Sage Journals: Discover world-class research

Abstract

Objectives:

Ruxolitinib is used to treat myelofibrosis, polycythemia vera, and steroid-refractory graft-versus-host disease following allogeneic stem cell transplantation. This study aimed to determine the association between ruxolitinib and adverse events by evaluating case reports published between January 2014 and March 2024 in the Japanese Adverse Drug Event Report database.

Methods:

The signals for the ruxolitinib–adverse event association were identified using propensity score-adjusted reporting odds ratio analysis. Data obtained from the drug–gene interaction, drug signature, search tool for chemical interactions, and interaction reference index databases were used to construct a drug–gene interaction network. Functional and pathway enrichment analyses were performed using the Disease Ontology Semantic and Enrichment and ReactomePA R packages.

Results:

The propensity score-adjusted reporting odds ratio for ruxolitinib-associated adverse events was as follows: anemia, 18.49 (95% confidence interval (CI): 16.15–21.16); myelosuppression, 4.70 (95% CI: 3.54–6.24); pancytopenia, 1.97 (95% CI: 1.23–3.16); cardiac failure, 2.29 (95% CI: 1.60–3.28); hepatic function abnormal, 1.60 (95% CI: 1.15–2.23); herpes zoster, 6.40 (95% CI: 4.35–9.41); pneumonia, 2.96 (95% CI: 2.35–3.73); renal impairment, 1.34 (95% CI: 0.94–1.90); sepsis, 5.14 (95% CI: 3.75–7.05); interstitial lung disease, 0.33 (95% CI: 0.21–0.52); deep vein thrombosis, 0.32 (95% CI: 0.07–1.44); hemorrhage, 1.99 (95% CI: 1.05–3.75). We also assessed 3015 human genes that directly or indirectly interact with ruxolitinib. The molecular complex detection plug-in of Cytoscape was used to detect 24 clusters. Several genes were enriched in the biological processes of “anemia” and “bacterial infections,” identified as significant ruxolitinib-related disease terms.

Conclusions:

This retrospective analysis using the Japanese Adverse Drug Event Report database indicated potential associations between ruxolitinib and adverse events, including anemia and bacterial infections. Future research should explore the underlying pharmacological mechanisms using functional enrichment analysis of ruxolitinib-associated genes related to blood toxicity and bacterial infections.

Keywords

Ruxolitinib drug-interaction network adverse events Japanese Adverse Drug Event Report database

Introduction

Ruxolitinib, which primarily inhibits Janus kinase (JAK)1 and JAK2, is used to treat various diseases, including myelofibrosis (MF), polycythemia vera (PV), and steroid-refractory graft-versus-host disease following allogeneic stem cell transplantation.^1
–5 PV is a clonal myeloproliferative neoplasm attributed to mutations in the JAK2 gene and is primarily characterized by an increase in red blood cell mass.¹ MF is a progressive, chronic myeloproliferative neoplasm that adversely impacts the daily lives of patients.² In patients with MF, the levels of pro-inflammatory cytokines (signaling through JAK1 and JAK2) are elevated.²

JAK inhibitors are therapeutic agents targeting the cytokine signaling pathways, playing important roles in inflammatory diseases and immune system dysfunctions.⁵ The JAK family comprises four enzymes, namely, JAK1, JAK2, JAK3, and tyrosine kinase 2 (TYK2), involved in distinct cytokine signaling cascades. JAK1 primarily regulates inflammatory and innate immune responses; JAK2 is essential for erythropoiesis, myelopoiesis, and platelet production; JAK3 supports lymphocyte proliferation and homeostasis; TYK2 mediates antiviral responses.^5
–9

JAK inhibitors, including tofacitinib, baricitinib, and upadacitinib, are associated with various adverse events (AEs), including an increased risk of infection, thrombosis, liver dysfunction, and hematologic abnormalities.^10
–13 Similarly, ruxolitinib is linked to various AEs, such as an increased infection risk, hematologic abnormalities, and liver dysfunction; hematologic abnormalities occur more frequently than with other JAK inhibitors.^1
–5 Previous studies using the Spontaneous Reporting System (SRS) and bioinformatics have explored reproductive toxicology,¹⁴ digoxin-associated anemia,¹⁵ neuropsychiatric AEs of montelukast,¹⁶ and gentamicin-associated hearing loss.¹⁷ In this study, we examined ruxolitinib-associated AEs using SRS and identified genes responsible for characteristic hematological AEs, infections, and drug interactions using bioinformatics.

The pharmacological mechanisms underlying most AEs remain unclear. Most drugs exert their pharmacological effects and result in AEs through interactions with multiple proteins encoded by different genes. Analysis of drug–gene interactions has improved the understanding of drug toxicity. Integrated analysis using SRS (Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS), and Japanese Adverse Drug Event Report (JADER) database) data and drug–gene interactions can provide deep insights into AEs.^14
–17

In this study, we explored the ruxolitinib-associated disproportionality of AEs using propensity score-adjusted reporting odds ratio (PSW-ROR) analysis, leveraging data from the JADER database. Furthermore, we constructed a drug–gene interaction network using genes associated with ruxolitinib to gain insights into the toxicologi-cal mechanisms underlying ruxolitinib-related AEs, and conducted functional enrichment analysis of these genes to identify the potential toxicological mechanisms of ruxolitinib-related AEs.

Methods

Data source

Ruxolitinib was introduced in Japan in 2014; therefore, in this study, JADER data from January 2014 to March 2024 were downloaded from the Pharmaceuticals and Medical Devices Agency (PMDA) and the Japanese Regulatory Authority website (www.pmda.go.jp). All data from the JADER database were cleaned and anonymized in PMDA. The JADER database comprises the following four tables: patient demographic information (DEMO table), drug information (DRUG table), AEs (REAC table), and primary disease (HIST table). A relationship database was created using FileMaker Pro 17 Advanced software (FileMaker, Inc., Santa Clara, CA, USA) based on the American Standard Code for Information Interchange entity relationship diagram.¹⁸ Drug data included the following role codes assigned to each drug according to its association with AEs: “suspected,” “concomitant,” and “interacting drugs.” Data in the SRS database have been reported by healthcare professionals. Our analyses, restricted to the “suspected drugs,” provide the most robust results regarding the association of certain drugs with AEs.

Definition of AEs

AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA, www.meddra.org) version 23.1. To evaluate ruxolitinib-associated AEs, we employed system organ classes (SOCs) with the preferred terms (PTs) related to each SOC summarized in Table 1.

Table 1.

Crude ROR and inverse probability of treatment weighting-adjusted ROR of adverse events of ruxolitinib in the JADER from January 2014 to March 2024.

System organ class	Preferred term name	Total case (n)	Case (n)	Crude ROR (95% CI^a)	Adjusted ROR^b (95% CI)
Blood and lymphatic system disorders	Anemia	7962	349	16.87 (15.04–18.93)	12.53 (11.00–14.26)
	Myelosuppression	4467	52	3.86 (2.93–5.08)	3.58 (2.68–4.78)
	Pancytopenia	3621	14	1.25 (0.74–2.12)	1.56 (0.97–2.51)
Cardiac disorders	Atrial fibrillation	1621	4	0.80 (0.30–2.13)	0.62 (0.20–1.90)
Cardiac disorders	Cardiac failure	5443	35	2.10 (1.51–2.94)	1.84 (1.28–2.64)
Gastrointestinal disorders	Gastrointestinal hemorrhage	3023	15	1.61 (0.97–2.68)	0.98 (0.51–1.89)
Hepatobiliary disorders	Hepatic function abnormal	9178	34	1.20 (0.86–1.69)	1.21 (0.86–1.71)
	Liver disorder	4296	16	1.21 (0.74–1.98)	0.99 (0.58–1.72)
	Portal vein thrombosis	239	3	4.11 (1.31–12.83)	4.62 (1.56–13.7)
Infections and infestations	Bronchiolitis	347	5	4.73 (1.95–11.44)	2.71 (0.84–8.77)
	Bronchopulmonary aspergillosis	766	12	5.16 (2.91–9.13)	10.15 (6.69–15.39)
	Herpes zoster	1726	26	4.98 (3.37–7.34)	4.90 (3.29–7.28)
	Pneumonia	10,887	90	2.75 (2.23–3.40)	2.25 (1.78–2.84)
	Sepsis	3233	38	3.88 (2.81–5.35)	4.02 (2.92–5.53)
Investigations	Hemoglobin decreased	1990	123	22.25 (18.46–26.81)	18.18 (14.8–22.33)
	Lymphocyte count decreased	1580	33	6.96 (4.92–9.85)	8.31 (6.02–11.47)
	Neutrophil count decreased	9259	29	1.01 (0.70–1.46)	1.13 (0.79–1.60)
	Platelet count decreased	11,106	235	7.56 (6.60–8.65)	6.28 (5.42–7.28)
	White blood cell count decreased	5933	51	2.83 (2.15–3.74)	4.13 (3.26–5.23)
Metabolism and nutrition disorders	Hypertriglyceridaemia	227	3	4.33 (1.38–13.53)	5.09 (1.76–14.75)
Nervous system disorders	Cerebral hemorrhage	4200	13	1.00 (0.58–1.73)	0.74 (0.39–1.40)
	Dizziness	3292	2	0.20 (0.05–0.78)	0.18 (0.04–0.78)
	Neuropathy peripheral	3348	3	0.29 (0.09–0.90)	0.34 (0.12–0.98)
Renal and urinary disorders	Renal impairment	9624	39	1.32 (0.96–1.81)	1.05 (0.73–1.50)
Respiratory, thoracic and mediastinal disorders	Interstitial lung disease	21,952	24	0.35 (0.23–0.52)	0.26 (0.17–0.42)
Vascular disorders	Deep vein thrombosis	2108	2	0.31 (0.08–1.23)	0.26 (0.06–1.19)
Vascular disorders	Hemorrhage	1946	10	1.67 (0.90–3.11)	1.49 (0.76–2.89)

Confidence interval.

ROR was adjusted for confounding factors using inverse probability of treatment weighting based on propensity scores.

Signal detection

We used the crude ROR for administrative safety evaluation to examine the association between ruxolitinib and AEs. The crude ROR was used to calculate the odds of reporting an AE for a drug relative to all other AEs associated with that drug, compared to the odds for all other drugs in the JADER database.¹⁹

This ratio is derived from a two-by-two contingency table, and results are presented as point estimates along with the corresponding 95% confidence interval (CI). A signal was considered positive if the lower limit of the 95% CI exceeded one and there were two or more reports.¹⁹

Crude ROR calculations may be influenced by confounding factors related to differences in patient background. The crude ROR can be partially adjusted using multiple logistic regression analysis, offering the possibility of controlling for covariates.^20,21 Propensity score (PS) methods have been recently used with stabilized inverse probability of treatment weighting (stabilized IPTW) to adjust for potential confounders.^22
–26 We applied a PS method using stabilized IPTW to adjust for potential confounding factors in estimating adjusted RORs of AEs associated with ruxolitinib and refined the safety signal with a dedicated correction to detect possible confounding factors in the database. PS methods are statistical techniques employed to reduce confounding in observational studies by constructing matched pairs of treated and untreated individuals with similar PSs, thereby balancing covariates between groups. However, a primary limitation of PS matching is the potential reduction in sample size due to the exclusion of unmatched subjects.

In contrast, stabilized IPTW retains all subjects in the analysis by assigning weights based on their PSs. The stabilized weight for each individual is defined as:

w_{i} = P (T) / P (T | X)

(1)

where P(T) denotes the marginal probability of recei-ving the treatment, and P(T∣X) represents the PS conditional on covariates X. This approach mitigates the influence of extreme weights while preserving the full sample size, thereby enhancing statistical efficiency and reducing bias.

PS was estimated using a logistic regression model, in which the treatment status (treated or not treated with ruxolitinib) was regressed on the following covariates: sex, age category (<60 years vs ⩾60 years), and reporting year. The IPTW was calculated as follows:

For patients treated with ruxolitinib:

I P T W = (p r o p o r t i o n o f t r e a t e d p a t i e n t s) / P S

(2)

For patients not treated with ruxolitinib:

I P T W = (p r o p o r t i o n o f n o t t r e a t e d p a t i e n t s) / (1 - P S)

(3)

After applying stabilized IPTW, adjusted RORs were estimated using a weighted logistic regression model with IPTW as the frequency weight. ROR was derived from model coefficients, and 95% CIs were obtained using the standard error of the estimated coefficients.

All analyses were conducted using Python (version 3.12.3) (Python Software Foundation, Wilmington, DE, USA). The following Python libraries were utilized: Pandas (version 1.4.4) (Pandas Development Team via NumFOCUS, Austin, TX, USA) and NumPy (version 1.21.5) (NumPy Developers via NumFOCUS, Austin, TX, USA) for data processing, statsmodels (version 0.13.2) (statsmodels Development Team via NumFOCUS, Austin, TX, USA) for multiple logistic regression and generalized linear models, and scikit-learn (version 1.0.2) (scikit-learn Development Team via NumFOCUS, Austin, TX, USA) for PS estimation.

Drug–gene interaction network

The drug–gene interaction network was constructed using data from multiple sources: drug–gene interactions from the drug–gene interaction database (DGIdb, https://www.dgidb.org), drug signatures from the drug signatures database (DSigDB, http://dsigdb.tanlab.org), and broader interaction data from the “search tool for interactions of chemicals” (STITCH, https://stitch.embl.de). Genes indirectly associated with ruxolitinib were identified using iRefIndex 20.0 (“9606. mitab,” https://irefindex.vib.be).²⁷ The molecular complex detection (MCODE) algorithm (version 2.0.3) (Bader Lab, University of Toronto, Toronto, ON, Canada),²⁸ a plug-in in Cytoscape version 3.10.1 (Cytoscape Consortium, San Diego, CA, USA), was used to identify highly interconnected regions within protein–protein interaction networks,²⁹ with the following criteria: degree cutoff, 2; node score cutoff, 0.2; k-core, 2, and maximum depth from the seed, 100. Functional analysis and visualization of genes and gene clusters were performed using the R software (version 4.2.0) (R Foundation, Vienna, Austria). We used the Disease Ontology Semantic and Enrichment analysis (DOSE) package (YuLab-SMU, Southern Medical University, Guangzhou, China),³⁰ which implements methods for measuring semantic similarities between Disease Ontology terms and gene products. Accordingly, simple clusters, including triangles and squares, were excluded from the obtained cluster, and the remaining clusters were subjected to DOSE analysis conducted using the “enrichDO” function of the DOSE package (version 3.24.2) (YuLab-SMU, Southern Medical University, Guangzhou, China) in R with a p-value cutoff of 0.05, a p-value adjustment method of “BH,” and a q-value cutoff of 0.05. We performed a pathway enrichment analysis using ReactomePA,³¹ designed for Reactome pathway-based analysis, to investigate the pathways associated with gene clusters. Pathway analysis was performed using the “enrichPathway” function from the ReactomePA package (version 1.42.0) (YuLab-SMU, Southern Medical University, Guangzhou, China) with a p-value cutoff of 0.05.

Results

ROR analysis

The JADER database included 583,504 reports from January 2014 to March 2024. After excluding cases with missing age or sex information, 502,961 reports (86.2%) were included in the analysis. Within this dataset, ruxolitinib was reported as the suspected drug in 1285 cases, with 703 cases (54.7%) of male patients and 582 (45.3%) of female patients. Additionally, 179 cases (13.9%) were patients aged < 60 years, whereas 1106 cases (86.1%) were patients aged ⩾ 60 years. The crude RORs of AEs associated with ruxolitinib were as follows: anemia, 21.45 (95% CI: 20.19–22.79); myelosuppression, 4.82 (95% CI: 4.21–5.53); pancytopenia, 1.41 (95% CI: 1.11–1.81); cardiac failure, 2.90 (95% CI: 2.47–3.40); hepatic function abnormal, 1.02 (95% CI: 0.86–1.22); herpes zoster, 6.01 (95% CI: 4.98–7.25); pneumonia, 3.57 (95% CI: 3.22–3.96); renal impairment, 1.85 (95% CI: 1.59–2.14); sepsis, 4.17 (95% CI: 3.56–4.89); interstitial lung disease, 0.39 (95% CI: 0.32–0.48); deep vein thrombosis (DVT), 0.37 (95% CI: 0.18–0.76); hemorrhage, 2.48 (95% CI: 1.88–3.28; Table 1).

Standardized mean differences (SMDs) were calculated for sex, age category, and reporting year to evaluate covariate balance following IPTW application. SMD below 0.1 indicates an acceptable balance.²³

Following IPTW adjustment, SMD values were 0.123 for the reporting year, 0.029 for sex, and 0.046 for the age category. While the SMD for the reporting year was just over 0.1, we deemed this acceptable as the distribution was not significantly imbalanced visually or clinically.

In the initial analysis, unstabilized IPTW values had a mean of 1.980 (min: 1.000; max: 2110.505) and a median (Q1–Q3) of 1.002 (1.001–1.003), whereas stabilized IPTW values had a mean of 1.000 (min: 0.395; max: 5.392 and a median (Q1–Q3) of 1.000 (0.999–1.001). These stabilized weights improved the balance and stability of the analysis.

The PSW-ROR of AEs associated with ruxolitinib was as follows: anemia, 18.49 (95% CI: 16.15–21.16); myelosuppression, 4.70 (95% CI: 3.54–6.24); pancytopenia, 1.97 (95% CI: 1.23–3.16); cardiac failure, 2.29 (95% CI: 1.60–3.28); hepatic function abnormal, 1.60 (95% CI: 1.15–2.23); herpes zoster, 6.40 (95% CI: 4.35–9.41); pneumonia, 2.96 (95% CI: 2.35–3.73); renal impairment, 1.34 (95% CI: 0.94–1.90); sepsis, 5.14 (95% CI: 3.75–7.05); interstitial lung disease, 0.33 (95% CI: 0.21–0.52); DVT, 0.32 (95% CI: 0.07–1.44); hemorrhage, 1.99 (95% CI: 1.05–3.75; Table 1).

Drug–gene interaction network

Our initial search across DGIdb, DSigDB, and STITCH databases identified 14 seed genes that interacted with ruxolitinib, namely CRLF2, CSF3R, JAK1, JAK2, JAK3, MPL, PARP1, PARP2, PARP3, PARP4, PLAUR, SRC, TYK2, and VHL, all of which were from strain 9606. The mitab dataset in the iRefIndex was integrated into a network comprising 18,270 nodes (genes) and 791,934 edges (interactions).

Genes that directly or indirectly interacted with the 14 seed genes were integrated into a network of 3015 nodes and 159,496 edges. Using the MCODE plug-in, 24 clusters were identified in the network. To elucidate their biological significance, functional enrichment analysis of these clusters was performed using a pathway analysis. The top two clusters related to ruxolitinib are shown in Table 2 and Figure 1; the remaining 22 are shown in Supplemental Data (Supplemental material).

Table 2.

Clusters of networks analyzed by MCODE^a analysis on ruxolitinib-related genes.

Cluster	Score (Density * #Nodes)	Nodes	Edges	Node IDs
1	55.408	99	2715	ALYREF, CPSF2, CSTF3, CYLD, DDX23, DNAJC8, EFTUD2, EIF1AX, EIF2S2, EIF3C, EIF3D, EIF3K, GTF2F1, HNRNPA0, HNRNPH2, HNRNPU, HNRNPUL1, ILF3, LSM2, NCL, NUDT21, PABPC1, PCBP1, PCBP2, POLR2A, POLR2B, POLR2C, POLR2D, POLR2E, POLR2F, POLR2G, POLR2H, POLR2I, POLR2J, POLR2K, POLR2L, PPP2CA, PPP2R1A, PPP2R2A, PRPF6, PRPF8, RPL10, RPL10A, RPL11, RPL12, RPL13, RPL14, RPL18A, RPL19, RPL21, RPL22, RPL23A, RPL27, RPL27A, RPL29, RPL3, RPL35, RPL35A, RPL37A, RPL5, RPL6, RPL7A, RPS10, RPS11, RPS15A, RPS16, RPS19, RPS2, RPS23, RPS27, RPS27A, RPS29, RPS3, RPS3A, RPS5, RPS6, RPS8, RPS9, SF3A1, SF3A2, SF3B3, SF3B4, SF3B6, SMC1A, SNRNP200, SNRNP70, SNRPA1, SNRPB, SNRPD2, SNRPF, SRP72, SRRM1, SRSF2, STAU1, SUGP1, UBA52, UPF2, UPF3B, YBX1
2	19.025	241	2283	ABCE1, ABI1, ACTC1, ACTL6A, ACTR2, ACTR3, ADIPOQ, AIFM1, AIMP1, APLNR, APOBEC3C, APOBEC3G, APP, AQR, ARF6, ARHGAP24, ARRB1, ASXL2, ATP2A2, ATXN3, BAG3, BANF1, BRK1, BTC, BUB3, C3, CACYBP, CALM1, CAPZA1, CAPZA2, CBX1, CBX3, CCDC47, CCNC, CCR1, CCR2, CCR5, CD247, CD36, CD3G, CDKN1A, CDKN2A, CETN2, CFTR, COP1, CSE1L, CSF2, CSF2RA, CSF2RB, CSNK1A1, CTTN, CXCR4, CYFIP2, DCTN1, DDX55, DHX57, DHX8, DRD4, EGF, EIF5A, ELOVL5, EMD, ERBB3, EREG, ERH, ESR2, FAF1, FAM98A, FCGR1A, FCGR2A, FCGR3A, FIP1L1, FLII, GAB1, GATAD2A, GATAD2B, GJA1, GMNN, GRB2, GTF2B, GTF2E1, H2AC21, H2AC4, H2AZ1, H3-3A, H3C13, HADHA, HBEGF, HERC5, HMGA1, IGF2R, IK, IL3, IL3RA, IL5, IL5RA, IMMT, INPP5D, INPPL1, IPO5, ISG15, ITGB1, ITK, JAK1, KIF5B, KPNA1, LAMTOR1, LCP2, LGALS3BP, LOXL2, LSM12, MAGED2, MAU2, MBD3, MCM6, MED7, MET, MRPL23, MRPS14, MRPS15, MRPS22, MRPS23, MRPS33, MTA3, MYBL2, MYL12A, NCAPD2, NCK1, NEK4, NOLC1, NRG1, NRG2, NRG4, NVL, PDGFRB, PDIA3, PDPK1, PDS5A, PFDN1, PFDN4, PFDN5, PFDN6, PFN1, PGRMC1, PIK3CA, PIK3CD, PINX1, PLCG1, PLEC, PLK1, PNO1, POLA1, POLD1, POLE, POLR1A, POLR1C, PRDX1, PRDX3, PREB, PSIP1, PSMA2, PSMA6, PSMB1, PSMC3, PSMD12, PSMD13, PSMD2, PSME1, PTGES3, PTPN1, PUM1, PXN, RAB2A, RAB9A, RAD21, RANGAP1, RB1, RB1CC1, RBBP7, RCC2, RECQL5, RFC1, RFC2, RNF123, RRP15, RSF1, RXRA, S1PR1, S1PR2, SAFB2, SAP130, SAP30, SCAF1, SDAD1, SEC22B, SERPINH1, SHC2, SHC3, SIN3A, SIRT1, SIRT6, SLC25A6, SMARCA2, SMARCE1, SMC2, SMURF1, SMURF2, SNCA, SPTAN1, STAG2, STRAP, STUB1, SUMO2, SUPT5H, SUPT6H, SUZ12, SYK, TAF13, TBL3, TKT, TNPO1, TOE1, TP53BP1, TPM3, TPT1, TRIM24, TUBA3C, UBA1, URB1, USP14, UTP14A, VARS1, VAV1, WAS, WASL, WDR36, YTHDC2, YWHAB, YWHAH, YY1, ZNF330

Network scoring: Include loops = false, Degree cutoff = 2, Cluster finding: node score cutoff = 0.2, Haircut = true, Fluff = false, K-core = 2, Max depth from seed = 100.

Figure 1.

Gene interaction networks obtained using the molecular complex detection plug-in in cytoscape. (a) Cluster 1 (99 genes), (b) Cluster 2 (241 genes).

For Cluster 1, the genes interacting with ruxolitinib were enriched in pathways such as “L13a-mediated translational silencing of ceruloplasmin expression,” “nonsense-mediated decay (NMD),” “NMD enhanced by the exon junction complex (EJC),” “eukaryotic translation initiation,” “cap-dependent translation initiation,” and “formation of a pool of free 40S subunits” (Table 3 and Figure 2). Cluster 1 was enriched in anemia-associated genes, including RPL5, RPL11, RPL27, RPL35, RPL35A, RPS10, RPS19, RPS27, RPS29, and RPS15A (Table 4 and Figure 3).

Table 3.

Pathway enrichment analysis of ruxolitinib-associated gene clusters using the ReactomePA^a package.

Cluster	Pathway	Gene ratio	Adjusted p-value	Gene
1	Influenza viral RNA transcription and replication	51/97	<0.001	GTF2F1, POLR2A, POLR2B, POLR2C, POLR2D, POLR2E, POLR2F, POLR2G, POLR2H, POLR2I, POLR2J, POLR2K, POLR2L, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
	Influenza infection	51/97	<0.001	GTF2F1, POLR2A, POLR2B, POLR2C, POLR2D, POLR2E, POLR2F, POLR2G, POLR2H, POLR2I, POLR2J, POLR2K, POLR2L, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
	L13a-mediated translational silencing of ceruloplasmin expression	44/97	<0.001	EIF1AX, EIF3C, EIF3D, EIF3K, EIF2S2, PABPC1, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
	Nonsense-mediated decay (NMD)	44/97	<0.001	PABPC1, PPP2CA, PPP2R1A, PPP2R2A, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52, UPF2, UPF3B
	Nonsense-mediated decay (NMD) enhanced by the exon junction complex (EJC)	44/97	<0.001	PABPC1, PPP2CA, PPP2R1A, PPP2R2A, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52, UPF2, UPF3B
	Eukaryotic translation initiation	44/97	<0.001	EIF1AX, EIF3C, EIF3D, EIF3K, EIF2S2, PABPC1, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
	Cap-dependent translation initiation	44/97	<0.001	EIF1AX, EIF3C, EIF3D, EIF3K, EIF2S2, PABPC1, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
	Formation of a pool of free 40S subunits	42/97	<0.001	EIF1AX, EIF3C, EIF3D, EIF3K, RPL3, RPL5, RPL6, RPL10, RPL11, RPL12, RPL13, RPL14, RPL19, RPL21, RPL22, RPL27, RPL29, RPL35, RPL7A, RPL10A, RPL18A, RPL23A, RPL27A, RPL35A, RPL37A, RPS2, RPS3, RPS5, RPS6, RPS8, RPS9, RPS10, RPS11, RPS16, RPS19, RPS23, RPS27, RPS29, RPS3A, RPS15A, RPS27A, UBA52
2	PIP3 activates AKT signaling	34/214	<0.001	BTC, CDKN1A, EGF, ERBB3, EREG, ESR2, GAB1, GATAD2A, GATAD2B, GRB2, HBEGF, LAMTOR1, MBD3, MET, MTA3, NRG1, NRG2, NRG4, PDGFRB, PDPK1, PIK3CA, PIK3CD, PSMA2, PSMA6, PSMB1, PSMC3, PSMD2, PSMD12, PSMD13, PSME1, RBBP7, STUB1, SUZ12, VAV1
	Intracellular signaling by second messengers	36/214	<0.001	BTC, CALM1, CDKN1A, EGF, ERBB3, EREG, ESR2, GAB1, GATAD2A, GATAD2B, GRB2, HBEGF, LAMTOR1, MBD3, MET, MTA3, NRG1, NRG2, NRG4, PDGFRB, PDPK1, PIK3CA, PIK3CD, PLCG1, PSMA2, PSMA6, PSMB1, PSMC3, PSMD2, PSMD12, PSMD13, PSME1, RBBP7, STUB1, SUZ12, VAV1
	RAF/MAP kinase cascade	34/214	<0.001	ARRB1, BTC, CALM1, CSF2, CSF2RA, CSF2RB, EGF, ERBB3, EREG, GRB2, HBEGF, IL3, IL5, IL3RA, IL5RA, JAK1, MET, NRG1, NRG2, NRG4, PDGFRB, PIK3CA, PSMA2, PSMA6, PSMB1, PSMC3, PSMD2, PSMD12, PSMD13, PSME1, SHC2, SHC3, SPTAN1, YWHAB
	MAPK1/MAPK3 signaling	34/214	<0.001	ARRB1, BTC, CALM1, CSF2, CSF2RA, CSF2RB, EGF, ERBB3, EREG, GRB2, HBEGF, IL3, IL5, IL3RA, IL5RA, JAK1, MET, NRG1, NRG2, NRG4, PDGFRB, PIK3CA, PSMA2, PSMA6, PSMB1, PSMC3, PSMD2, PSMD12, PSMD13, PSME1, SHC2, SHC3, SPTAN1, YWHAB
	PI3K events in ERBB2 signaling	11/214	<0.001	BTC, EGF, ERBB3, EREG, GAB1, GRB2, HBEGF, NRG1, NRG2, NRG4, PIK3CA
	Interleukin receptor SHC signaling	13/214		CSF2, CSF2RA, CSF2RB, GRB2, IL3, IL5, IL3RA, IL5RA, INPP5D, INPPL1, JAK1, PIK3CA, PIK3CD
	MAPK family signaling cascades	34/214	<0.001	ARRB1, BTC, CALM1, CSF2, CSF2RA, CSF2RB, EGF, ERBB3, EREG, GRB2, HBEGF, IL3, IL5, IL3RA, IL5RA, JAK1, MET, NRG1, NRG2, NRG4, PDGFRB, PIK3CA, PSMA2, PSMA6, PSMB1, PSMC3, PSMD2, PSMD12, PSMD13, PSME1, SHC2, SHC3, SPTAN1, YWHAB
	Interleukin-3, interleukin-5 and GM-CSF signaling	15/214	<0.001	CSF2, CSF2RA, CSF2RB, GRB2, IL3, IL5, IL3RA, IL5RA, INPP5D, INPPL1, JAK1, PIK3CA, PIK3CD, SYK, VAV1

p-value cutoff of 0.05.

Figure 2.

Dot plot of the pathway analysis performed using the ReactomePA package. (a) Cluster 1 and (b) Cluster 2.

Table 4.

Disease ontology enrichment of ruxolitinib-related genes using the disease ontology semantic and enrichment^a package.

Cluster	DO ID	DO term	Gene ratio	Adjusted p-value	Gene
1	1339	Diamond-Blackfan anemia	10/67	<0.001	RPL5, RPL11, RPL27, RPL35, RPL35A, RPS10, RPS19, RPS27, RPS29, RPS15A
	1342	Congenital hypoplastic anemia	10/67	<0.001	RPL5, RPL11, RPL27, RPL35, RPL35A, RPS10, RPS19, RPS27, RPS29, RPS15A
	12,449	Aplastic anemia	11/67	<0.001	RPL5, RPL11, RPL27, RPL35, RPL35A, RPS6, RPS10, RPS19, RPS27, RPS29, RPS15A
	2355	Anemia	11/67	<0.001	RPL5, RPL11, RPL27, RPL35, RPL35A, RPS6, RPS10, RPS19, RPS27, RPS29, RPS15A
	5603	T-cell acute lymphoblastic leukemia	3/67	0.042	RPL5, RPL10, RPL11
	3492	Mixed connective tissue disease	2/67	0.047	SNRNP70, SNRPB
2	5520	Head and neck squamous cell carcinoma	16/191	0.001	ADIPOQ, ATP2A2, CDKN1A, CDKN2A, CTTN, CXCR4, ERBB3, FCGR3A, GJA1, HBEGF, ITGB1, MET, PIK3CA, RB1, RFC1, SMARCA2
	9169	Wiskott-Aldrich syndrome	5/191	0.001	ACTR2, ACTR3, NCK1, WAS, WASL
	1542	Head and neck carcinoma	18/191	0.001	ADIPOQ, ATP2A2, CCR5, CDKN1A, CDKN2A, CTTN, CXCR4, EGF, ERBB3, FCGR3A, GJA1, HBEGF, ITGB1, MET, PIK3CA, RB1, RFC1, SMARCA2
	11,934	Head and neck cancer	18/191	0.001	ADIPOQ, ATP2A2, CCR5, CDKN1A, CDKN2A, CTTN, CXCR4, EGF, ERBB3, FCGR3A, GJA1, HBEGF, ITGB1, MET, PIK3CA, RB1, RFC1, SMARCA2
	3376	Bone osteosarcoma	8/191	0.001	BUB3, CDKN1A, CDKN2A, IGF2R, ITGB1, MET, PXN, RB1
	612	Primary immunodeficiency disease	20/191	0.002	ABCE1, APOBEC3G, C3, CCR2, CCR5, CD247, CD3G, CXCR4, FCGR2A, FCGR3A, IL5, ISG15, ITK, LCP2, MET, PIK3CD, SYK, VAV1, WAS, YY1
	50,338	Primary bacterial infectious disease	18/191	0.002	ADIPOQ, ARF6, C3, CCR2, CCR5, CD36, CD247, CFTR, CSF2, CXCR4, EREG, FCGR1A, FCGR2A, HBEGF, IL5, ITGB1, NRG1, SNCA
	9500	Leukocyte disease	11/191	0.003	ATP2A2, CFTR, CSF2, CXCR4, FCGR2A, FCGR3A, FIP1L1, IL5, IL5RA, VAV1, WAS
	104	Bacterial infectious disease	19/191	0.008	ADIPOQ, ARF6, C3, CCR2, CCR5, CD36, CD247, CFTR, CSF2, CXCR4, EREG, FCGR1A, FCGR2A, HBEGF, IL5, ITGB1, NRG1, SMARCA2, SNCA
	80,524	Thyroid gland adenocarcinoma	11/191	0.008	CDKN1A, CDKN2A, CXCR4, EGF, IK, MET, PIK3CA, PRDX1, RB1, RXRA, TPM3

p-value cutoff of 0.05, p-value adjustment method of “BH,” and q-value cutoff of 0.05.

Figure 3.

Dot plot for disease ontology semantic and enrichment analysis using the disease ontology semantic and enrichment package. (a) Cluster 1 and (b) Cluster 2.

Cluster 2 was enriched in bacterial infection-associated genes, including interleukin-5 (IL-5), colony-stimulating factor 2 (CSF2, also known as GM-CSF), C-C motif chemokine receptor (CCR)2 and CCR5, Fc gamma receptor IIA (FCGR2A), and Fc gamma receptor IIIA (FCGR3A; Table 4 and Figure 3). In Cluster 2, the genes were enriched in various pathways including “PIP3 activates AKT signaling,” “intracellular signaling by second messengers, RAF/MAP kinase cascade, MAPK1/MAPK3 signaling, PI3K events in ERBB2 signaling,” “interleukin receptor SHC signaling, MAPK family signaling cascades,” and “interleukin-3, interleukin-5, and GM-CSF signaling” (Table 3 and Figure 2).

Data from the DOSE package were used to characterize diseases associated with ruxolitinib. Genes involved in anemia (adjusted p-value = 0.0003) in Cluster 1 and primary bacterial infectious disease (adjusted p-value = 0.0022) in Cluster 2 were significantly enriched (Table 4 and Figure 3).

Discussion

In this study, we propose a methodology to apply safety signals—specifically, disproportionality indices such as RORs, which are widely used in pharmacovigilance—to bioinformatics to identify genes potentially associated with AEs. AEs are associated with the pharmacological effects of drugs; therefore, these genes can potentially serve as therapeutic targets for managing AEs. This will help to facilitate the development of novel treatments or broaden the application of existing drugs. This study presents a practical approach within the exploratory domain of translational research and partially addresses the limitations of crude ROR by integrating PS methods and multiple logistic regression analysis. To the best of our knowledge, this is the first study to adjust RORs using PSW through IPTW, a novel contribution to the field.

Ruxolitinib is a JAK inhibitor used widely to treat MF and PV. However, its associated AEs may limit ruxolitinib’s therapeutic efficacy. Reports from the JADER database have linked ruxolitinib to major AEs. The lower limits of the 95% CIs of crude RORs related to the SOC of “blood and lymphatic system disorders,” “infections and infestations,” and “investigations” were >1, and the signal was significant.

In Phase 3 studies of PV and MF, the most frequent AEs leading to dose adjustment or interruption of ruxolitinib treatment were anemia and thrombocytopenia.^1,2 Additionally, many patients with MF develop anemia during the disease and require red blood cell transfusion.

Conversely, bleeding episodes, attributed to either thrombocytopenia or qualitative platelet dysfunction, can complicate the clinical course of MF.³² In the JADER database, significant signals were detected in the RORs associated with gastrointestinal hemorrhage and hemorrhage. These findings necessitate the careful management of hematopoietic disorder risks in patients receiving ruxolitinib therapy. Infections, such as pneumonia, sepsis, and herpes zoster, are also frequently reported common AEs associated with JAK inhibitors besides ruxolitinib, including tofacitinib and baricitinib, which are used for chronic inflammatory diseases.^10
–13 Since the immunosuppressive effects of ruxolitinib can increase the risk of infection, infection control measures and early therapeutic intervention are essential.

Owing to the growing concern regarding thromboembolic events in patients undergoing treatment with JAK inhibitors, the FDA has mandated a “boxed warning” for most of these drugs currently available.³³ However, ruxolitinib lacks such a mandate. Setyawan et al. have detected a signal for portal vein thrombosis (PVT) in the FAERS database,³⁴ consistent with our findings from the JADER analysis. In contrast, no signal was detected for DVT in either study. Given the observed overreporting of PVT, the signal of which is based on three cases with possible concerns about random noise, associated with ruxolitinib in the present study, additional analyses are required. In the JADER dataset, including reports from January 2013 to March 2024, PVT was reported only in patients treated with ruxolitinib, not in those receiving other JAK inhibitors (tofacitinib, peficitinib, baricitinib, filgotinib, and upadacitinib; Supplemental Table 4). This is due to potential confounding by indication (e.g., PVT risk in myeloproliferative neoplasms). A direct comparison and analysis of PVT risk among JAK inhibitors was not feasible. Stratified analysis (e.g., PVT in ruxolitinib vs non-ruxolitinib-treated MF/PV patients) or comparison of PVT incidence with relevant epidemiological data will provide deeper insights.

A comprehensive analysis of thrombosis risk among JAK inhibitors using JADER suggested older age (>60 years) as a risk factor for more extensive embolic and thrombotic events across all JAK inhibitors examined.³⁵

JAK inhibitors are associated with increased risk of infection, thrombosis, liver dysfunction, and hematologic abnormalities, among others.^10
–13 The results of this disproportionality analysis are consistent with those previously reported.^10
–13 Moreover, gene enrichments in the biological processes of anemia and bacterial infection identified in this study correspond to the pharmacological effects of ruxolitinib.

Integrated analysis using FAERS data and drug–gene interaction analysis data provides novel insights into AEs.^14
–16 Previously, we employed analogous methodologies to analyze montelukast and identified 1144 interacting human genes, generating 14 clusters, which included genes enriched in mood disorders.¹⁶ Lin et al.¹⁵ conducted a drug–gene interaction network analysis similar to our study, focusing on digoxin-associated anemia using FAERS. Wu et al.¹⁴ integrated FAERS data with drug–gene network analysis to investigate variations in the reproductive toxicity of alopecia drugs. In this study, we employed the aforementioned comparable method,^14
–16 wherein drug–gene interactions were queried in public databases to elucidate the toxicological mechanisms underlying ruxolitinib-related AEs. Specifically, 3015 human genes interacting with ruxolitinib were identified. We generated 24 clusters using the MCODE software (Bader Lab, University of Toronto, Toronto, ON, Canada). The results yielded two clusters that included diseases likely associated with ruxolitinib.

Cluster 1 was enriched with anemia-associated genes, including RPL5, RPL11, RPL27, RPL35, RPL35A, RPS10, RPS19, RPS27, RPS29, and RPS15A (Table 4 and Figure 3). Pathways related to intracellular protein synthesis, primarily in the endoplasmic reticulum, were identified and located downstream of JAK. Ruxolitinib-associated anemia likely occurs due to JAK2-mediated inhibition of erythropoietin (EPO) signaling in red blood cell progenitor cells. The EPO binds to the EPO receptor, activating JAK2.^36,37 This, in turn, phosphorylates STAT5, promoting the proliferation and differentiation of erythroid progenitor cells.^38,39 Furthermore, ruxolitinib reduces the allele burden of JAK2 V617F mutations and suppresses the autonomous proliferation of EPO-independent erythroid progenitor cells.^40,41 A somatic mutation in JAK2 V617F is found in >90% of patients with PV and half of those with MF.^42
–45 ROR analysis in JADER detected a signal for ruxolitinib-associated red blood cell depletion. Our functional enrichment analysis suggests that pathways related to intracellular protein synthesis in the endoplasmic reticulum indirectly regulate AEs related to ruxolitinib, even if the gene is not a direct drug target.

Cluster 2 was enriched in genes involved in bacterial infections and malignancy, including IL-5, CSF2, CCR2, CCR5, FCGR2A, and FCGR3A. Activation of eosinophils by IL-5 is crucial for immune responses, such as those elicited in allergic diseases and parasitic infections.⁴⁶ IL-5 binds to the IL-5 receptor to activate the JAK2-mediated signaling pathway and regulates eosinophil differentiation, proliferation, survival, and activation.⁴⁷ CCR2, a G protein-coupled receptor, is primarily expressed on monocytes and T cells and induces their migration to sites of inflammation.⁴⁸ The signaling pathways of CCR5, another G protein-coupled receptor, involve phosphatidylinositol 3-kinase/protein kinase B and JAK/signal transducer and activator of transcription, key in inflammatory responses and immune regulation.⁴⁹

However, CCR2 and CCR5 may show cross-talk with the JAK-STAT pathway.^49,50 FCGR2A, expressed on macrophages and neutrophils, induces phagocytosis and antibody-dependent cellular cytotoxicity by binding to IgG antibodies.⁵¹ Meanwhile, FCGR3A, expressed on natural killer cells (NK) and macrophages, activates cell-mediated immune responses via IgG binding.⁵¹ JAK inhibition increases the risk of infection because of its mechanism of action.

Signals for pneumonia, sepsis, and shingles have been detected in the JADER database. Pneumonia is a frequently reported AE associated with JAK inhibitors.^9
–13 JAK inhibition suppresses IL-6 and IL-2 signaling, decreasing the activity of T cells and NK and increasing the risk of infection.^52,53 While JAK inhibition increases the risk of infection through immunosuppression, the source of the increased risk of infection includes the effects of underlying diseases and concomitant medications. JAK inhibitors are suspected to increase the risk of malignant tumor development, although the data remain inconclusive.^32,54

Limitation

This study, which is based on spontaneous reports obtained from the JADER database, has some limitations. First, potential biases might have been introduced due to over-reporting, under-reporting, and missing data. Second, external factors, such as safety communications and market dynamics, might have influenced the reporting patterns. Finally, given that the JADER dataset contains limited patient and clinical data (e.g., comorbidities and concomitant medications), our findings should not be interpreted with caution. Spontaneous reports of AEs, such as “anemia” in this study, may be overrepresented due to clinicians’ heightened awareness of hematologic risks. Additional biases include notoriety bias, wherein the number of spontaneous reports increases following safety alerts issued by regulatory authorities.^55
–57 Such warnings may increase RORs. Another relevant phenomenon is the Weber effect,^58
–60 which describes the tendency for spontaneous reporting of AEs to sharply increase following a drug’s approval, subsequently plateau, and eventually decline over time; however, this effect is not consistently observed across all drugs.⁶⁰ To mitigate these biases, PS methods can be employed to balance groups based on potential confounding variables.^61,62 Furthermore, several studies have attempted to adjust ROR estimates by incorporating factors such as reporting year, age, and sex into the model using multivariate logistic regression analyses.^63
–65

In Japanese health insurance, individuals ⩾65 years or older are classified as elderly; this age classification is also seen in the package insert for ruxolitinib.⁶⁶ However, in the JADER database, age information has been modified and published in 10-year increments by the PMDA to protect personal information, making it difficult to accurately classify individuals at exactly 65 years of age. Defining the elderly population as those aged ⩾ 70 years may be more appropriate for ruxolitinib, given its characteristics. Nevertheless, in this study, individuals ⩾60 years old were classified and analyzed as elderly.

The ROR is merely an indicator of disproportionality and only identifies statistical associations, not causal relationships, and does not establish causation nor account for actual exposure rates. In this study, given the limited individuals’ background information (e.g., concomitant medications, underlying diseases, and patient characteristics), we calculated a partially adjusted PSW-ROR by controlling for age, sex, and reporting year.

Using only these three covariates may not fully control for all potential confounders. The models for estimating PS and calculating IPTW inherently involve estimation error and uncertainty; therefore, caution is essential to avoid overinterpretation of disproportional signals. Alternately, some approaches have been proposed to address these issues in a high-dimensional context.^67
–69 It may be possible to quantify the potential impact of reporting bias using sensitivity analysis or external validation (e.g., comparing JADER signals to FAERS or published clinical trial data). Future investigations with prospective study designs or multivariable analyses with comprehensive covariate data should validate these findings and clarify the risk profiles of ruxolitinib and other JAK inhibitors.

To date, the drug–gene interactions previously examined have not been validated in experimental models or in vitro or in vivo experiments due to our limited knowledge of disease-related proteins and their interactions. Clusters 1 and 2 may explain the drug’s mechanism of action. These might not necessarily be off-target, unexpected AEs. In other words, the drug is effective, and dose-effect relationships should be assessed, rather than addressed by drug–gene interaction analysis. Therefore, the association between ruxolitinib and this gene must be experimentally confirmed.

Initially, we identified 14 proteins that interacted with ruxolitinib; however, the seed genes suggested by databases, such as DGIdb, used in this study are not direct pharmacological targets, and many may be indirectly affected by the drug. We utilized protein-protein interaction networks to integrate these 14 affected proteins into a larger network (iRefIndex). Our findings revealed that the network was enriched in genes associated with anemia and bacterial infection. Specifically, RPL5, RPL11, RPL27, RPL35, RPL35A, RPS10, RPS19, RPS27, RPS29, and RPS15A were associated with anemia, and IL5, CSF2, CCR2, CCR5, FCGR2A, and FCGR3A with bacterial infections. Furthermore, even within small clusters, we observed specific associations. For cerebrovascular disease, a cluster containing four genes included three associated genes: COL4A1, ELANE, and MMP12. Although these genes are not direct targets of ruxolitinib, future studies are necessary to determine whether specific “hotspots” within such protein-protein interaction networks can distinguish between drugs associated with anemia and those that lead to bacterial infections.

The drug–gene network analysis used in this study relies on public databases, such as DGIdb and STITCH, which may include non-causal relationships, limiting conclusions concerning mechanisms without experimental validation. Nevertheless, this approach, which uses gene enrichment, provides novel insights into ruxolitinib-associated AEs and will facilitate future research. Given the nature of exploratory network analysis, over-interpretation of clusters as direct toxicity mechanisms should be avoided, and functional validation, including in vitro and in vivo, is required to clarify gene and pathway associations.

Conclusion

Ruxolitinib is associated with significant AEs that require careful consideration. Our retrospective analysis of the JADER database suggested anemia, likely due to the suppression of erythropoiesis, as a major AE, underscoring the importance of vigilant management in patients. Pneumonia, sepsis, and other infections linked to the immunosuppressive effects of the drug have been frequently reported. Genetic analysis revealed that genes potentially influencing anemia and infection through interactions with ruxolitinib, including those related to the endoplasmic reticulum and immune response genes, such as IL5, CSF2, CCR2, CCR5, FCGR2A, and FCGR3A, were enriched. Due to the limitations of spontaneous JADER reports, such as reporting biases, incomplete patient information, and unverified drug–gene interactions, the findings should be interpreted cautiously. Further studies for validation and risk mitigation remain warranted.

Supplemental Material

sj-jpg-1-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-jpg-1-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-jpg-2-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-jpg-2-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-jpg-3-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-jpg-3-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-jpg-4-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-jpg-4-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-10-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-10-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-11-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-11-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-12-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-12-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-13-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-13-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-5-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-5-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-6-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-6-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-7-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-7-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-8-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-8-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Supplemental Material

sj-xlsx-9-smo-10.1177_20503121251348420 – Supplemental material for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks

Supplemental material, sj-xlsx-9-smo-10.1177_20503121251348420 for Analysis of adverse events with ruxolitinib using real-world datasets and drug-interaction networks by Hideyuki Tanaka, Mika Maezawa, Mizuki Tanaka, Ryogo Umetsu, Sakiko Hirofuji, Koumi Miyasaka, Satoshi Nakao, Yuka Nokura, Moe Yamashita, Nanaka Ichihara, Kana Sugishita, Tomofumi Yamazaki, Kohei Shiota, Hirofumi Tamaki, Kazuhiro Iguchi and Mitsuhiro Nakamura in SAGE Open Medicine

Footnotes

Acknowledgements

Not applicable.

ORCID iDs

Mika Maezawa

Satoshi Nakao

Hirofumi Tamaki

Mitsuhiro Nakamura

Ethical considerations

Ethical approval was not sought for this study because the study was a database-related observational study and did not require the direct involvement of any research subjects. All results were obtained from data openly available online from the Pharmaceuticals and Medical Devices Agency website (). All the data from the Japanese Adverse Drug Event Report database were fully anonymized by the relevant regulatory authority before access was obtained.

Author contributions

H.T., M.M., M.T., R.U., and M.N. contributed to the overall study concept and design. H.T., M.M., S.H., K.M, S.N., Y.N., and M.N. performed data extraction and statistical analysis. H.T., M.M., and M.N. wrote the manuscript. M.T., R.U., S.H., K.M., S.N., Y.N., M.Y., N.I., K.S., T.Y., K.S., H.T., and K.I. revised the article to provide meaningful intellectual content. All authors have read and reviewed the manuscript.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was partially supported by JSPS KAKENHI (grant numbers 21K06646 and 21K11100). The funders had no role in the study design, data collection, analysis, decision to publish, or preparation of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.

Supplemental material

Supplemental material for this article is available online.

References

Kiladjian

Zachee

Hino

, et al. Long-term efficacy and safety of ruxolitinib versus best available therapy in polycythaemia vera (RESPONSE): 5-year follow up of a phase 3 study. Lancet Haematol 2020; 7(3): e226–e237.

Verstovsek

Mesa

Gotlib

, et al. Efficacy, safety, and survival with ruxolitinib in patients with myelofibrosis: results of a median 3-year follow-up of COMFORT-I. Haematologica 2015; 100(4): 479–488.

Fan

Huo

Yang

, et al. Efficacy and safety of ruxolitinib in steroid-refractory graft-versus-host disease: a meta-analysis. Front Immunol 2022; 13: 954268.

Verstovsek

Mesa

Gotlib

, et al. A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med 2012; 366(9): 799–807.

Verstovsek

Mesa

Gotlib

, et al. Long-term treatment with ruxolitinib for patients with myelofibrosis: 5-year update from the randomized, double-blind, placebo-controlled, phase 3 COMFORT-I trial. J Hematol Oncol 2017; 10(1): 55.

Traves

Murray

Campigotto

, et al. JAK selectivity and the implications for clinical inhibition of pharmacodynamic cytokine signalling by filgotinib, upadacitinib, tofacitinib and baricitinib. Ann Rheum Dis 2021; 80(7): 865–875.

Hammarén

Virtanen

Raivola

, et al. The regulation of JAKs in cytokine signaling and its breakdown in disease. Cytokine 2019; 118: 48–63.

Morris

Kershaw

Babon

JJ.

The molecular details of cytokine signaling via the JAK/STAT pathway. Protein Sci 2018; 27(12): 1984–2009.

O’Shea

Kontzias

Yamaoka

, et al. Janus kinase inhibitors in autoimmune diseases. Ann Rheum Dis 2013; 72(Suppl 2): ii111–ii115.

10.

Cohen

Tanaka

Mariette

, et al. Long-term safety of tofacitinib up to 9.5 years: a comprehensive integrated analysis of the rheumatoid arthritis clinical development programme. RMD Open 2020; 6(3): e001395.

11.

Taylor

Takeuchi

Burmester

, et al. Safety of baricitinib for the treatment of rheumatoid arthritis over a median of 4.6 and up to 9.3 years of treatment: final results from long-term extension study and integrated database. Ann Rheum Dis 2022; 81(3): 335–343.

12.

Harrington

Harkins

Conway

Janus kinase inhibitors in rheumatoid arthritis: an update on the efficacy and safety of tofacitinib, baricitinib and upadacitinib. J Clin Med 2023; 12(20): 6690.

13.

Cohen

van Vollenhoven

Winthrop

, et al. Safety profile of upadacitinib in rheumatoid arthritis: integrated analysis from the SELECT phase III clinical programme. Ann Rheum Dis 2021; 80(3): 304–311.

14.

Differences in reproductive toxicology between alopecia drugs: an analysis on adverse events among female and male cases. Oncotarget 2016; 7(50): 82074–82084.

15.

Lin

Feng

, et al. Digoxin-induced anemia among patients with atrial fibrillation and heart failure: clinical data analysis and drug–gene interaction network. Oncotarget 2017; 8(34): 57003–57011.

16.

Umetsu

Tanaka

Nakayama

, et al. Neuropsychiatric adverse events of montelukast: an analysis of real-world datasets and drug–gene interaction network. Front Pharmacol 2021; 12: 764279.

17.

Tanaka

Matsumoto

Satake

, et al. Gentamicin-induced hearing loss: a retrospective study using the Food and Drug Administration adverse event reporting system and a toxicological study using drug–gene network analysis. Heliyon 2021; 7(7): e07429.

18.

The pharmaceuticals and medical devices agency, Entity Relation (ER) diagram for each table in JADER, https://www.pmda.go.jp/safety/info-services/drugs/adr-info/suspected-adr/0004.html (2025, accessed 4 May 2025).

19.

Poluzzi

Raschi

Piccinni

, et al. Data mining techniques in pharmacovigilance: analysis of the publicly accessible FDA adverse event reporting system (AERS). In: Karahoca

(ed.) Data mining applications in engineering and medicine. London, UK: IntechOpen, 2012, pp. 265–302.

20.

Abe

Umetsu

Uranishi

, et al. Analysis of polypharmacy effects in older patients using Japanese Adverse Drug Event Report database. PLoS One 2017; 12(12): e0190102.

21.

Nakao

Hasegawa

Shimada

, et al. Evaluation of anti-infective-related clostridium difficile-associated colitis using the Japanese Adverse Drug Event Report database. Int J Med Sci 2020; 17(7): 921–930.

22.

Rosenbaum

Rubin

DB.

The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70(1): 41–55.

23.

Austin

PC.

An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46(3): 399–424.

24.

Robins

Hernán

Brumback

Marginal structural models and causal inference in epidemiology. Epidemiology 2000; 11(5): 550–560.

25.

Ross

Raebel

, et al. Use of stabilized inverse propensity scores as weights to directly estimate relative risk and its confidence intervals. Value Health 2010; 13(2): 273–277.

26.

Cole

Hernán

MA.

Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008; 168(6): 656–664.

27.

Razick

Magklaras

Donaldson

IM.

iRefIndex: a consolidated protein interaction database with provenance. BMC Bioinform 2008; 9: 405.

28.

Bader

Hogue

CW.

An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinform 2003; 4: 2.

29.

Shannon

Markiel

Ozier

, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003; 13(11): 2498–2504.

30.

Wang

Yan

, et al. DOSE: an R/Bioconductor package for disease ontology semantic and enrichment analysis. Bioinformatics 2015; 31(4): 608–609.

31.

QY.

ReactomePA: an R/Bioconductor package for reactome pathway analysis and visualization. Mol Biosyst 2016; 12(2): 477–479.

32.

Kremyanskaya

Atallah

Hoffman

, et al. Clarifying the use of ruxolitinib in patients with myelofibrosis. Oncology (Williston Park) 2013; 27(7): 706–714.

33.

An Official Website of U.S. Food and Drug Administration. FDA requires warnings about increased risk of serious heart-related events, cancer, blood clots, and death for JAK inhibitors that treat certain chronic inflammatory conditions, https://www.fda.gov/drugs/fda-drug-safety-podcasts/fda-requires-warnings-about-increased-risk-serious-heart-related-events-cancer-blood-clots-and-death (2022, accessed 9 November 2024).

34.

Setyawan

Azimi

Strand

, et al. Reporting of thromboembolic events with JAK inhibitors: analysis of the FAERS database 2010–2019. Drug Saf 2021; 44(8): 889–897.

35.

Tanaka

Maezawa

Nakao

, et al. Analysis of adverse events with Janus Kinase inhibitors reported to spontaneous reporting system. Pharm Biomed Res 2024; 10(3): 203–228.

36.

Bittorf

Jaster

Lüdtke

, et al. Requirement for JAK2 in erythropoietin-induced signalling pathways. Cell Signal 1997; 9(1): 85–89.

37.

Abraham

Haikarainen

Vuorio

, et al. Molecular basis of JAK2 activation in erythropoietin receptor and pathogenic JAK2 signaling. Sci Adv 2024; 10(10): eadl2097.

38.

Tóthová

Tomc

Debeljak

, et al. STAT5 as a key protein of erythropoietin signalization. Int J Mol Sci 2021; 22(13): 7109.

39.

Țichil

Mitre

Zdrenghea

, et al. A review of key regulators of steady-state and ineffective erythropoiesis. J Clin Med 2024; 13(9): 2585.

40.

Vannucchi

Verstovsek

Guglielmelli

, et al. Ruxolitinib reduces JAK2 p.V617F allele burden in patients with polycythemia vera enrolled in the RESPONSE study. Ann Hematol 2017; 96(7): 1113–1120.

41.

Deininger

Radich

Burn

, et al. The effect of long-term ruxolitinib treatment on JAK2p.V617F allele burden in patients with myelofibrosis. Blood 2015; 126(13): 1551–1554.

42.

Shi

Yuan

, et al. JAK2 V617F stimulates proliferation of erythropoietin-dependent erythroid progenitors and delays their differentiation by activating Stat1 and other nonerythroid signaling pathways. Exp Hematol 2016; 44(11): 1044–1058.e5.

43.

Baxter

Scott

Campbell

, et al. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet 2005; 365(9464): 1054–1061.

44.

James

Ugo

Le Couédic

, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature 2005; 434(7037): 1144–1148.

45.

Kralovics

Passamonti

Buser

, et al. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med 2005; 352(17): 1779–1790.

46.

Ikutani

Yanagibashi

Ogasawara

, et al. Identification of innate IL-5-producing cells and their role in lung eosinophil regulation and antitumor immunity. J Immunol 2012; 188(2): 703–713.

47.

Ogata

Kouro

Yamada

, et al. JAK2 and JAK1 constitutively associate with an interleukin-5 (IL-5) receptor alpha and betac subunit, respectively, and are activated upon IL-5 stimulation. Blood 1998; 91(7): 2264–2271.

48.

She

Ren

Chen

, et al. Functional roles of chemokine receptor CCR2 and its ligands in liver disease. Front Immunol 2022; 13: 812431.

49.

Wang

Yukselten

Nuwagaba

, et al. JAK/STAT signaling pathway affects CCR5 expression in human CD4+ T cells. Sci Adv 2024; 10(12): eadl0368.

50.

Kok

Hong

Kuo

, et al. Oncostatin M-induced CCL2 transcription in osteoblastic cells is mediated by multiple levels of STAT-1 and STAT-3 signaling: an implication for the pathogenesis of arthritis. Arthritis Rheum 2009; 60(5): 1451–1462.

51.

Junker

Gordon

Qureshi

Fc gamma receptors and their role in antigen uptake, presentation, and T cell activation. Front Immunol 2020; 11: 1393.

52.

Muhammad

Fan

Hai

, et al. Reigniting hope in cancer treatment: the promise and pitfalls of IL-2 and IL-2R targeting strategies. Mol Cancer 2023; 22: 121.

53.

Xue

Yao

, et al. Evolving cognition of the JAK-STAT signaling pathway: autoimmune disorders and cancer. Sig Transduct Target Ther 2023; 8: 204.

54.

Russell

Stovin

Alveyn

, et al. JAK inhibitors and the risk of malignancy: a meta-analysis across disease indications. Ann Rheum Dis 2023; 82(8): 1059–1067.

55.

Pariente

Gregoire

Fourrier-Reglat

, et al. Impact of safety alerts on measures of disproportionality in spontaneous reporting databases. Drug Saf 2007: 30(10): 891–898.

56.

De Bruin

van Puijenbroek

Egberts

ACG

, et al. Non-sedating antihistamine drugs and cardiac arrhythmias – biased risk estimates from spontaneous reporting systems? Br J Clin Pharmacol 2002; 53(4): 370–374.

57.

Motola

Vargiu

Leone

, et al. Influence of regulatory measures on the rate of spontaneous adverse drug reaction reporting in Italy. Drug Saf 2008; 31(7): 609–616.

58.

Wallenstein

Fife

Temporal patterns of NSAID spontaneous adverse event reports. Drug Saf 2001; 24(3): 233–237.

59.

Hartnell

Wilson

JP.

Replication of the Weber effect using postmarketing adverse event reports voluntarily submitted to the United States Food and Drug Administration. Pharmacotherapy 2004; 24(6): 743–749.

60.

McAdams

Governale

Swartz

, et al. Identifying patterns of adverse event reporting for four members of the angiotensin II receptor blockers class of drugs: revisiting the Weber effect. Pharmacoepidemiol Drug Saf 2008; 17(9): 882–889.

61.

Nakao

Hasegawa

Umetsu

, et al. Pharmacovigilance study of anti-infective-related acute kidney injury using the Japanese Adverse Drug Event Report database. BMC Pharmacol Toxicol 2021; 22(1): 47.

62.

Wang

Lang

Wang

, et al. Propensity score-adjusted three-component mixture model for drug-drug interaction data mining in FDA Adverse Event Reporting System. Stat Med 2020; 39(7): 996–1010.

63.

van Puijenbroek

Egberts

Meyboom

, et al. Signalling possible drug-drug interactions in a spontaneous reporting system: delay of withdrawal bleeding during concomitant use of oral contraceptives and itraconazole. Br J Clin Pharmacol 1999; 47(6): 689–693.

64.

Suzuki

Umetsu

, et al. Analysis of the interaction between clopidogrel, aspirin, and proton pump inhibitors using the FDA Adverse Event Reporting System database. Biol Pharm Bull 2015; 38(5): 680–686.

65.

Takeyama

Sai

Imatoh

, et al. Influence of Japanese Regulatory Action on denosumab-related hypocalcemia using Japanese Adverse Drug Event Report database. Biol Pharm Bull 2017; 40(9): 1447–1453.

66.

Novarits. JAKAVI® tablet, oral solution. Package insert, https://www.pmda.go.jp/PmdaSearch/iyakuDetail/ResultDataSetPDF/300242_4291034F1029_1_13 (2025, accessed 4 May 2025).

67.

Schuemie

Ryan

Hripcsak

, et al. A systematic approach to improving the reliability and scale of evidence from health care data. arVix, https://arxiv.org/pdf/1803.10791.pdf (2018, accessed 4 May 2025).

68.

Tian

Schuemie

Suchard

MA.

Evaluating large-scale propensity score performance through real-world and synthetic data experiments. Int J Epidemiol 2018; 47(6): 2005–2014.

69.

Hripcsak

Ryan

Duke

, et al. Characterizing treatment pathways at scale using the OHDSI network. Proc Natl Acad Sci USA 2016; 113(27): 7329–7336.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB

0.50 MB

0.29 MB

0.30 MB

0.01 MB

0.02 MB

0.01 MB

0.02 MB