Macular oedema (ME) is retinal swelling that can impair vision and, in severe cases, lead to irreversible visual loss. ME has multiple causes including diabetes and ocular disease but can also be drug‑induced. Case reports have implicated paclitaxel (PTX), a cytotoxic agent used widely in oncology, as a potential cause of ME. Evidence from large‑scale datasets has been limited because ME is a rare adverse event and clinical trials are underpowered to detect rare ocular toxicity. The authors used a large spontaneous reporting system to detect signals that could guide clinical monitoring and hypothesis generation.
The study analysed anonymised reports from the Japanese Adverse Drug Event Report (JADER) database covering April 2004 through June 2025. The full JADER repository initially contained 1,615,041 reports. After excluding records with missing sex, age, body weight or reporting year, reports involving docetaxel or cabazitaxel, and records registering both conventional PTX and nab‑PTX together, the final analytic dataset comprised 585,738 reports. Within this set, 123 reports of ME were identified.
ME events were defined using MedDRA/J preferred terms for 'macular oedema' and 'cystoid macular oedema'. Extracted variables included sex, age (categorised in 10‑year increments in the DEMO file), body weight (10‑kg intervals), reporting year, presence of co‑existing diabetes mellitus (defined by relevant MedDRA terms), and concomitant use of prostaglandin analogue eye drops. Reports listing conventional PTX and nanoparticle albumin‑bound PTX (nab‑PTX) were identified for formulation‑specific analyses.
The primary approach was disproportionality analysis using reporting odds ratios (RORs) from two‑by‑two contingency tables. A positive signal required at least three ME cases and a lower bound of the 95% confidence interval for the ROR exceeding 1.0. To adjust for potential confounding, multivariable logistic regression models estimated adjusted RORs incorporating covariates chosen based on previous reports: sex, age ≥70 years, body weight <50 kg, co‑existing diabetes mellitus, and co‑administration of prostaglandin analogue eye drops. Analyses were conducted in R, with P < 0.05 considered statistically significant.
In the analysed dataset of 585,738 reports, 52.68% were male. Reports involving patients aged 70 years or older comprised 42.14% of records, and 34.78% of reports documented body weight under 50 kg. Overall exposure to the target drugs appeared in 2.76% of reports (16,192 reports), with 11,563 reports for conventional PTX and 4,629 for nab‑PTX. Co‑existing diabetes mellitus was present in 13.57% of reports, and 0.30% of reports included concomitant prostaglandin analogue eye drops.
Across the dataset, 123 reports of ME were identified. Disproportionality analysis produced significant signals for PTX formulations. The multivariable logistic regression estimated an adjusted ROR for overall PTX of 19.5 (95% CI 13.5–28.3), indicating elevated reporting of ME among PTX‑exposed reports compared with non‑PTX reports in the database. When evaluated by formulation, conventional PTX produced an adjusted ROR of 9.4 (95% CI 5.5–16.1), while nab‑PTX produced a larger adjusted ROR of 47.0 (95% CI 30.4–72.8).
After adjusting for covariates, co‑existing diabetes mellitus was associated with increased reporting of ME (adjusted ROR 1.8; 95% CI 1.2–2.9). Concomitant use of prostaglandin analogue eye drops showed a markedly higher adjusted ROR for ME reporting (19.9; 95% CI 8.6–45.7). Age ≥70 years, female sex, and body weight <50 kg were included in the multivariable models but the article reports the significant associations specifically for diabetes and prostaglandin analogue eye drop co‑administration.
These findings represent signal detection from a spontaneous reporting system and cannot establish causality or incidence. Spontaneous reports are subject to reporting bias, underreporting, missing data, and lack of denominator exposure data; these limitations restrict interpretation of absolute risk. The dataset required substantial exclusions for missing demographic fields, and the analyses depended on MedDRA coding and accurate capture of concomitant medications. The authors present results as hypothesis‑generating rather than confirmatory.
Within the constraints of pharmacovigilance data, the study suggests clinicians should be aware of a potential association between paclitaxel formulations and reported macular oedema, particularly with nab‑PTX and in patients with diabetes or those using prostaglandin analogue eye drops. The authors recommend consideration of closer ophthalmologic monitoring for these higher‑risk groups during PTX therapy. Further prospective studies or detailed pharmacoepidemiologic investigations are needed to quantify absolute risk and clarify causality.
The minimal dataset and code used for the analyses are available via Figshare at the DOI reported in the source article. The original raw data derive from the publicly accessible PMDA JADER database; no registration is required to access the source JADER files. The study authors note compliance with PMDA terms of use and that data processing steps were applied to the downloaded files.