Personalized vs Fixed-Activity Dosimetry¹⁷⁷Lu · ⁹⁰Y

Quick Answer

Almost every external-beam radiation oncologist prescribes in absorbed dose (Gray), yet most radioligand therapy (RPT) is still given as a fixed administered activity — the same number of gigabecquerels to every patient, regardless of how much radiation actually reaches the tumor or the organs at risk. This is a striking inconsistency: RPT is radiotherapy, and absorbed dose varies several-fold between patients given identical activity. Dosimetry measures the dose actually delivered (from post-therapy SPECT/CT or PET), and personalized, dosimetry-guided dosing adjusts activity to hit a tumor-dose target while respecting organ limits. The evidence that this matters is strongest in ⁹⁰Y radioembolization (DOSISPHERE-01 showed personalized dosimetry improved response in HCC) and is accumulating for PRRT and Lu-PSMA. The historical barrier is practical, not conceptual: classical dosimetry requires multiple SPECT/CT time points and laborious segmentation. Single-time-point protocols and AI-automated segmentation/dosimetry are collapsing that burden — which is why automated dosimetry is widely seen as the change that will finally make personalized RPT routine.

Why fixed activity leaves benefit on the table

Give ten patients the same activity of ¹⁷⁷Lu-PSMA and the tumor and kidney absorbed doses will differ widely — driven by tumor burden, receptor density, renal function, and clearance kinetics. Fixed dosing therefore under-treats some patients (tumor dose below the threshold for response) while risking avoidable toxicity in others (organ dose near tolerance). Because RPT delivers ionizing radiation, the same dose–response and dose–toxicity relationships that govern external-beam therapy apply — which is the core argument that activity, not absorbed dose, is the wrong prescription unit.

What the evidence shows

Personalization is best proven where dosimetry is most tractable:

The counter-argument is honest: for β-emitters like ¹⁷⁷Lu, registration trials used fixed activity and showed benefit, prospective outcome trials proving personalized dosing is superior are still maturing, and dosimetry adds cost and complexity. The field is moving toward personalization, but the level-1 evidence base is still being built outside radioembolization.

The practical bottleneck — and how it falls

Classical dosimetry is demanding: serial SPECT/CT at several time points (e.g. ~4, 24, 96 h), image registration, organ and tumor segmentation, time–activity curve fitting, and dose calculation — per cycle, per patient. Two developments dismantle this:

• Single-time-point (STP) dosimetry — validated simplifications estimate absorbed dose from one post-therapy scan, cutting patient visits and workload with acceptable accuracy for many organs.
• AI-automated segmentation and dosimetry — deep learning contours organs and lesions and computes voxel dose in minutes, removing the manual step that made routine dosimetry impractical.

Together these turn dosimetry from a research-lab exercise into something a busy service can actually run every cycle — the operational key to personalization at scale.

High-Yield Pearls

• RPT is radiotherapy, yet most is dosed as fixed activity, not absorbed dose — a several-fold inter-patient dose variation.
• Dosimetry measures delivered dose (post-therapy SPECT/CT/PET); personalized dosing targets tumor dose within organ limits.
• DOSISPHERE-01 (⁹⁰Y in HCC) is the strongest evidence that personalized dosimetry improves response.
• The barrier was practical (multi-time-point SPECT/CT + manual segmentation) — single-time-point protocols and AI automation are removing it.
• Higher tumor absorbed dose generally tracks better outcomes; kidney/marrow dose tracks toxicity.

Common Pitfalls

• Treating administered activity as if it were the delivered dose — they diverge widely.
• Assuming fixed dosing is optimal because registration trials used it — that shows efficacy, not optimality.
• Ignoring single-time-point/AI options and dismissing dosimetry as too laborious for routine use.

Related Pages

• Framework: Dosimetry in theranostics, MIRD dosimetry — worked examples; evidence: DOSISPHERE / Y-90 radioembolization; frontier: AI & machine learning in nuclear medicine.

Self-Check

Key References

• DOSISPHERE-01 (Garin E, et al., Lancet Gastroenterol Hepatol 2021) — personalized dosimetry in ⁹⁰Y radioembolization for HCC.
• MIRD/EANM dosimetry guidance; validation studies of single-time-point dosimetry and AI-automated organ/lesion segmentation for RPT.