Three numbers govern the design of nearly every radiopharmaceutical therapy (RPT) trial: 23 Gy to the kidneys, 2 Gy to the bone marrow, and 200 mCi (7.4 GBq) per cycle (using lutetium-177 compounds). However, none were derived from data specific to the agents they now constrain. With 30+ phase III RPT trials active and α-emitting agents entering the clinic, every new program inherits these values as if they were validated dose limits. They are not.
RPT occupies a unique position between radiation therapy and systemic drug development. Unlike external beam radiotherapy (EBRT), where absorbed dose is prescribed to a defined target volume, RPT is prescribed by administered activity, a surrogate that ignores the patient-specific biodistribution that determines how much radiation reaches the tumor and the organs at risk. Unlike conventional chemotherapy, where the therapeutic agent is the molecule itself, in RPT the molecule is only the vehicle; the radiation it delivers is the treatment. Yet the regulatory and clinical framework treats RPT like a drug: fixed dose, fixed schedule, toxicity-based modifications. This framework was built on dose constraints borrowed from EBRT and iodine-131 thyroid therapy, adopted when the field was in its infancy and dosimetry was impractical. As the US Food and Drug Administration (FDA) acknowledged in its August 2025 draft guidance, these EBRT-derived limits have lessened applicability to RPT.
The 200 mCi (7.4 GBq) per cycle was not a maximum tolerated dose. In the first Lu-177-DOTATATE clinical report, Kwekkeboom et al escalated through 100, 150, and 200 mCi in patients and stopped. The stated reason was pharmacological: 200 mCi of carrier-added Lu-177 labeled onto 180-300 µg of peptide, and higher peptide mass would saturate somatostatin receptors and reduce tumor uptake. The cumulative dose of 750-800 mCi was separately capped by the 23 Gy kidney and 2 Gy marrow limits. No formal phase I stopping rules governed the per-cycle escalation.
This constraint was a function of the radiochemistry available in 2001. Carrier-added Lu-177 from NRG Petten and MURR had limited specific activity (approximately 20-30 Ci/mg), so activity and peptide mass scaled together: more GBq meant inescapably more µg.Modern Lutathera uses no-carrier-added Lu-177 (specific activity >3,000 GBq/mg) and achieves 7.4 GBq with only 100 µg, roughly half the original mass.At today's molar activity, 15 GBq could be labeled onto the same 200 µg that Kwekkeboom considered acceptable. The per-cycle activity has not been revisited in over 20 years.
Successive series of 131 and 504 patients confirmed the safety of 200 mCi. NETTER-1 adopted it verbatim,and the dose now sits on both Lutathera and Pluvicto labels. The Lutathera label reports a mean cumulative kidney dose of 18-20 Gy over four cycles, below even the EBRT-derived threshold it was supposed to approach.
The same pattern of pragmatic adoption without validation applies to the kidney dose limit. The 23 Gy threshold originates from the 1991 consensus by Emami et al.,which estimated the TD 5/5 for bilateral whole-kidney EBRT based on clinical experience from the 1940s through the 1980s. The authors acknowledged that some values were based purely on the experience of the clinicians. The threshold applies to dose rates of approximately 1 Gy per minute. RPT delivers kidney dose three orders of magnitude lower, where the biologically effective dose (BED) for the same physical dose is substantially reduced.Wessels et al quantified the effect: the same 23 Gy yields a BED of 37-40 Gy for Y-90 but only 27-30 Gy for Lu-177. Even the EBRT community has lowered its own estimate: QUANTEC revised kidney tolerance to 15-18 Gy.17The RPT community continues to cite the 1991 value rather than the 2010 revision. One possible explanation is practical: the QUANTEC estimate of 15-18 Gy would place Lutathera's mean kidney dose above the threshold, creating a regulatory conflict for an already approved drug. Whether or not this is the reason, the pattern of selective citation, using the older, higher value while the originating field has moved to a lower one, weakens the argument that 23 Gy represents a validated safety boundary for RPT.
The most cited PRRT kidney study, Bodei et al,proposed BED thresholds of 28 and 40 Gy from 28 patients (23 Y-90, five Lu-177). The Lu-177 patients contributed zero toxicity events. BED correlated with toxicity (P= .036), but absorbed dose did not (P= .68). These are Y-90 BED thresholds, from a sample smaller than many phase I trials, applied to Lu-177 without validation. The kidney doses themselves were measured using planar imaging with In-111 surrogates for Y-90, a method carrying 20%-30% uncertainty.
Radiopeptide uptake concentrates in the proximal tubules, where the radiosensitive glomeruli reside in the outer cortex.Y-90's 11 mm beta range cross-irradiates glomeruli; Lu-177's 1.7 mm range does not.For the same whole-kidney dose, glomerular exposure is 2-3-fold higher with Y-90.
A parallel story applies to bone marrow. Benua et al observed approximately 250 I-131 treatments at Memorial Sloan Kettering and found severe complications in one (4.2%) of 24 patients at blood doses of 1-2 Gy versus five (15.2%) of 33 (one fatal) at 2-3 Gy.The 200 rad (2 Gy) limit was drawn as an empirical line between these zones, not from a formal dose-response analysis. The measurement itself was of dose to blood, not to marrow, using a surrogate model that assumes no specific marrow binding, carries at least a factor-of-2 uncertainty, and misses 68% of marrow dose that comes from cross-irradiation by surrounding organs.Dorn et al later showed that 3 Gy was safe with no permanent marrow failure in 124 patients, suggesting even the original limit was conservative for I-131.
These inherited limits interact with a further problem: most pivotal RPT trials enrolled heavily pretreated patients, and the toxicity attributed to the radiopharmaceutical may reflect prior treatment. VISION required prior androgen receptor pathway inhibitors and taxane chemotherapy.PSMAfore tested the same drug at the same dose in taxane-naïve patients.The hematological difference is notable: Grade 3 to 4 thrombocytopenia was 9% in VISION versus 2.7% in PSMAfore, and anemia was 15% versus 7%. Although these were distinct trials with different eligibility criteria and patient populations, the magnitude of the difference is consistent with the known myelosuppressive effects of prior taxane exposure and raises the possibility that a substantial fraction of hematological toxicity in VISION reflected preexisting marrow compromise rather than radiopharmaceutical effect.
The same pattern holds for PRRT. Therapy-related myeloid neoplasm rates range from 0.7% in first-line NETTER-2 to 20% in patients heavily pretreated with alkylating agents, where MDS showed monosomy 7, a cytogenetic hallmark of alkylating-induced disease.Dose limits derived from heavily pretreated populations may not apply to earlier-line patients with intact organ function and could result in unnecessary activity reductions. If so, trials designed around these limits risk testing subtherapeutic doses, potentially obscuring the benefit of otherwise effective agents.
As RPTs move to earlier-line treatment and healthier populations, the dosimetry question shifts. In late-stage diffuse disease, escalation has typically meant pushing activity until organ toxicity defines the ceiling. In locally advanced or oligometastatic disease, where tumor volumes are confined and definable, prospective tumor dose prescription becomes feasible and arguably safer than toxicity-driven escalation. Prescribing radiation dose to a defined target is standard practice in every other form of radiotherapy. There is no reason RPT should be the exception, particularly for alpha emitters, where EBRT-derived normal tissue limits are even less applicable.
