteria: the repeat measurement meets the 95% gamma action limit at the department's chosen criteria AND the point dose agrees within tolerance AND the independent calculation agrees. If any of those remain outside tolerance, escalate to physics and the RO; do not treat on the argument that "the point dose was fine." Document the investigation and the resolution - including whether the TPS or the QA process was at fault - so the next plan does not repeat it.

A couch-model error is a systematic planning error, not just a QA artifact: posterior beams passing through the couch are attenuated more than the TPS predicted, so the PTV may be underdosed and the rectum/other OARs may receive less than calculated. Consequences:

• The clinical plan itself may need recalculation with the correct couch structure and density; the difference is usually small per beam but can be clinically meaningful for a posterior-weighted 7-field prostate plan at 78 Gy.
• All plans using posterior-oblique fields in the department should be reviewed for the same issue - this is a process failure, not a single-patient failure.
• Verification: recalculate with the corrected couch model, re-run patient-specific QA (should now pass), and compare the corrected vs. original dose distributions to the PTV and rectum before fraction 1. If the difference is clinically significant, the RO must approve a re-plan.
• Documentation and QA records: record the root cause, the corrective action, and add a couch-model check to the annual/monthly QA and to the plan-check checklist. Report it through the department's incident/quality system.