Non-small-cell lung cancer (NSCLC) remains the leading cause of cancer-related mortality worldwide, accounting for approximately 85% of all lung cancer cases (1, 2). Rearrangements of the anaplastic lymphoma kinase (ALK) gene occur in 3–7% of NSCLC cases and confer marked sensitivity to ALK tyrosine kinase inhibitors (TKIs) (3, 4). The identification of ALK rearrangements has enabled the development of molecularly targeted therapies, substantially improving outcomes for patients with advanced ALK-positive NSCLC (5, 6). Crizotinib, a first-generation ALK TKI approved in 2011, established a new standard of care for this molecular subtype. Compared to chemotherapy, crizotinib significantly prolongs median progression-free survival (7.7 vs. 3.0 months) and achieves a higher objective response rate (70.0%) with a median duration of response of 27.8 months (7–9).

Crizotinib has demonstrated efficacy in treating NSCLC patients with ROS1 rearrangements and c-MET mutations, thereby broadening its clinical indications (5, 6). However, its use is associated with a range of adverse events (AEs), including elevated transaminases, visual disturbances, bradycardia, QT interval prolongation, gastrointestinal symptoms (nausea, vomiting, diarrhea, constipation), and interstitial pneumonia (8, 10). Evidence from observational cohort studies on ocular toxicity remains inconsistent, with reporting frequency of vision disorders as low as 12.5 per 1000 person-years in Europe and 47.8 per 1000 person-years in the United States (11). Such underestimation is likely due to the mild and transient nature of most visual AEs, which are frequently underreported. Accordingly, a large-scale real-world investigation is warranted to comprehensively evaluate the ocular safety profile of crizotinib.

The FDA Adverse Event Reporting System (FAERS) and VigiBase are two major spontaneous reporting databases that provide critical real-world evidence for post-marketing pharmacovigilance (12). FAERS, maintained by the U.S. Food and Drug Administration, is extensively utilized to detect safety signals related to AEs associated with approved drugs and biologics (13). In parallel, VigiBase—developed and managed by the Uppsala Monitoring Centre on behalf of the World Health Organization—serves as the global repository for suspected adverse drug reactions, facilitating early detection of emerging safety concerns across diverse populations and therapeutic contexts (14).

Given the increasing clinical use of crizotinib, a comprehensive understanding of its adverse effect profile—particularly ocular toxicity—is essential for optimizing patient safety and informing therapeutic strategies. This study utilizes standardized data from FAERS and VigiBase to evaluate the risk of crizotinib-associated ocular AEs, aiming to refine its safety profile and provide evidence-based guidance for healthcare professionals and policy decision-makers in the management of these events.

To date, pharmacovigilance studies specifically addressing ocular AEs associated with crizotinib remain limited. This study presents, to our knowledge, the first comprehensive real-world evaluation of ocular toxicity linked to crizotinib using data from both the FAERS and VigiBase databases. Through disproportionality analysis across two independent international reporting systems, we identified consistent and robust safety signals for several ocular PTs, including visual impairment, vision blurred, and photopsia. TTO analysis further indicated that these AEs typically emerge early in the treatment course, highlighting the importance of prompt clinical monitoring.

Visual disturbances were observed in approximately 62% of patients in early clinical trials of crizotinib (17), with most events classified as low-grade and considered by investigators to have limited clinical significance (18). However, these ocular effects have not been systematically assessed in post-marketing surveillance. The present study addresses this gap by quantifying the association between crizotinib and ocular AEs using data from two large-scale pharmacovigilance databases. Our findings underscore the need to incorporate routine ophthalmologic monitoring into the clinical management of patients receiving crizotinib, particularly during the initial phase of treatment when such events are most likely to occur.

As shown in Table 1, ocular AEs associated with crizotinib were more frequently reported in female patients (49.19% in FAERS; 49.62% in VigiBase) than in male patients (40.21% in FAERS; 40.60% in VigiBase). This sex disparity may be attributed to the higher prevalence of ALK gene rearrangements among non-smoking females with NSCLC (19), leading to greater exposure to crizotinib in this population. Additionally, a year-by-year decline in crizotinib-related AE reports was observed from 2014 to 2025, which likely reflects the increasing clinical adoption of newer-generation ALK inhibitors with superior efficacy and safety profiles.

Our analysis of the FAERS and VigiBase databases confirms a strong association between crizotinib and a range of visual disturbances, including blurred vision, diplopia, photophobia, and vitreous floaters. In addition, robust disproportionality signals were observed for less frequently reported terms such as halo vision, amaurosis, and visual field defects. Prior clinical studies have consistently described these ocular AEs as low grade (Grade 1–2) and self-limiting (20). For instance, pivotal trials reported that approximately 60% of patients experienced vision-related symptoms, yet none required dose adjustments due to these effects (21). However, even transient visual disturbances can interfere with safety-critical activities. Expert recommendations caution that symptoms such as photopsia or progressive floaters may signal serious retinal pathology and advise patients to exercise caution when performing tasks like driving or operating machinery (20, 22).

The visual disturbances associated with crizotinib are likely attributable to its effects on retinal signal transduction pathways. In addition to ALK inhibition, crizotinib targets c-MET and ROS1 kinases (23). While ALK protein is expressed in the retinal ganglion and inner nuclear layers, c-MET appears absent from the neural retina. Experimental data have shown that crizotinib alters retinal electrophysiology—specifically, reducing the electroretinogram b-wave amplitude during dark adaptation—an effect not observed with ALK-selective inhibitors (22). Studies further demonstrated that crizotinib more potently disrupted ON/OFF ganglion cell responses compared to alectinib, indicating broader retinal functional impairment (24). These findings suggest that crizotinib’s off-target inhibition of MET and ROS1, which are expressed in retinal tissue, contributes to its ocular toxicity. By disrupting photoreceptor–bipolar cell transmission and ganglion cell signaling, crizotinib may induce characteristic symptoms such as photopsia, flashes, and blurred vision (25). Although uncommon, inflammatory or vascular complications—including optic disc edema with vasculitic leakage—have also been reported, resolving after switching to alternative ALK inhibitors (26). Collectively, these data support a biologically plausible mechanism for crizotinib-induced visual AEs, driven by interference with retinal kinase signaling pathways.

Based on these findings, we recommend incorporating proactive ophthalmologic monitoring into clinical protocols for patients receiving crizotinib. A comprehensive baseline ophthalmic examination should be conducted prior to treatment initiation to document pre-existing conditions and establish a reference point (27). Follow-up evaluations, ideally within the first month of therapy, are advised to detect early-onset visual symptoms. Throughout the treatment course, clinicians should routinely inquire about visual disturbances and document any reported symptoms. Consistent with regulatory guidance, patients should receive regular visual assessments—monthly when feasible—and be instructed to promptly report any new or worsening visual changes. Symptoms such as blurred or double vision, photopsia, halos, or visual field defects should trigger timely referral to an ophthalmologist or neuro-ophthalmologist. Unexplained visual loss or severe ocular events warrant urgent specialist evaluation. Patient education is also critical: individuals should be informed of the potential for mild visual AEs and advised to exercise caution when engaging in visually demanding activities (e.g., driving or operating machinery) until symptoms are stable (28). In clinical practice, this includes preemptive discussions about visual risks and clear guidance on activity restrictions should significant symptoms occur.

Our findings are consistent with, and expand upon, prior observations from clinical trials and small cohort studies. Visual disturbances associated with crizotinib were well documented in early-phase trials, with approximately 60–62% of patients reporting mild symptoms, typically within the first week of treatment (21). In these controlled settings, the visual effects were generally considered trivial and rarely led to treatment modification. Our large-scale pharmacovigilance analysis corroborates that most ocular AEs are indeed low-grade; however, it also reveals specific patterns—such as elevated reporting odds for photopsia, vitreous floaters, and visual brightness—that may be underrecognized in clinical trials. Prospective ophthalmologic substudies have similarly reported frequent objective findings (e.g., changes in refraction and retinal thickness on optical coherence tomography) in crizotinib-treated patients, though without serious clinical sequelae (29). Notably, Solomon et al. found that such objective abnormalities often did not correlate with patient-reported symptoms, suggesting that many visual effects of crizotinib may be subtle or subclinical (29). Our study complements these limited-sample investigations by leveraging global real-world pharmacovigilance data. By aggregating reports across diverse populations, indications, and extended treatment durations, FAERS and VigiBase capture rare or delayed events—such as optic nerve disorders—that smaller trials are likely to overlook.

Nonetheless, the intrinsic limitations of spontaneous reporting systems must be considered when interpreting these findings. Neither FAERS nor VigiBase contains denominator data (i.e., the total number of patients exposed), precluding calculation of true incidence or risk estimates (30). Underreporting is a well-recognized issue, particularly for transient, low-grade events that may not prompt medical attention. Conversely, reporting bias may inflate associations, as known or anticipated AEs are more likely to be reported. Additionally, confounding by co-medications or comorbid conditions may obscure causality. Disproportionality analysis is designed to detect statistical signals but does not establish a causal relationship between a drug and an AEs. Furthermore, both databases lack granular clinical details, such as precise onset timing or ophthalmologic examination findings, limiting mechanistic interpretation (31). Despite these limitations, the use of two independent databases enhances robustness and mitigates source-specific biases. In summary, while spontaneous reporting cannot substitute for prospective studies, our findings offer complementary real-world evidence: crizotinib-associated ocular AEs are generally mild, occur early in therapy, and follow recognizable patterns that support current monitoring practices. Future prospective studies or registries with systematic ophthalmologic assessment are warranted to clarify incidence, mechanisms, and long-term outcomes.

This pharmacovigilance analysis of FAERS and VigiBase revealed consistent disproportionality signals linking crizotinib to multiple ocular AEs. Strong associations were identified for blurred vision, diplopia, photophobia, vitreous floaters, halo vision, amaurosis, and visual field defects. These events typically occurred early in treatment and, while predominantly low-grade, may transiently impair visual function, particularly in tasks requiring high acuity. Notably, some events such as halo vision and amaurosis, infrequently reported in trials, emerged as significant signals in real-world data. Despite the inherent limitations of spontaneous reporting systems—such as reporting bias and absence of incidence denominators—these findings underscore the need for heightened clinical vigilance. Further prospective studies and mechanistic investigations are warranted to validate these associations and clarify the pathophysiological basis of crizotinib-induced ocular toxicity.