Post-marketing safety analysis of nafamostat: A retrospective pharmacovigilance study using the Japanese Adverse Drug Event Report (JADER) database

Abstract

Nafamostat is a broad-spectrum serine protease inhibitor approved for pancreatitis, disseminated intravascular coagulation, and extracorporeal circulation anticoagulation. It also showed potential anti-SARS-CoV-2 activity during the COVID-19 pandemic, leading to expanded clinical use. This study aimed to explore post-marketing adverse event (AE) signals of nafamostat based on the Japanese Adverse Drug Event Report database so as to provide evidence for clinical safety management. Retrospective analysis was performed on AE reports retrieved from the Japanese Adverse Drug Event Report database between Q1 2004 and Q3 2024. Four disproportionality analysis algorithms, including reporting odds ratio, proportional reporting ratio, Bayesian confidence propagation neural network, and multi-item gamma Poisson shrinker, were adopted to detect AE signals. The distribution of AEs across system organ classes and preferred terms was summarized, and the time to onset was analyzed using the Weibull shape parameter test. A total of 2051 valid AE reports with nafamostat as the primary suspected drug were included, covering 20 system organ classes. The most prominent signals were observed in immune system disorders (n = 997) and vascular disorders (n = 370). Twenty-four preferred terms met the screening criteria of all 4 algorithms, among which anaphylactic shock (n = 746), shock (n = 341), and hyperkalemia (n = 200) were the most frequently reported. Six potential novel signals not recorded in the drug label were identified, namely acquired factor V deficiency, increased viral load, burning sensation, retroperitoneal hemorrhage, abdominal wall hematoma, and device-related thrombosis. Further bias analysis confirmed that 3 signals were false-positive results caused by confounding by indication or protopathic bias, while burning sensation had clear biological plausibility. The median time to onset was 1 day (interquartile range: 1–9 days); 58.55% of AEs occurred on the 1st day of administration, and 82.63% developed within 30 days. The Weibull test indicated that the AE risk peaked at the initial medication stage and gradually decreased over time. This real-world study clarified the safety profile of nafamostat, identifying severe allergic reactions and hyperkalemia as the main AEs. Of 6 initially detected unlabeled signals, 3 were confirmed as false-positive due to confounding bias, while the remaining 3 (bleeding complications and abnormal burning sensation) warrant clinical monitoring. Close monitoring within 24 hours after administration is highly recommended to reduce drug-related risks.

1. Introduction

Nafamostat (FUT-175) is a synthetic, broad-spectrum serine protease inhibitor with activity against various enzymes, including trypsin, components of the complement and coagulation cascades, and the contact system.^[1–4] Initially approved in Japan in 1986 for treating acute and chronic pancreatitis, disseminated intravascular coagulation, and extracorporeal circulation anticoagulation therapy, its therapeutic profile has drawn renewed attention during the COVID-19 pandemic (https://www.pmda.go.jp/PmdaSearch/iyakuDetail/GeneralList/3999407). Both clinical^[5–8] and preclinical studies^[9–11] have indicated its ability to interfere with SARS-CoV-2 S protein-mediated membrane fusion, suggesting antiviral potential.

While therapeutic benefits are important, drug safety and adverse reactions warrant equal consideration. Hyperkalemia is a recognized adverse effect of nafamostat.^[12–14] Given its clinical utilization, ongoing post-marketing surveillance is crucial. Spontaneous reporting databases such as the Japanese Adverse Drug Event Report (JADER) database play a vital role in capturing post-approval safety signals that may be underrepresented in controlled clinical trials.^[15] Analyzing JADER data enables a more comprehensive assessment of nafamostat’s safety profile. This study aims to investigate adverse event (AE) signals associated with nafamostat using the JADER database to support evidence-based clinical practice.

2. Methods

2.1. Study design and data sources

This retrospective pharmacovigilance study utilized data from the JADER database, maintained by Japan’s Pharmaceuticals and Medical Devices Agency, an administrative body under the Ministry of Health, Labour and Welfare responsible for regulating pharmaceuticals and medical devices.^[16] Publicly accessible JADER data files from the 1st quarter of 2004 to the 3rd quarter of 2024 were downloaded from the Pharmaceuticals and Medical Devices Agency website. The database comprises 4 datasets: demographic information, drug information, AE, and primary illnesses.^[17] Reports related to nafamostat were extracted, and all preferred terms (PTs) were mapped to the corresponding system organ classes (SOCs) according to Medical Dictionary for Regulatory Activities version 27.1. To maintain terminology consistency, earlier version terms were programmatically aligned with this version. Duplicate entries were removed based on a unique report identifier (primaryid). Only reports where nafamostat was designated as the primary suspected (PS) drug were included.

This study did not require ethics approval or informed consent, as it used only publicly available, anonymized AE data from the JADER database, with no new personal data or interventions introduced.

2.2. Time to onset (TTO) of analysis of nafamostat-associated AEs

TTO was defined as the number of days from drug initiation to event onset, with 1 day added to avoid 0-day values when the event occurred on the start date.^[18] Only the reports with available TTO data were analyzed. Reports with inaccurate, missing, or erroneous data were excluded to ensure the accuracy of TTO.

The incidence of AEs varies over time after the initiation of drug treatment. TTO analysis was conducted using the Weibull shape parameter (WSP) test to describe the risk that the incidence of AEs increases or decreases over time.^[19]

The shape of the Weibull distribution was described by 2 parameters: scale (α) and shape (β). The scale parameter α of the Weibull distribution determines the scale of the distribution function. The shape parameter β of the Weibull distribution determines the shape of the distribution function. Three hazard types are described in the WSP test: early failure type means the hazard of the AEs decreases over time (β < 1 and 95% confidence interval [CI] < 1); random failure type means the hazard of the AEs constantly occurs over time (β was equal to or nearly 1 and its 95% CI included the value 1); wear-out failure type means the hazard of the AEs increases over time (β > 1 and 95% CI > 1).

2.3. Statistical analysis

Disproportionality analyses were conducted using 4 established algorithms: the reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and the multi-item gamma Poisson shrinker (MGPS)-based empirical Bayesian geometric mean.^[20] The following thresholds were used for signal detection: ROR with the lower limit of the 95% CI > 1; PRR ≥ 2, χ² ≥ 4; BCPNN with the lower limit of the 95% CI of the information component (IC025) > 0; MGPS with the lower limit of the 95% CI of the empirical Bayesian geometric mean ≥2. Corresponding formulas and thresholds are provided in Table S1, Supplemental Digital Content 1. Signals were defined as statistically significant associations identified by these methods. Potential novel signals were considered as significant PTs not currently listed in the approved prescribing information.^[21] All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing).

3. Results

3.1. Descriptive analysis

From the 1st quarter of 2004 to the 3rd quarter of 2024, 2051 AE reports listing nafamostat as the PS drug were identified (Table 1). The cohort consisted of 38.42% females (n = 788), 60.95% males (n = 1250), and 0.63% with unspecified sex (n = 13). Age distribution was as follows: 0.73% (n = 15) under 10 years, 23.45% (n = 481) between 10 and 60 years, and 74.16% (n = 1521) over 60 years. Regarding body weight, 45.49% (n = 933) of patients weighed between 50 and 100 kg. Healthcare professionals submitted the majority of reports (96.05%).

Table 1.

Clinical characteristics of reports with nafamostat from the JADER database (the 1st quarter of 2004 to the 3rd quarter of 2024).

Characteristics	Case number, n	Case proportion, %
Sex
Female	788	38.42
Male	1250	60.95
Unknown	13	0.63
Age (yr)
<10	15	0.73
10–60	481	23.45
≥60	1521	74.16
Unknown	34	1.66
Weight (kg)
<50	548	26.72
50–100	933	45.49
≥100	4	0.20
Unknown	566	27.60
Reporter
Physician	1681	81.96
Health-professional	91	4.44
Pharmacist	198	9.65
Unknown	79	3.85
Consumer	2	0.10
Indications (top 5)
Renal and urinary disorders	862	22.27
Surgical and medical procedures	432	11.16
Metabolism and nutrition disorders	373	9.64
Vascular disorders	337	8.71
Cardiac disorders	276	7.13
Outcomes of adverse events
Death	123	4.60
Recovery	1849	69.17
Remission	479	17.92
After affects	16	0.60
No recovery	88	3.29
Unknown	118	4.41

JADER = Japanese Adverse Drug Event Report.

The most common therapeutic indications were renal and urinary disorders (n = 862), surgical and medical procedures (n = 432), metabolism and nutrition disorders (n = 373), vascular disorders (n = 337), and cardiac disorders (n = 276). Clinical outcomes were predominantly recovery (69.17%) or remission (17.92%).

Notably, the annual count of reported AEs related to nafamostat exhibited a declining trend over time (Fig. 1).

Figure 1.

Number of reported cases of nafamostat-related adverse events in the Japanese Adverse Drug Event Report (JADER) database from the 1st quarter of 2004 to the 3rd quarter of 2024.

3.2. Signal detection of nafamostat at the system organ class level

Figure 2 and Table 2 illustrate the distribution of AEs across SOCs for nafamostat as the PS drug. Significant signals meeting all 4 algorithms were observed for immune system disorders and vascular disorders. Additionally, investigations, metabolism and nutrition disorders, and product issues showed significant signals in at least 1 algorithm.

Figure 2.

The distribution of adverse events induced by nafamostat at the system organ class (SOC) level.

Table 2.

Signal strength of reports of nafamostat at the SOC level in JADER database.

System organ class (SOC)	N	ROR (95% CI)	PRR (χ2)	EBGM (EBGM05)	IC (IC025)
Immune system disorders	997	19.19 (17.74–20.77)*	12.41 (10,565.85)*	12.17 (11.25)*	3.61 (1.94)*
Vascular disorders	370	6.03 (5.4–6.74)*	5.34 (1326.78)*	5.3 (4.74)*	2.41 (0.74)*
Investigations	287	1.15 (1.02–1.3)*	1.13 (4.9)	1.13 (1)	0.18 (−1.49)
Metabolism and nutrition disorders	266	2.44 (2.15–2.77)*	2.3 (202.53)*	2.29 (2.02)*	1.2 (−0.47)
General disorders and administration site conditions	123	0.69 (0.57–0.82)	0.7 (16.62)	0.7 (0.59)	−0.51 (−2.18)
Respiratory, thoracic and mediastinal disorders	110	0.55 (0.45–0.66)	0.57 (39.14)	0.57 (0.47)	−0.82 (−2.48)
Gastrointestinal disorders	108	0.49 (0.41–0.6)	0.51 (54.02)	0.51 (0.42)	−0.96 (−2.63)
Skin and subcutaneous tissue disorders	102	0.61 (0.5–0.74)	0.63 (24.27)	0.63 (0.51)	−0.68 (−2.34)
Blood and lymphatic system disorders	89	0.49 (0.4–0.61)	0.51 (45.08)	0.51 (0.41)	−0.97 (−2.64)
Hepatobiliary disorders	59	0.54 (0.41–0.69)	0.55 (23.21)	0.55 (0.42)	−0.87 (−2.54)
Nervous system disorders	55	0.2 (0.16–0.26)	0.22 (168.88)	0.22 (0.17)	−2.19 (−3.86)
Cardiac disorders	39	0.34 (0.25–0.47)	0.35 (47.97)	0.35 (0.26)	−1.5 (−3.16)
Infections and infestations	23	0.1 (0.07–0.15)	0.11 (189.12)	0.11 (0.07)	−3.24 (−4.91)
Musculoskeletal and connective tissue disorders	16	0.22 (0.13–0.36)	0.22 (44.67)	0.22 (0.14)	−2.17 (−3.83)
Injury, poisoning and procedural complications	11	0.14 (0.08–0.25)	0.14 (59.32)	0.14 (0.08)	−2.82 (−4.49)
Renal and urinary disorders	9	0.08 (0.04–0.16)	0.09 (91.4)	0.09 (0.04)	−3.54 (−5.21)
Product issues	4	2.65 (0.99–7.09)	2.65 (4.1)*	2.64 (0.99)	1.4 (−0.27)
Psychiatric disorders	2	0.04 (0.01–0.14)	0.04 (51.79)	0.04 (0.01)	−4.77 (−6.44)
Endocrine disorders	2	0.04 (0.01–0.18)	0.04 (41.74)	0.04 (0.01)	−4.49 (−6.16)
Eye disorders	1	0.02 (0–0.17)	0.02 (38.92)	0.02 (0)	−5.33 (−7)

CI = confidence interval, EBGM = empirical Bayesian geometric mean, EBGM05 = the lower limit of 95% CI of EBGM, IC = information component, IC025 = the lower limit of 95% CI of the IC, JADER = Japanese Adverse Drug Event Report, PRR = proportional reporting ratio, ROR = reporting odds ratio, SOC = system organ class, χ² = chi-squared.

^*Indicates statistically significant signals in algorithm.

3.3. Signal detection of nafamostat at the preferred term level

Twenty-four PTs showed significant disproportionality across all 4 algorithms (Table S2, Supplemental Digital Content 2). The most frequent significant PTs were anaphylactic shock (n = 746), shock (n = 341), and hyperkalaemia (n = 200). Figure 3 presents these PTs categorized by ROR magnitude.

Figure 3.

Forest plot of preferred terms (PTs) signals meeting positive criteria across 4 disproportionality analysis algorithms (ranked according to reporting odds ratio) with the x-axis displayed on a logarithmic scale (log ROR). CI = confidence interval, PT = preferred term, ROR = reporting odds ratio, SOC = system organ class.

Comparison with the official drug labeling revealed 6 PTs not previously documented: acquired factor V deficiency, increased viral load, burning sensation, retroperitoneal hemorrhage, abdominal wall hematoma, and device-related thrombosis. These 6 unlabeled preferred terms will be further stratified and evaluated for confounding bias and biological plausibility in Section 4.

3.4. TTO analysis of nafamostat-associated AEs

Among the 2051 primary suspected reports, valid TTO data were obtained for 2147 individual AE records, as a single report may contain multiple AEs with different onset times. Analysis of available onset data revealed a median time to AE onset of 1 day (interquartile range 1–9 days). Most AEs (82.63%) occurred within 30 days, with 58.55% (1257/2147) emerging on the 1st day. A small proportion (3.82%) of AEs occurred as late as 360 days post-administration. The non-monotonic count between the 31 to 60 day and 61 to 90 day bins arises from inconsistent free-text onset date documentation in raw JADER spontaneous reports, rather than a true clinical risk fluctuation. WSP analysis yielded a β value of 0.325 (95% CI: 0.316–0.334), with β < 1 indicating a decreasing hazard function over time. This statistically confirms that the risk of nafamostat-associated AEs is highest immediately after treatment initiation, consistent with the observed concentration of events on day 1. The TTO distribution is presented in Figure 4.

Figure 4.

(A) The distribution of time to onset of AEs induced by nafamostat. (B) Weibull fitting of time to onset (TTO) for nafamostat-related AEs. AE = adverse event, CI = confidence interval, IQR = interquartile range.

4. Discussion

Continuous post-marketing surveillance is indispensable for comprehensively evaluating drug safety and balancing the clinical benefits and risks of medications. In this retrospective pharmacovigilance study, 4 mainstream disproportionality analysis methods, namely ROR, PRR, BCPNN, and MGPS, were applied to mine AE signals associated with nafamostat based on the JADER database. The real-world data from spontaneous reporting systems can effectively complement the limitations of controlled clinical trials, which often exclude complex comorbidities and fail to capture rare or long-term adverse reactions.^{[6,8,22–26]}

In terms of demographic characteristics of included AE reports, male patients accounted for 60.95% of all cases, notably higher than female patients (38.42%). This gender discrepancy is primarily attributed to the clinical application scenarios of nafamostat. The drug is widely used for renal diseases, metabolic disorders, and cardiovascular diseases, conditions that have a higher prevalence in male populations.^[27] Relevant epidemiological data also confirm that men have a higher incidence of renal replacement therapy and cardiovascular diseases,^[28] leading to greater drug exposure and more AE reports among males.

Age stratification showed that patients aged 60 years and above occupied 74.16% of all reports. This phenomenon is closely linked to the aging population in Japan and the growing number of elderly patients requiring dialysis and extracorporeal circulation treatment.^[29–31] Elderly patients generally have declined physiological function and impaired renal clearance capacity, making them a high-risk group for adverse reactions during nafamostat administration, which deserves close clinical attention.

The annual number of nafamostat-related AE reports presented a gradual declining trend during the study period. This trend can be partially explained by the Weber effect, a common rule for spontaneous reporting: AE reporting volume usually peaks within 2 to 3 years after a drug is widely used and then decreases steadily.^[32] In addition, the off-label use of nafamostat for anti-SARS-CoV-2 treatment gradually faded after the COVID-19 pandemic, also contributing to the reduction of annual reports. It should be noted that the current data cannot fully confirm whether improved rational drug use is the dominant factor for the declining reporting trend.

Disproportionality analysis revealed that nafamostat-related adverse reactions involved multiple SOCs. The most prominent AE signals were concentrated in immune system disorders and vascular disorders, followed by laboratory test abnormalities and metabolic disorders. Consistent with existing safety data, immune-related reactions including anaphylactic shock, shock, and hypersensitivity were the most frequently reported AEs.^[33–36] The underlying mechanism has been well elucidated: nafamostat is a potent inhibitor of human diamine oxidase (DAO). Since DAO is the key enzyme responsible for extracellular histamine degradation, DAO inhibition impairs histamine clearance in vivo, thereby inducing or exacerbating hypersensitivity and anaphylactoid reactions, especially in patients receiving hemodialysis or extracorporeal circulation support.^[37]

Hyperkalaemia was another prominent and well-documented adverse reaction of nafamostat. Pharmacologically, nafamostat inhibits epithelial sodium channels in the renal collecting duct, which reduces renal potassium excretion and consequently elevates serum potassium levels.^[12,38] This mechanism is similar to that of amiloride. Given the high proportion of elderly and renally impaired patients in the study population, routine monitoring of serum potassium is essential during clinical medication. Beyond potassium metabolism, epithelial sodium channels inhibition also affects sodium homeostasis, which has been reported in patients receiving nafamostat during mechanical circulatory support.^[39]

At the PT level, 24 significant AE signals were identified by all 4 analytical algorithms. Combined with the official prescribing information of nafamostat, 6 newly detected PTs were further classified and interpreted according to clinical context, pharmacological characteristics, and potential biases.

The 1st category consists of false-positive signals caused by confounding biases, including acquired factor V deficiency, increased viral load, and device-related thrombosis. These statistical associations do not represent true drug-induced adverse reactions. Acquired factor V deficiency is a preexisting coagulopathy. Nafamostat is preferentially selected as a regional anticoagulant for patients with this disease because systemic heparin is contraindicated due to bleeding risks.^[13,40] The detected signal is a typical case of confounding by indication, rather than a direct adverse effect of nafamostat.^[41] Increased viral load was only reported during the COVID-19 pandemic, when nafamostat was explored as a potential anti-SARS-CoV-2 agent. This finding reflects therapeutic failure or natural disease progression, namely protopathic bias, instead of drug toxicity. Multiple clinical trials have also verified the limited antiviral efficacy of nafamostat against SARS-CoV-2,^[6,8] further supporting this interpretation. Device-related thrombosis occurs in patients receiving extracorporeal life support such as extracorporeal membrane oxygenation and continuous renal replacement therapy. Nafamostat is applied to prevent circuit thrombosis, and the occurrence of this complication indicates that the prothrombotic artificial surface overwhelms the regional anticoagulant effect of the drug, rather than nafamostat promoting thrombosis formation.

The 2nd category is novel adverse reactions with biological plausibility: burning sensation. This neurological symptom is not listed in the current drug instructions, and multiple mechanisms can explain its occurrence. On the 1 hand, nafamostat promotes peripheral vasodilation via nitric oxide release, increasing local skin temperature.^[42,43] On the other hand, as a serine protease inhibitor, it modulates the activity of protease-activated receptors on nociceptive nerve endings.^[44,45] Relevant animal experiments have confirmed that nafamostat can alleviate visceral hypersensitivity by regulating the serine protease-activated receptor signaling pathway.^[44] Collectively, these mechanisms provide solid biological evidence for nafamostat-induced abnormal skin sensation. This signal requires further prospective clinical validation.

The 3rd category covers bleeding-related events: retroperitoneal hemorrhage and abdominal wall hematoma. Existing studies have reported these gastrointestinal and soft tissue bleeding complications in patients treated with nafamostat.^[46] Long-term immobilization and local muscle injury in critically ill patients may lead to lumbar artery rupture and subsequent retroperitoneal hemorrhage.^[46,47] Multiple comparative studies on anticoagulant regimens for extracorporeal membrane oxygenation have demonstrated that nafamostat is associated with a higher bleeding risk than heparin.^[48] Nevertheless, published systematic reviews show inconsistent conclusions regarding its bleeding and thrombosis risks,^[49] which may be related to heterogeneous dosing regimens and varied enrolled populations. Prospective studies focusing on high-risk populations are urgently needed to clarify this issue.

TTO analysis showed that the median onset time of nafamostat-related AEs was 1 day, with 58.55% of reactions occurring on the 1st day of medication and 82.63% occurring within 30 days. The WSP test yielded a β value of 0.325 (95% CI: 0.316–0.334), confirming that the risk of adverse reactions decreases gradually over time. These results clearly indicate that the highest risk period is the initial stage of administration. Clinicians must strengthen intensive monitoring within the 1st 24 hours after drug use, especially for life-threatening events such as anaphylactic shock, severe hyperkalemia, and acute hemorrhage.

Several limitations of this study must be acknowledged. First, as a retrospective analysis based on a spontaneous reporting database, the research is inevitably affected by inherent biases including underreporting, selective reporting, and incomplete clinical information. In addition, the causal relationship between nafamostat and AEs cannot be definitively verified merely through spontaneous reports. Second, the JADER database lacks detailed data on drug dosage, treatment duration, and combined medications. For this reason, we failed to conduct quantitative analysis on dose–response relationships or evaluate the risk of drug-drug interactions. Third, baseline conditions such as primary diseases, coagulation function, and renal function were not fully recorded, making it impossible to eliminate the interference of multiple confounding factors. Fourth, all data were derived from Japan, so the research conclusions may have limited generalizability to populations in other regions.

5. Conclusion

In conclusion, this pharmacovigilance study systematically analyzed the AE signals of nafamostat using real-world JADER data. We clarified known adverse reactions, interpreted newly detected signals, and distinguished false-positive results caused by biases. The findings provide evidence for standardizing clinical medication, strengthening early safety monitoring, and optimizing anticoagulant regimens for extracorporeal circulation. For clinical practice, we recommend enhanced surveillance within 24 hours after nafamostat administration, regular detection of serum potassium in elderly and renally impaired patients, and dynamic monitoring of coagulation indicators in patients receiving extracorporeal support. More prospective clinical studies are required in the future to further confirm the safety profile of nafamostat in diverse populations.

Author contributions

Conceptualization: Yi Yin.

Funding acquisition: Yi Yin.

Investigation: Yi Yin.

Supervision: Guiju Zhang.

Validation: Guiju Zhang.

Visualization: Guiju Zhang.