The study’s foundation rests on two of the world’s largest spontaneous reporting repositories. The researchers queried the United States Food and Drug Administration Adverse Event Monitoring System, known as AEMS and commonly referred to in the literature as FAERS, alongside the Japanese Adverse Drug Event Report database, or JADER, covering the period from January 2015 through June 2025. This window captures essentially the entire clinical era of PD-1 and PD-L1 inhibitors, the antibody drugs that block the programmed cell death 1 pathway and its ligand, thereby preventing tumors from co-opting a natural immune checkpoint to shield themselves. Because spontaneous databases record reports submitted by clinicians, pharmacists and patients rather than systematically collected clinical data, they cannot establish incidence rates on their own. What they can reveal, through carefully applied statistical tools, is whether a particular drug-event pairing is reported more often than would be expected by chance, a principle known as disproportionality.

To detect such signals, the team deployed two complementary analytical approaches. The first was the reporting odds ratio, or ROR, a frequentist measure that compares the odds of an immune thrombocytopenia report among patients taking PD-1 or PD-L1 inhibitors against the odds among reports involving all other drugs. The second was a Bayesian framework, which shrinks estimates for drug-event combinations with sparse data and thereby reduces false alarms. By requiring agreement across these methods, the researchers aimed to filter out statistical noise and isolate robust signals. In the American database, they identified 288 reports of immune thrombocytopenia among 157,788 total reports involving PD-1 or PD-L1 inhibitors, a proportion of 0.18 percent. In the Japanese database, the yield was higher: 322 reports among 69,205 relevant reports, or 0.47 percent, a difference that may reflect reporting culture, coding practices, or genuine differences in the treated populations between the two countries.

The headline statistical finding is unambiguous. Every individual PD-1 and PD-L1 inhibitor examined in the study generated a significant disproportionality signal for immune thrombocytopenia in the American database, indicating that the association is not driven by a single agent but is a class effect spanning the entire therapeutic family. In JADER, the overall reporting odds ratio was 3.57, with a 95 percent confidence interval of 3.13 to 4.06, meaning that reports of immune thrombocytopenia were roughly three and a half times more likely to involve these checkpoint inhibitors than other drugs. The American analysis produced an even stronger estimate. Crucially, the researchers recognized that Japanese reports appear in both databases, since Japanese cases can be submitted to the FDA system as well, and they therefore repeated the American analysis after excluding reports originating from Japan. The signal not only survived but remained strong, with a reporting odds ratio of 5.72 and a confidence interval of 4.83 to 6.77, demonstrating that the association is not an artifact of duplicated data flowing between the two national systems.

Beyond confirming that the signal exists, the study probed when these events occur, and the timing analysis yielded one of the most clinically actionable insights. The median time to onset in the American database was 43 days, with an interquartile range stretching from 13 to 126 days, meaning that half of all reported cases arose within roughly six weeks of starting treatment. To characterize the underlying hazard pattern, the team fitted a Weibull distribution to the time-to-onset data and calculated a shape parameter of 0.70, with a 95 percent confidence interval of 0.64 to 0.80. Because this value falls below 1, the distribution follows what safety researchers call an early-failure pattern: the risk of an immune thrombocytopenia report is concentrated in the early phase of treatment and gradually declines thereafter. This statistical signature echoes what has been observed for other immune-related adverse events of checkpoint inhibitors, many of which cluster in the first weeks to months of therapy when the newly unleashed immune response is at its most volatile.

Yet the Weibull analysis also carried a cautionary note. Although reports clustered early, they did not stop early. Cases continued to be recorded well beyond the initial treatment period, with the upper reaches of the interquartile range extending past four months and a tail of reports arriving even later. For clinicians, this means that vigilance for platelet counts cannot be confined to the first few infusion cycles. A patient who has tolerated a PD-1 inhibitor for half a year can still develop immune-mediated platelet destruction, and the study’s data suggest that the hazard, while front-loaded, never fully disappears during the treatment period. The authors frame this as a call for continued clinical vigilance throughout the course of immune checkpoint inhibitor therapy, with particular attention to platelet measurements in the early weeks but without assuming that later courses of treatment are risk-free.

To explore which patients might be most susceptible, the researchers conducted an exploratory multivariable logistic regression on the report-level data from the American database. This analysis adjusted for multiple characteristics simultaneously and identified two factors associated with a higher likelihood of an immune thrombocytopenia report: age between 65 and 85 years, and body weight of 75 kilograms or less. Both findings are biologically plausible. Older patients tend to have more dysregulated immune systems, a phenomenon sometimes described as immunosenescence, and lower body weight may correlate with altered drug exposure, comorbidity burdens, or baseline hematological vulnerability. The authors are careful to characterize this regression as exploratory, since spontaneous reporting data are subject to confounding, reporting biases and missing information, and the analysis identifies patterns in reports rather than causal risk factors in patients. Even so, the results give clinicians a concrete profile to watch: older, lighter patients early in the course of checkpoint inhibitor therapy.

The demographic picture of the reports themselves adds further texture. In the American database, men accounted for 53.8 percent of the immune thrombocytopenia reports, and individuals aged 65 or older accounted for 58.4 percent, a distribution that mirrors both the demographics of the cancer populations receiving these drugs and the age-related patterns of immune-related adverse events more broadly. Immune thrombocytopenia itself is not a trivial complication. Depending on severity, it can present as petechiae, bruising, mucosal bleeding or, in the worst cases, life-threatening hemorrhage, and it may force interruption or discontinuation of an otherwise effective cancer therapy. Because checkpoint inhibitors are now given to hundreds of thousands of patients worldwide across lung cancer, melanoma, renal cell carcinoma, head and neck cancer and many other indications, even a rare event translates into a meaningful absolute number of affected individuals.

The study’s limitations are those inherent to pharmacovigilance databases, and the authors acknowledge them directly. Spontaneous reports lack denominators, so the 0.18 and 0.47 percent figures represent proportions of reports rather than true incidence rates among treated patients. Reporting is subject to stimulated surveillance, media attention and regulatory alerts, and misclassification of events is possible. The regression findings are exploratory and cannot establish causation. Nevertheless, the convergence of evidence across two independent national databases, two statistical methodologies, a class-wide signal covering every individual drug, a sensitivity analysis that controlled for cross-database duplication, and a coherent early-failure timing pattern gives the findings substantial weight. The authors position their work as a foundation for active-surveillance studies, which follow patients prospectively and can measure true incidence, and they emphasize that the practical message for oncology teams is straightforward: monitor platelet counts with particular care in the first weeks of PD-1 or PD-L1 inhibitor treatment, remain alert throughout therapy, and pay special attention to older patients of lower body weight, so that this rare but potentially serious immune complication can be caught and managed before it becomes dangerous.

Subject of Research: Immune thrombocytopenia associated with PD-1/PD-L1 inhibitor therapy assessed through real-world pharmacovigilance database analysis

Article Title: A multidimensional assessment of immune thrombocytopenia reports involving PD-1/PD-L1 inhibitors: a real-world pharmacovigilance study using AEMS and JADER databases

Article References: Li, Y., Yang, Q., Pan, Y., & Mai, H. (2026). A multidimensional assessment of immune thrombocytopenia reports involving PD-1/PD-L1 inhibitors: a real-world pharmacovigilance study using AEMS and JADER databases. BMC Pharmacology and Toxicology. https://doi.org/10.1186/s40360-026-01246-5

Image Credits: AI Generated

DOI: 10.1186/s40360-026-01246-5

Keywords: PD-1 inhibitors, PD-L1 inhibitors, immune thrombocytopenia, immune checkpoint inhibitors, pharmacovigilance, FAERS, JADER, disproportionality analysis, immune-related adverse events, drug safety, platelets, immunotherapy