📄 medicine

Bayesian Reappraisal of the Safety of [177Lu]Lu-PSMA Radioligand Therapy Across Clinical Settings

This Bayesian reappraisal of [177Lu]Lu-PSMA radioligand therapy across diverse clinical settings reveals that while severe neutropenia and acute renal toxicity are uncommon, the incidence of grade 3 or higher anemia and thrombocytopenia varies significantly, being notably higher in heavily pretreated or marrow-compromised patients, with long-term follow-up suggesting potential delayed renal function declines.

Original authors: Zijie Wang, Hiroshi Fukushima, Shugo Yajima, Motohiro Fujiwara, Hiroyuki Sato, Akihiro Hirakawa, Yosuke Yasuda, Hajime Tanaka, Soichiro Yoshida, Yasuhisa Fujii

Published 2026-09-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Zijie Wang, Hiroshi Fukushima, Shugo Yajima, Motohiro Fujiwara, Hiroyuki Sato, Akihiro Hirakawa, Yosuke Yasuda, Hajime Tanaka, Soichiro Yoshida, Yasuhisa Fujii

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Prostate cancer that has spread throughout the body and no longer responds to standard hormone treatments is a formidable challenge. For years, doctors have turned to chemotherapy, but many patients eventually run out of options. Recently, a different approach has emerged that targets the cancer cells directly using a tiny, radioactive particle. This treatment, known as radioligand therapy, works like a guided missile. It uses a molecule that seeks out a specific protein found on the surface of prostate cancer cells. Once attached, it delivers a dose of radiation that damages the cancer's DNA and kills it. Because the molecule is designed to stick to the cancer, it spares most healthy tissue, but it does pass through the kidneys and bone marrow, which are the body's blood-making factories. This creates a delicate balance: the treatment must be strong enough to fight the cancer without causing dangerous side effects in these vital organs.

A team of researchers set out to understand exactly how safe this treatment is across different types of patients. They looked at data from eight different groups of people who had received the therapy. These groups varied widely in their medical history. Some had never received a specific type of chemotherapy called taxanes, while others had already tried and failed multiple rounds of it. Some had cancer that had heavily invaded their bones, while others had kidney problems before they even started the new treatment. The researchers wanted to know if the safety profile changed depending on these factors. Instead of simply adding up all the numbers to get one average, they used a sophisticated method to look at each group separately, acknowledging that a patient with a weak bone marrow might react differently than a patient with a strong one.

The most significant finding was that the risk of low blood counts depends heavily on the patient's starting condition. The treatment can cause a drop in red blood cells, which leads to anemia, and a drop in platelets, which are needed for clotting. In groups of patients who had already undergone extensive prior chemotherapy or whose bone marrow was already compromised by cancer, these drops were more frequent and severe. In one group of patients with widespread bone marrow involvement, the estimated chance of a severe drop in red blood cells was nearly one in five. In contrast, for patients who had not yet received taxane chemotherapy, the risk was much lower, estimated at roughly six percent. However, a third type of blood cell, the neutrophil, which fights infection, remained relatively stable. Severe drops in these cells were uncommon across almost all groups, suggesting that the treatment does not wipe out the body's infection-fighting capabilities as often as it affects blood clotting or oxygen transport.

The kidneys, which filter waste from the blood, also showed a nuanced picture. In the major clinical trials, severe, immediate kidney failure was rare. Only a small fraction of patients experienced a sharp, dangerous decline in kidney function. However, when researchers looked at patients who were followed for a longer period, a different pattern appeared. About half of the patients in a long-term follow-up group saw their kidney filtration rate drop by at least fifteen percent over a year. This suggests that while the treatment might not cause a sudden, catastrophic failure, it can lead to a slow, gradual decline in kidney health over time. This is an important distinction, as it means doctors need to monitor kidney function carefully for years, not just during the weeks of treatment.

Other side effects were also mapped out. Dry mouth was a very common complaint, affecting more than half of the patients in some groups, but it was almost always mild. Severe dry mouth that caused significant pain or inability to eat was extremely rare. Similarly, feelings of nausea and fatigue were frequent but usually manageable. The treatment did lead some patients to stop therapy early due to side effects, but this happened less often in the newer trials involving patients who had not yet received taxane chemotherapy compared to those who had. Fatal events directly linked to the treatment were very uncommon, occurring in less than two percent of patients in the largest studies.

The researchers also compared the results from different trials to see if the treatment was safer in specific settings. They found a high probability that patients who had not yet received taxane chemotherapy experienced fewer severe blood-related side effects than those who had. This aligns with the idea that patients with a stronger bone marrow reserve can better withstand the treatment. However, when looking at smaller groups of patients, such as those in a specific Japanese trial, the data was too limited to draw firm conclusions. The uncertainty was wide, meaning that while the numbers looked different, the researchers could not be sure if those differences were real or just due to the small number of people involved.

Ultimately, this analysis confirms that the treatment is generally safe but requires careful selection and monitoring. The risk of severe anemia and low platelets is not uniform; it rises significantly in patients who are already heavily treated or whose bone marrow is struggling. The risk to the kidneys appears to be more about long-term, slow changes rather than immediate injury. By understanding these differences, doctors can better predict who might need extra support during treatment and who can expect a smoother course. The study does not claim to have solved all safety questions, particularly regarding the long-term effects on the kidneys, but it provides a clearer, more detailed map of the risks involved, helping to guide the use of this powerful therapy for the right patients at the right time.

Technical Summary: Bayesian Reappraisal of the Safety of [177Lu]Lu-PSMA Radioligand Therapy

Problem Statement
Lutetium-177 prostate-specific membrane antigen ([177Lu]Lu-PSMA) radioligand therapy is an established treatment for metastatic castration-resistant prostate cancer (mCRPC), expanding from post-taxane settings into earlier lines of therapy. While pivotal trials have established its efficacy, the safety profile—specifically regarding myelosuppression (anemia, thrombocytopenia, neutropenia), renal impairment, and salivary gland toxicity—remains difficult to interpret consistently across diverse clinical populations. Existing literature presents heterogeneous cohorts varying in prior treatment history, bone marrow reserve, baseline renal function, and follow-up duration. Reporting crude incidence rates alone, particularly in small cohorts or those with zero observed events, fails to adequately convey estimation uncertainty. Furthermore, direct pooling of highly heterogeneous data may be misleading. There is a need for a unified framework to estimate toxicity endpoints across distinct clinical settings while quantifying uncertainty without forcing a single pooled incidence rate.

Methodology
This study employed a targeted Bayesian reappraisal of prespecified clinical datasets rather than a systematic review. The analysis utilized published aggregate data from eight distinct sources representing varied clinical contexts:

Pivotal Randomized Trials: VISION (post-taxane), PSMAfore (taxane-experienced), and TheraP (post-docetaxel).
Special Populations and Real-World Cohorts: A Japanese phase II trial, a US expanded-access program, a cohort with diffuse bone marrow involvement, a cohort with baseline renal impairment, and a long-term renal follow-up cohort.

Statistical Framework:

Model: For each cohort $i$ and safety outcome, the observed event count $x_i$ was modeled using a Bayesian binomial distribution: $x_i \sim \text{Binomial}(n_i, p_i)$ , where $n_i$ is the evaluable denominator and $p_i$ is the cohort-specific event probability.
Priors: A weakly informative normal prior was applied on the logit scale: $\text{logit}(p_i) \sim \text{Normal}(0, 2.5^2)$ .
Zero-Event Handling: Cohorts with zero observed events were analyzed directly without adding artificial events (e.g., adding 0.5), preserving the integrity of the data while allowing the posterior distribution to reflect uncertainty.
Estimation: Posterior distributions were evaluated using deterministic numerical integration. Results are reported as posterior medians with equal-tailed 95% credible intervals (CrI).
Comparisons: Post hoc exploratory comparisons were performed between selected cohorts (e.g., PSMAfore vs. VISION) by calculating the posterior probability $P(p_A < p_B | D_A, D_B)$ . These comparisons were explicitly defined as unadjusted, descriptive, and hypothesis-generating, not causal estimates.

Key Results

Hematologic Toxicity:
- Anemia and Thrombocytopenia: Grade 3 or higher incidence varied significantly by setting. Estimates were lowest in the taxane-experienced PSMAfore cohort (approx. 6% for anemia, 2% for thrombocytopenia) and highest in heavily pretreated or marrow-compromised cohorts (e.g., diffuse bone marrow involvement: ~23% anemia, ~25% thrombocytopenia; US expanded-access: ~18% anemia).
- Neutropenia: Grade 3 or higher neutropenia remained uncommon (<5%) across most cohorts, including those with marrow involvement, though estimates were slightly higher in the marrow-compromised group.
- Comparative Probabilities: The posterior probability that PSMAfore had a lower incidence of Grade 3+ anemia than VISION was 99.8%, and for thrombocytopenia, 99.9%. For neutropenia, this probability was 94.6%.
Renal Safety:
- Acute Events: Severe acute renal events (Grade 3+) were uncommon in pivotal trials (approx. 1–4%). In a cohort with baseline renal impairment (GFR ≤60 mL/min), one patient experienced worsening from Grade 2 to Grade 3.
- Long-term Decline: In a long-term follow-up cohort (≥12 months), approximately 45% of patients experienced a ≥15% decline in estimated glomerular filtration rate (eGFR), and 23.6% experienced a ≥30% decline.
Other Toxicities:
- Dry Mouth: Any-grade dry mouth was common (up to 60% in some trials), but Grade 3+ events were rare (<1%).
- Systemic Symptoms: Nausea and fatigue were predominantly low-grade; Grade 3+ fatigue ranged from 0.4% to 5.9% in large trials.
- Discontinuation and Mortality: Treatment discontinuation due to adverse events was approximately 12% in VISION and 5% in PSMAfore. Fatal adverse events were uncommon, with no treatment-related deaths attributed to the drug in TheraP or the Japanese phase II trial.

Key Contributions

Unified Uncertainty Quantification: The study provides a unified Bayesian framework that estimates toxicity risks across heterogeneous clinical settings without calculating a misleading single pooled incidence. It explicitly quantifies uncertainty, particularly in small cohorts or those with zero events, where traditional frequentist methods might yield misleadingly precise or undefined results.
Context-Specific Safety Profiles: By analyzing distinct cohorts separately, the paper delineates how safety profiles shift based on clinical context. It highlights that while severe neutropenia is generally rare, anemia and thrombocytopenia risks are substantially elevated in patients with diffuse bone marrow involvement or those heavily pretreated.
Differentiation of Acute vs. Long-term Renal Risk: The analysis distinguishes between rare severe acute renal events and the more common, gradual decline in renal function observed in long-term follow-up, suggesting that short-term safety assessments may not fully capture delayed renal toxicity.
Exploratory Comparative Analysis: The paper offers probabilistic comparisons (e.g., 99.8% probability of lower anemia in PSMAfore vs. VISION) that contextualize the safety differences between taxane-experienced and post-taxane populations, while rigorously noting the limitations of these unadjusted comparisons.

Significance and Claims
The paper claims that [177Lu]Lu-PSMA therapy is associated with variable risks of Grade 3+ anemia and thrombocytopenia depending on the clinical setting, with higher risks observed in heavily pretreated or marrow-compromised patients, whereas severe neutropenia remains uncommon. It asserts that severe acute renal toxicity is rare, but long-term monitoring is necessary to detect gradual declines in renal function.

The authors emphasize that Bayesian analysis is particularly valuable for interpreting safety data in smaller or zero-event cohorts, preventing the misinterpretation of "0%" incidence as an absence of risk. The study concludes that while the therapy is generally safe, the variation in hematologic toxicity across settings underscores the importance of baseline marrow reserve and disease burden in risk assessment. The authors explicitly state that their exploratory comparisons are hypothesis-generating and do not constitute causal evidence of comparative safety, as they are unadjusted for differences in patient selection, disease burden, and follow-up duration. The findings support the need for baseline assessment and longitudinal monitoring of renal and hematologic function, especially as the therapy moves into earlier treatment lines.

Technical Summary: Bayesian Reappraisal of the Safety of [177Lu]Lu-PSMA Radioligand Therapy

More like this