The 90Y microsphere dose that disagrees with the treatment plan

The most important point is that 99mTc-MAA is a surrogate, not the therapy itself. Systematically check:

1. Catheter position and flow. MAA particles are ~10-90 um and distribute by first-pass capillary trapping. If the microcatheter was more proximal at MAA time than at treatment (or vice versa), the lobar distribution differs. Compare the angiographic position and the extrahepatic/segmental distribution on both scans.
2. Arteriovenous shunting. A 4% lung shunt is acceptable, but a higher tumor-to-normal shunt at treatment can divert activity away from tumor and into normal parenchyma. Recompute the lung shunt from the post-treatment PET.
3. Reconstruction and quantification differences. MAA SPECT and 90Y PET have very different resolution, scatter, and partial-volume behavior. A 6.2 cm lesion is large enough that partial volume is modest, but the normal-liver VOI near the tumor will be contaminated by spill-in. Confirm the same VOI definitions, same reconstruction (e.g., TOF-OSEM with resolution recovery), and same segmentation method on both datasets.
4. Assumed versus measured activity. Verify the actual calibrated activity in the vial and the residual in the delivery system; a 1.4 GBq nominal dose can differ from delivered dose.
5. Partition model assumptions. The partition model requires a tumor-to-normal ratio (T/N). If the T/N was estimated from a region that included necrotic or hypovascular tumor, the tumor dose is overestimated and the normal-liver dose underestimated. This is consistent with the observed direction of the error (predicted tumor high, predicted normal low).
6. Tumor biology. Necrotic core, prior locoregional therapy, or altered arterial supply can make MAA uptake unrepresentative.

In practice, the MAA-based prediction is the weaker dataset; the post-treatment PET is the ground truth for what was delivered. The discrepancy itself is the finding to act on.

The post-treatment PET shows a normal-liver dose of 71 Gy, which is well above the commonly cited safety threshold for lobar resin microsphere therapy (roughly 40-50 Gy to the normal liver in a single session, and lower for whole-liver or compromised reserve). The tumor received only 61 Gy, which is below the typical tumoricidal range for HCC. This is the opposite of the intended plan and is a safety signal.

Recommendation: do not commit to a second lobar treatment in 6 weeks on the current data. Before any second treatment:

• Reassess liver function (bilirubin, albumin, INR, Child-Pugh, MELD) and imaging for radiation-induced liver disease or biliary injury.
• Recompute the cumulative normal-liver dose, not just the second-session dose. Cumulative normal-liver tolerance is the binding constraint.
• Repeat dosimetry with a better predictive dataset. Options include a repeat MAA with careful catheter matching, or a scout 90Y PET-based approach, and explicit measurement of the T/N ratio with necrotic core excluded from the tumor VOI.
• Consider a segmental rather than lobar approach if the tumor is confined to one segment, to spare normal parenchyma.
• Document the discrepancy and the rationale in the chart and discuss it with the multidisciplinary team.

The judgment being tested: the post-treatment scan is not just a QA artifact; it changes the risk-benefit calculation for the next fraction. The correct answer is to slow down, not to repeat the same plan with a dose adjustment based on a model that has already been shown to be wrong in this patient.