Analysis of anticancer drug associated adverse reactions in depressive patients from vigibase Authors: Basile Chretien, Charles Dolladille, Kazuki Nishida, Branko Aleksic, Joachim Alexandre, Jean-Matthieu L’orphelin Published: 2025 ABSTRACT AbstractDepressive symptomatology is prevalent among oncology patients and may influence the perception of adverse drug reactions (ADRs) to anticancer drugs. Understanding this relationship is crucial for optimizing treatment strategies and improving ADR management. This pharmacovigilance study analyzed VigiBase® data from 1968 (year of its establishment) to 2024. Focusing on reports involving antineoplastic or immunomodulating agents, a case/non-case disproportionality analysis was conducted to assess the reporting and seriousness of ADRs in patients receiving antidepressants. Multivariable analyses were performed for all ADRs and the most frequently reported ADRs. Reporting Odds Ratios (RORs) were considered significant if the 95% credible interval lower bound exceeded 1. Among 8,804,863 reports, 428,102 involved patients receiving antidepressants. Antidepressant use was associated with a higher likelihood of serious ADRs reporting(58%vs45%; adjusted-ROR:1.78, 95%CI:1.76–1.79). Increased reporting was observed for most tested ADRs, including fatigue(ROR:2.19, 95%CI:2.13–2.25), pain(ROR:2.79, 95%CI:2.69–2.89), and headache(ROR:2.56, 95%CI:2.46–2.62). The timing of antidepressant initiation did not significantly impact ADR reporting. Antidepressant use in oncology patients is associated with increased ADR reporting and seriousness, emphasizing the need for heightened clinical vigilance. Effective management of depressive symptomatology may improve ADR perception, treatment adherence, and patient well-being. Prospective studies are required to confirm these findings and inform targeted interventions.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-28563-9.Keywords: Anticancer drugs, Depression, Adverse reaction, Antidepressants, ManagementSubject terms: Cancer, Psychology, Medical research Introduction Adverse Drug Reactions (ADRs) present a significant challenge in oncology, affecting various organ systems, with variations in frequency and seriousness depending on the oncological treatment utilized1–5. Patients receiving anticancer drugs may experience hematological toxicity, gastrointestinal disorders, and neurotoxicity; those undergoing targeted therapy may suffer from dermatological and cardiovascular toxicities; Immune Checkpoint Inhibitors (ICIs) can lead to immune-related Adverse Drug Reactions (irADRs). Consequently, all patients undergoing oncology treatment are likely to encounter at least one ADRs during the course of their therapy. The majority of ADRs are classified as Common Terminology Criteria for Adverse Events (CTCAE) grade 1–2 and predominantly occur in the skin or gastrointestinal tract6, though approximately 10% are severe, classified as grade 3–47,8. Nevertheless, the documentation and emphasis on ADRs in publications derived from randomized controlled trials (RCTs)9 appear inadequate for achieving comprehensive understanding of these events. Psychiatric ADRs often remain underreported and are not consistently listed in the efficacy RCTs9. However, they have the potential to significantly deteriorate the quality of life for patients who are already grappling with the psychological stress associated with cancer and its treatment10. The interplay between anticancer drugs, depressive symptomatology, and anxiety represents a critical area of investigation, given the substantial psychological burden posed by cancer and its therapeutic interventions. Depressive symptomatology affects approximately 20%11 and anxiety 10% of cancer patients12, irrespective of cancer type and severity13,14. These rates are significantly higher than those observed in the general population12,15, resulting heightened pain and fatigue; and escalating healthcare expenditures accompanied by increased hospital stays16 and resource consumption. Indeed, numerous studies have indicated that depressive symptomatology serves as an independent predictor of poor survival rates among patients with advanced cancer17–19, with Sancassiani et al. reporting a strong correlation between major depressive disorder and premature mortality in cancer patients20–23. In Ikhile’s review study24, a unidirectional perspective was proposed between ADRs and depressive symptomatology, advocating for integrated care to address both physical and mental health challenges in oncology patients24. While this perspective is valid, we contend that the relationship between ADRs stemming from anticancer drugs and depressive symptomatology is bidirectional: ADRs can induce depressive symptomatology, while depressive symptomatology may contribute to the seriousness of ADRs or enhance the overall experience of such events. Conversely, ADRs are not an absolute criteria, but are influenced by contextual elements25 as well as inter-individual variations. In a pro-inflammatory oncological context, inflammation of the insular cortex or a history of preexisting depressive symptomatology may lead to reduced interoceptive accuracy which affects bodily signals perception26. Thus, the emergence of an AE and the subsequent grading thereof, could be subject to distortion27. While the CTCAE gradation of ADRs is standardized to ensure consistent communication among clinical teams, patients, and regulatory entities, the perception and reporting of particular ADRs may differ among depressed patients due to their altered state of mind. This raises questions about whether the tool is applied consistently across patient populations. This gradation remains pivotal to oncology management, as severe grade necessitate treatment discontinuation28. Given the limitations of RCTs in capturing psychiatric ADRs and the subjective nature of ADR perception, pharmacovigilance databases provide a valuable source of real-world data, enabling large-scale analysis of rare and underreported events. Disproportionality analysis further facilitates the detection of potential associations between antidepressant use and ADR seriousness in oncology patients, helping to bridge critical knowledge gaps in pharmacovigilance. Thus, we propose the implementation of a disproportionality analysis in Vigibase® to ensure the association between depressive symptomatology and the incidence of specific ADRs reporting, as well as the evaluation of their seriousness, allowing for discussion around variability in the reporting and classification of serious adverse events, which may be influenced by the patient’s psychological condition. Administration of antidepressants was used as a proxy indicator for depressive symptomatology, given that VigiBase® does not systematically record psychiatric diagnoses. While this approach does not capture all cases of depression, it allows for the identification of a subset of oncology patients likely experiencing depressive symptoms. Materials and methods Population/Data source Our analyses utilized an automatically deduplicated version of VigiBase®, the World Health Organization (WHO) comprehensive Individual Case Safety Report (ICSR) database29. This database encompasses over 39 million reports from 120 nations since 1968, integrating administrative data (including reporting country, report type, and reporter qualifications), patient demographics (gender, age), reaction onset dates, outcome characterization using MedDRA version 27.1 terminology, WHO causality assessment, and detailed drug information (name, administration and cessation dates, induction period, prescribed indications, dosage). Each AE was classified as either ‘serious’ or ‘non-serious’ in accordance with WHO criteria, with ‘serious’ including outcomes such as death, life-threatening situations, hospitalization or its prolongation, persistent incapacity or disability, and clinically significant events as determined by the reporting physician30. As the data do not contain personally identifiable information and are collected for regulatory purposes, this study did not require ethical approval from an Institutional Review Board (IRB) in accordance with local and international ethical guidelines. Descriptive analysis of the population To characterize the study population, we extracted demographic and administrative variables from VigiBase® reports involving antineoplastic or immunomodulating agents. Patients were stratified based on antidepressant exposure, which served as a proxy for depressive symptomatology. Descriptive statistics were computed for age category, sex, WHO region, and seriousness of reported adverse drug reactions (ADRs). Comparisons between patients treated and not treated with antidepressants were performed using Pearson’s chi-square tests for categorical variables. To quantify the strength of associations, we calculated Cramér’s V for each comparison. Effect sizes were interpreted according to conventional thresholds (small: V < 0.1, medium: V ≈ 0.3, large: V > 0.5). These analyses aimed to assess the distributional differences between groups and contextualize the observed disproportionality signals. Case/Non-Case study in VigiBase® For our analyses, ICSRs were limited to those involving liable antineoplastic or immunomodulating agents (per the Anatomical Therapeutic Chemical (ATC) Classification code L; hereafter referred to as antineoplastics) reported in VigiBase from 1968 to July 29th, 2024. Administration of antidepressants was used as an indicator for identifying patients experiencing depressive symptomatology. These agents were identified according to the ATC classification for antidepressants (code N06A). Drugs included in those ATC classes are available in Supplementary Table 1. No exclusion criteria were applied on those drugs in order to avoid missing potentially depressed individuals or patients treated with oncology drugs. The seriousness of reported ADRs was ascertained using the aforementioned WHO pharmacovigilance criteria. No restriction on age was applied for better completeness. Table 1Description of the population. Variable N Antidepressants p-value 2 Cramér’s V0 N = 8,376,76111 N = 428,1021 Age category 5,863,885< 0.0010.0380-27days7,850 (0.1%)181 (< 0.1%)28days-23months35,380 (0.6%)126 (< 0.1%)2–11 years old85,136 (1.5%)623 (0.2%)12–17 years old83,982 (1.5%)2,167 (0.7%)18–45 years old1,180,145 (21%)65,623 (22%)45–64 years old2,353,588 (42%)143,033 (47%)65–74 years old1,142,697 (21%)57,975 (19%)75 + years old671,220 (12%)34,159 (11%)Unknown2,816,763124,215 Sex 8,230,431< 0.0010.062Female4,874,588 (62%)308,828 (75%)Male2,941,959 (38%)105,056 (25%)Unknown560,21414,218 WHO Region 8,804,863< 0.0010.087African Region20,957 (0.3%)675 (0.2%)Eastern Mediterranean Region96,097 (1.1%)1,109 (0.3%)European Region1,963,729 (23%)82,516 (19%)Region of the Americas4,960,136 (59%)328,170 (77%)South-East Asia Region218,235 (2.6%)1,080 (0.3%)Western Pacific Region1,117,607 (13%)14,552 (3.4%) Serious 8,303,757< 0.0010.051No4,313,038 (55%)169,957 (42%)Yes3,590,089 (45%)230,673 (58%)Unknown473,63427,472 Seriousness 3,330,829< 0.0010.082Caused/Prolonged Hospitalization1,065,277 (34%)98,826 (44%)Congenital anomaly/Birth defect4,583 (0.1%)216 (< 0.1%)Death459,906 (15%)24,103 (11%)Disabling/Incapacitating43,423 (1.4%)5,666 (2.5%)Life threatening99,721 (3.2%)8,166 (3.6%)Other1,430,950 (46%)89,992 (40%)Unknown5,272,901201,1331n (%); Description of the population. ‘0’ indicates patients not treated with antidepressants; ‘1’ indicates patients treated with antidepressants. 2Pearson’s Chi-squared test Description of the population. 1n (%); Description of the population. ‘0’ indicates patients not treated with antidepressants; ‘1’ indicates patients treated with antidepressants. 2Pearson’s Chi-squared test Primary Outcome: We aimed to investigate whether serious reports were more reported in oncology patients treated with antidepressants compared to those not receiving such treatment. To address this, a univariate disproportionality analysis was conducted, followed by a multivariable disproportionality analysis adjusted for age category and sex to mitigate bias. Secondary Outcomes: The 20 most frequently reported ADRs among patients treated with antineoplastic or immunomodulating agents were identified using the Preferred Term level of the MedDRA classification. Disproportionality analyses were performed for each of these adverse drug reactions to assess whether they were more reported in oncology patients treated with antidepressants compared to those not receiving antidepressants. A similar analysis was also conducted at the High Level Group Term (HLGT) level as an unadjusted sensitivity analysis. The 20 most frequently reported ADRs among patients treated with antineoplastic or immunomodulating agents were identified using the Preferred Term level of the MedDRA classification. Disproportionality analyses were performed for each of these adverse drug reactions to assess whether they were more reported in oncology patients treated with antidepressants compared to those not receiving antidepressants. A similar analysis was also conducted at the High Level Group Term (HLGT) level as an unadjusted sensitivity analysis. The same 20 most reported ADRs were utilized to create subsets within the population, allowing us to ascertain whether serious reports were more reported in oncology patients treated with antidepressants compared to those not receiving such treatment. A similar analysis was also conducted at the HLGT level as an unadjusted sensitivity analysis. The same 20 most reported ADRs were utilized to create subsets within the population, allowing us to ascertain whether serious reports were more reported in oncology patients treated with antidepressants compared to those not receiving such treatment. A similar analysis was also conducted at the HLGT level as an unadjusted sensitivity analysis. To examine the potential impact of the timing of depressive symptomatology onset (determining whether antidepressant treatment began before or after the initiation of antineoplastics), we assessed whether the seriousness of reports was contingent upon the timing of the onset of depressive symptomatology through disproportionality analyses. This classification was possible only for reports where both start dates were available. The same analyses were also conducted for each of the 20 most reported ADRs in this population. This strategy guarantees sufficient case counts for reliable disproportionality estimates and reflects the most common patient experiences in real-world pharmacovigilance. Administrative events such as drug ineffective, off-label use, and product dose omission were intentionally retained because they provide meaningful insights into treatment adherence, prescribing behaviors, and patient perception—factors particularly relevant in the context of depressive symptomatology. Additionally, we explored whether the onset time of depressive symptomatology influenced the reporting of each of these 20 ADRs using the same methodology. To examine the potential impact of the timing of depressive symptomatology onset (determining whether antidepressant treatment began before or after the initiation of antineoplastics), we assessed whether the seriousness of reports was contingent upon the timing of the onset of depressive symptomatology through disproportionality analyses. This classification was possible only for reports where both start dates were available. The same analyses were also conducted for each of the 20 most reported ADRs in this population. This strategy guarantees sufficient case counts for reliable disproportionality estimates and reflects the most common patient experiences in real-world pharmacovigilance. Administrative events such as drug ineffective, off-label use, and product dose omission were intentionally retained because they provide meaningful insights into treatment adherence, prescribing behaviors, and patient perception—factors particularly relevant in the context of depressive symptomatology. Additionally, we explored whether the onset time of depressive symptomatology influenced the reporting of each of these 20 ADRs using the same methodology. The disproportionality analysis in a case/non-case format has been developed to examine the association between drug exposure and AE reporting31, and can additionally be employed to compare the seriousness of these ADRs. We utilized reporting odds ratios (RORs) and their 95% confidence intervals (CIs) for each of our analyses. A signal was considered statistically significant if the lower bound of the 95% confidence interval for the ROR exceeded 1. P-values were calculated using Wald tests within logistic regression models. Model fit was assessed using Tjur’s R², which quantifies the difference in mean predicted probabilities between cases and non-cases, providing an interpretable measure of discrimination for logistic regression models. Multicollinearity was also assessed using variance inflation factors (VIFs). A sensitivity analysis was also done only with only keeping antidepressant drugs when considered suspect. The analyses adhered to the READUS-PV guidelines and were executed using R version 4.4.1, including the vigicaen package32,33. This study protocol was registered on ClinicalTrials.gov under the identifier NCT06801158. Results Descriptive characteristics VigiBase® reports included 39,240,217 ICSRs at the time of analysis, of which 8,804,863 pertained to liable antineoplastic or immunomodulating agents relevant to our analyses. In this population, reports involving antidepressants amounted to a total of 428,102 ICSRs (4.9%, Supplementary Fig. 1). A higher proportion of women were represented among patients treated with antidepressants (75%) compared to those not treated with them (62%). The most frequently reported age category in both groups was 45–64 years old (47% and 42%, respectively), and most reports originated from the Americas (77% and 59%, respectively). Reports in the antidepressant group were more likely to be classified as serious (58% vs. 45%), with the predominant criteria for seriousness being hospitalization caused or prolonged (44%) and other implications (40%), while in the remaining population, other implications (46%) and hospitalization caused or prolonged (34%) were the most common criteria (Table 1). 88.4% of the reports were coming from pharmacist or physician, suggesting that the dataset is primarily composed of clinically validated reports (Supplementary Table 2). Missing data patterns were assessed by antidepressant exposure. In the study population, age category and sex showed higher missingness in patients not treated with antidepressants compared to those treated (age: 33.6% vs. 29.0%; sex: 6.7% vs. 3.3%; both p < 0.0001), while seriousness was slightly more missing among antidepressant users (6.4% vs. 5.7%; p < 0.0001). Although differences in missingness between groups were statistically significant, the absolute differences were small. No imputation was performed; analyses were restricted to complete cases.Patients treated with antidepressants exhibited a higher likelihood of reporting serious ADRs compared to those not receiving antidepressants (ROR: 1.78; 95%CI: 1.76–1.79). The sensitivity analysis yielded similar results (ROR: 1.54; 95%CI: 1.50–1.58). This trend was consistently observed across all 20 most frequently reported ADRs in this population (Table 2). Similar patterns were observed in the univariate analyses conducted across all reported adverse events, as detailed in Supplementary Table 3. Table 2Seriousness depending on antidepressant exposure.CategoryAdverse eventNN antidepressants (Serious)N antidepressants (Non-Serious)Univariate Model ROR (CI)pMultivariate Model* ROR (CI)p Overall All combined8,303,757230,673169,9571.63 (1.62–1.64)< 0.0011.78 (1.76–1.79)< 0.01 Gastrointestinal Nausea404,70016,55614,2612.41 (2.09–2.19)< 0.0012.34 (2.28–2.41)< 0.001Diarrhoea364,56514,21311,5381.76 (1.72–1.81)< 0.0011.93 (1.87–1.99)< 0.001Vomiting229,70910,5335,2612.28 (2.20–2.36)< 0.0012.56 (2.46–2.66)< 0.001 General/Constitutional Fatigue411,35519,52516,6152.04 (2.00–2.09.00.09)< 0.0012.19 (2.13–2.25)< 0.001Asthenia182,45610,3504,4282.29 (2.20–2.37)< 0.0012.53 (2.42–2.65)< 0.001Malaise174,1727,1044,8002.46 (2.36–2.55)< 0.0012.48 (2.37–2.60)< 0.001Pyrexia207,4458,9703,3922.11 (2.02–2.20)< 0.0012.23 (2.12–2.34)< 0.001 Neurological Headache277,74712,41311,8372.50 (2.43–2.56)< 0.0012.54 (2.46–2.62)< 0.001Dizziness163,7419,4946,7212.24 (2.17–2.32)< 0.0012.37 (2.28–2.46)< 0.001 Musculoskeletal Pain239,17713,3187,7252.68 (2.60–2.76)< 0.0012.79 (2.69–2.89)< 0.001Arthralgia219,22310,0196,9502.76 (2.67–2.85)< 0.0012.88 (2.77–2.99)< 0.001 Dermatological Rash241,0526,0736,6501.85 (1.79–1.92)< 0.0011.89 (1.81–1.97)< 0.001Pruritus203,2984,8736,2772.14 (2.06–2.22)< 0.0012.24 (2.14–2.34)< 0.001 Respiratory Dyspnoea204,84611,8473,6452.18 (2.10–2.26)< 0.0012.47 (2.35–2.58)< 0.001Pneumonia144,80712,1145052.15 (1.96–2.36)< 0.0012.50 (2.23–2.80)< 0.001 Drug Administration Injection site pain188,1122,3067,0132.59 (2.46–2.72)< 0.0012.20 (2.08–2.34)< 0.001Product dose omission173,8744,6457,7952.18 (2.10–2.27)< 0.0012.26 (2.15–2.37)< 0.001Off label use248,2606,5235,0971.66 (1.60–1.72)< 0.0011.81 (1.73–1.90)< 0.001Drug ineffective385,0828,63511,5991.94 (1.89–2.00.89.00)< 0.0011.87 (1.80–1.94)< 0.001 Mortality Death262,6189,665431.23 (0.92–1.69)0.1861.50 (1.08–2.16)0.023*: adjusted on age category and sex. Seriousness depending on antidepressant exposure. *: adjusted on age category and sex. Exceptions included death (ROR: 0.94; 95%CI: 0.92–0.96) and pruritus (ROR: 0.98; 95%CI: 0.96–1.00.96.00), which were marginally less reported among patients treated with antidepressants, as well as off-label use and injection site pain, which did not exhibit statistically significant differences between patients treated with antidepressants and those not receiving such treatment (p = 0.193 and p = 0.074, respectively). All remaining ADRs among the 20 tested were associated with an increased reporting rate in patients undergoing antidepressant therapy (Table 3). Similar patterns were observed in the univariate analyses conducted across all reported adverse events, as detailed in Supplementary Table 4. Table 3Adverse Drug Reactions depending on antidepressant exposure.CategoryAdverse eventNN antidepressant reportsN non-antidepressant reportsUnivariate Model ROR (CI)pMultivariate Model* ROR (CI)p Gastrointestinal Nausea404,70032,440391,5641.67 (1.65–1.69)< 0.0011.44 (1.42–1.46)< 0.001Diarrhoea364,56526,742353,5141.51 (1.49–1.53)< 0.0011.44 (1.42–1.46)< 0.001Vomiting229,70916,778228,8961.45 (1.43–1.48)< 0.0011.28 (1.26–1.31)< 0.001 General/Constitutional Fatigue411,35536,930383,8601.97 (1.94–1.99)< 0.0011.87 (1.85–1.90)< 0.001Asthenia182,45615,441174,7991.76 (1.73–1.79)< 0.0011.63 (1.60–1.66)< 0.001Malaise174,17212,311167,4971.45 (1.42–1.48)< 0.0011.41 (1.38–1.44)< 0.001Pyrexia207,44513,184214,9881.21 (1.18–1.23)< 0.0011.21 (1.18–1.23)< 0.001 Neurological Headache277,74725,634267,3031.93 (1.91–1.96)< 0.0011.77 (1.74–1.80)< 0.001Dizziness163,74117,405158,0962.20 (2.17–2.24)< 0.0012.01 (1.97–2.05)< 0.001 Musculoskeletal Pain239,17721,685227,4231.91 (1.88–1.94)< 0.0011.80 (1.77–1.83)< 0.001Arthralgia219,22317,465208,4891.67 (1.64–1.69)< 0.0011.51 (1.48–1.54)< 0.001 Dermatological Rash241,05213,761249,8151.08 (1.06–1.10)< 0.0011.05 (1.03–1.07)< 0.001Pruritus203,29812,090209,0851.13 (1.11–1.15)< 0.0010.98 (0.96–1.00.96.00)0.040 Respiratory Dyspnoea204,84616,503209,0851.57 (1.54–1.59)< 0.0011.41 (1.38–1.43)< 0.001Pneumonia144,80712,952138,6071.85 (1.82–1.89)< 0.0011.88 (1.84–1.92)< 0.001 Drug Administration Injection site pain188,1129,512181,9841.02 (1.00–1.04.00.04)0.0311.02 (1.00–1.05.00.05)0.074Product dose omission173,87412,443161,5401.52 (1.49–1.55)< 0.0011.35 (1.32–1.38)< 0.001Off label use248,26011,632237,1880.96 (0.94–0.98)< 0.0011.02 (0.99–1.04)0.193Drug ineffective385,08220,601370,8271.09 (1.08–1.11)< 0.0011.19 (1.16–1.21)< 0.001 Mortality Death262,61810,220269,1090.74 (0.72–0.75)< 0.0010.94 (0.92–0.96)< 0.001*: adjusted on age category and sex. Adverse Drug Reactions depending on antidepressant exposure. *: adjusted on age category and sex. Data regarding the timing of antidepressant initiation (prior to or following the initiation of antineoplastic therapy) was available for only 2,045 ICSRs. No significant difference in seriousness was observed between these two groups for any of the researched ADRs (Table 4). With the exception of pyrexia, which was less reported in individuals who commenced antineoplastics after initiating antidepressants compared to those who began antidepressant treatment subsequent to antineoplastics (ROR 0.43, 95%CI 0.23–0.79), no statistically significant differences were observed between the two groups for any of the other 19 ADRs analysed (Table 5). As anticipated in disproportionality analyses involving rare events and large datasets, Tjur’s R² values were consistently below 0.1. This low discriminative power is typical and does not affect the reliability of RORs, which remain valid for signal detection. VIFs were calculated to assess multicollinearity, and all values were below 2, indicating no concerning collinearity among covariates. Table 4Seriousness in antidepressant treated patients depending on antidepressant starting date (before or after antineoplastic).CategoryAdverse eventNAntidep. firstAntineop. firstUnivariate Model*¹ ROR (CI)pMultivariate Model*¹’² ROR (CI)pSeriousNon-SeriousSeriousNon-Serious Overall All combined1,6355562056252490.93 (0.74–1.15)0.4850.89 (0.71–1.12)0.329 Gastroi- ntestinal Nausea77141433162.06 (0.80–5.41)0.1361.56 (0.51–4.92)0.439Diarrhoea541991790.89 (0.29–2.80)0.8471.00 (0.24–4.26)0.999Vomiting6216926111.33 (0.45–3.93)0.6050.91 (0.24–3.27)0.884 General/ Constitutional Fatigue6919831111.19 (0.40–3.47)0.7550.68 (0.18–2.47)0.567Asthenia4310101853.60 (0.99–14.50)0.0582.37 (0.38–16.06)0.351Malaise38951771.35 (0.32–5.53)0.6761.64 (0.28–10.03)0.577Pyrexia523051612.67 (0.39–53.43)0.3891.09 (0.07–29.65)0.953 Neurological Headache561872471.33 (0.39–4.57)0.6421.31 (0.33–5.21)0.693Dizziness72171225180.98 (0.37–2.55)0.9680.83 (0.25–2.67)0.751 Musculoskeletal Pain27641433.11 (0.53–20.40)0.2103.33 (0.36–43.02)0.303Arthralgia421222170.50 (0.07–2.48)0.4310.22 (0.01–2.09)0.235 Dermatological Rash5616112091.53 (0.51–4.68)0.4501.06 (0.29–3.75)0.927Pruritus4411713130.64 (0.18–2.14)0.4680.83 (0.21–3.25)0.790 Respiratory Dyspnoea441152625.91 (1.09–45.89)0.0514.34 (0.60–45.38.60.38)0.166Pneumonia—————*³—*³— Drug Administration Injection site pain—————*³—*³—Product dose omission—————*³—*³—Off label use23821121.38 (0.14–13.58)0.7730.50 (0.01–10.86)0.661Drug ineffective26741232.29 (0.39–14.72)0.3580.41 (0.01–4.93)0.523 Mortality Death—————*³—*³—*¹: A positive ROR is in favour of a higher reporting for patients who started the neoplastic drug first. *²: adjusted on age category and sex. *³: Results for some variables could not be obtained due to insufficient variation or complete separation, which prevented proper model estimation. Seriousness in antidepressant treated patients depending on antidepressant starting date (before or after antineoplastic). Gastroi- ntestinal General/ Constitutional *¹: A positive ROR is in favour of a higher reporting for patients who started the neoplastic drug first. *²: adjusted on age category and sex. *³: Results for some variables could not be obtained due to insufficient variation or complete separation, which prevented proper model estimation. Table 5Adverse Drug Reactions in antidepressant treated patients depending on antidepressant starting date (before or after antineoplastic).CategoryAdverse eventNAntidep. firstAntineop. firstUnivariate Model*¹ ROR (CI)pMultivariate Model*¹’² ROR (CI)pSeriousNon-SeriousSeriousNon-Serious Gastro- intestinal Nausea8933907561,0491.47 (0.95–2.30)0.0871.16 (0.71–1.93)0.547Diarrhoea6835905331,0720.80 (0.49–1.29)0.3550.69 (0.40–1.17)0.170Vomiting7328912451,0601.38 (0.86–2.26)0.1861.34 (0.80–2.30)0.273 General/ Constitutional Fatigue8029911511,0541.52 (0.96–2.45)0.0771.48 (0.89–2.51)0.140Asthenia5124916271,0780.96 (0.55–1.68)0.8740.89 (0.49–1.60)0.689Malaise4315925281,0771.60 (0.86–3.10)0.1441.50 (0.78–3.00.78.00)0.236Pyrexia5737903201,0850.45 (0.25–0.77)0.0040.43 (0.23–0.79)0.007 Neurological Headache6628912381,0671.16 (0.71–1.92)0.5581.22 (0.71–2.14)0.477Dizziness8134906471,0581.18 (0.76–1.87)0.4631.21 (0.74–2.01)0.443 Musculoskeletal Pain3011929191,0861.48 (0.71–3.23)0.3061.48 (0.65–3.54)0.358Arthralgia4814926341,0712.10 (1.14–4.06)0.0211.93 (0.98–4.01)0.065 Dermatological Rash6630910361,0691.02 (0.62–1.68)0.9321.04 (0.63–1.76)0.867Pruritus5722918351,0701.36 (0.80–2.38)0.2591.46 (0.84–2.61)0.187 Respiratory Dyspnoea5521919341,0711.39 (0.81–2.45)0.2421.36 (0.77–2.46)0.295Pneumonia248932161,0891.71 (0.75–4.24)0.2172.64 (1.00–8.24.00.24)0.065 Drug Administration Injection site pain7193961,0995.13 (0.87–96.91)0.131*³—Product dose omission31939161,0891.70 (0.16–36.67)0.6641.86 (0.17–40.59)0.618Off label use2310930131,0921.11 (0.48–2.60)0.8100.97 (0.38–2.46)0.941Drug ineffective3513927221,0831.45 (0.74–2.97)0.2931.71 (0.81–3.85)0.174 Mortality Death201392771,0980.45 (0.17–1.11)0.0940.46 (0.17–1.13)0.101*¹: A positive ROR is in favour of a higher reporting for patients who started the neoplastic drug first. *²: adjusted on age category and sex. *³: Results for some variables could not be obtained due to insufficient variation or complete separation, which prevented proper model estimation. Adverse Drug Reactions in antidepressant treated patients depending on antidepressant starting date (before or after antineoplastic). Gastro- intestinal General/ Constitutional *¹: A positive ROR is in favour of a higher reporting for patients who started the neoplastic drug first. *²: adjusted on age category and sex. *³: Results for some variables could not be obtained due to insufficient variation or complete separation, which prevented proper model estimation. Discussion Our study indicates a significant over-reporting of ADRs among oncology patients treated with antidepressants. These findings suggest that depressive symptomatology may influence how patients perceive and report ADRs, which could have implications for treatment adherence and patient support strategies in oncology. The ratio of serious to non-serious events appears elevated within the oncology population receiving antidepressants (58/42) compared to those not treated with these agents (55/45). In a multivariate model, odds ratios were noted to be significantly higher for pain (ROR = 2.79, 95%CI: 2.69–2.89), headache (ROR = 2.56, 95%CI: 2.46–2.62), and respiratory discomfort (ROR = 2.50, 95%CI: 2.23–2.80), as well as for vomiting, asthenia, arthralgia, and rash, all elements that may be subject to psychosomatic influences. Our findings are consistent with a prospective study analysing 290 postoperative early-stage breast cancer patients, which established a significant correlation between depressive symptomatology and the incidence of grade 2 myelosuppression34. Although our study does not assess biological susceptibility, this supports the hypothesis that depressive symptomatology may influence how ADRs are perceived and reported. The same research team conducted a multi-center cross-sectional survey to evaluate the incidence of depressive symptomatology in cancer patients who experienced serious ADRs: among the 112 patients enrolled, the prevalence of moderate to severe depressive symptomatology was found to be 38.4%. Reduced cognitive function and lower overall quality of life were identified as risk factors for depressive symptomatology in these patients35. This contrasts with the results of our study, where only 6.04% of the serious documented ADRs were concurrently reported with antidepressant use. This discrepancy may be attributed to underreporting within pharmacovigilance systems, wherein concomitant medications are less often documented36,37. Nonetheless, the findings bolster our conclusion, as we still identify an over-reporting of ADRs and serious ADRs among oncology patients treated with antidepressants, despite the likelihood that many oncology patients categorized as not treated with antidepressants are, in fact, receiving such treatment, which diminishes the potential for detecting signals associated with the co-administration of antidepressants in our population. Consequently, the actually observed signal may be even stronger than reported. In our study, the reporting of depressive symptomatology among oncology patients was notably higher in women compared to men, with the predominant age category being 45–64 years. This aligns with a Scottish study where 71% were women with a mean age of 56.6 (SD 11.9) years38. Given that the 45–64 age category was associated with increased ADR reporting among patients treated with antidepressants, and considering the higher prevalence of depressive symptomatology in this demographic, it is plausible that depressive symptomatology contributes to differences in ADR reporting patterns in oncology patients. Off-label drug use was in proportion slightly less frequently reported among oncology patients treated with antidepressants compared to those not receiving them. However, when off-label use was reported in the antidepressant group, it was more often classified as serious. This suggests that patients with depressive symptomatology may perceive off-label treatments as more distressing or clinically significant, potentially influencing the seriousness classification of associated ADRs. This could arise from the reality that patients prescribed antidepressants often present with coexisting depressive symptomatology, known to correlate with increased severity of cancer-related symptoms, decreased treatment tolerance, and heightened risk of ADRs. Thus, the interplay of depressive symptomatology, off-label drug use, and anticancer drugs may be perceived as contributing to more severe clinical complications. The differentiated perception of pain among oncology patients treated with antidepressants compared to other oncology patients underscores distinct mechanisms underlying localized and generalized pain experiences. Global pain, often experienced as chronic or diffuse pain, is markedly modulated by central sensitization—a phenomenon wherein the central nervous system becomes hypersensitive to pain signals. In cases of depressive symptomatology, neurochemical imbalances, such as diminished serotonin and norepinephrine levels, disrupt descending pain modulation pathways, culminating in amplified pain perceptions39. Furthermore, brain regions associated with emotional and cognitive processing contribute to enhanced pain perception in depressive states40. These differences elucidate why oncology patients treated with antidepressants may not report heightened injection site pain, yet frequently experience increased generalized pain. Regarding the signal of gastroenterological ADRs observed in our study, existing literature indicates that functional digestive disorders often originate from dysregulation of the brain-gut axis41. The enteric nervous system governs intestinal motility, secretions42, blood flow, local immunity, and microbiota homeostasis43. Helicobacter pylori, in particular, has the capacity to activate the ERK/MAPK proliferation pathway44, contributing to oncogenesis. Furthermore, robust links between alterations in the intestinal microbiome and the onset of depressive symptomatology have been documented45,46, underscoring the imperatives of a holistic treatment approach that accommodates psychological variables. With respect to the signal of cutaneous ADRs identified in our study, an emerging concept termed psychodermatology47 emphasizes that skin conditions can reciprocally influence and be influenced by psychological states, thus highlighting the necessity for a holistic approach in patient care48. Given that the skin serves as the principal barrier between the host and external environment, a compromised skin barrier could instigate microbial dysbiosis49, facilitated by the continual interactions between the skin barrier and commensal microbiota50, both of which are mediated by psychological factors51. Additionally, our findings indicated a signal of fatigue among oncology patients undergoing antidepressant treatment, which is traditionally associated with depressive symptomatology and may further aggravate clinical outcomes52. These patients frequently experience discomfort, including aches and pains, irritable bowel syndrome, headaches, and subjective sensations of infection53, consistent with the observed signals in our study. Although our study does not directly assess the mechanisms underlying arthralgia, previous research suggests that depressive symptomatology may exacerbate pain perception through neuroinflammatory and psychosomatic pathways54. It is therefore plausible that the combination of anticancer drug-related ADRs and depressive symptomatology contributes to increased reporting of arthralgia in this population. It is reasonable to hypothesize that ADRs stemming from anticancer drugs, combined with symptomatology associated with depressive symptomatology, create a synergistic effect that significantly amplifies both the frequency and intensity of ADRs for these patients. The evaluation of the seriousness of all ADRs investigated in oncology patients exhibited no substantial dependency on the timing of antidepressant initiation, whether it transpired before or after the commencement of antineoplastic therapy. This finding implies that the initiation of antidepressants, relative to the onset of anticancer drugs, does not significantly impact the intensity or seriousness of ADRs experienced by patients. The results suggest that other factors, such as the individual characteristics of patients or the specific anticancer drugs employed, may exert a more pronounced influence on the seriousness of ADRs than does the sequence of antidepressant and anticancer drugs initiation. When depressive symptomatology occurs in the oncology population, it is generally treated analogously to depressive symptomatology in the broader population, including selective serotonin reuptake inhibitors (SSRIs)19,21,55,56 or psychotherapy and psychosocial support14. However, in numerous oncology clinical investigations, the differentiation between trait depressive symptomatology, defined as a stable predisposition toward depressive episodes, and state depressive symptomatology, categorized as a reactive condition, is often overlooked, despite the critical importance of this distinction within the oncology sphere13,14. Several studies have sought to identify oncology patients at higher risk of developing depressive symptoms, but findings have been inconsistent57,58. Approximately one in four patients experiencing depressive symptomatology exhibit clinically significant symptoms such as insomnia, irritability or anxiety disorder59. Various hypotheses have been proposed to elucidate the somatic underpinnings associated with the prevalence and incidence of anxiety and depressive disorders throughout oncology treatment. However, in cases of reactive depressive symptomatology, these somatic factors are likely secondary, as depressive symptoms often arise as a psychological response to the diagnosis of a serious or terminal condition. Studies frequently emphasize that neurological and psychiatric disorders are impactful and primarily driven by the underlying neuroinflammation, which serves as the basis for many ADRs60,61, such as the release of inflammatory cytokines by specific malignancies (e.g., pancreatic and lung cancers) or as adverse effects of therapies like anticancer drugs and corticosteroids13. Hormonal treatments, particularly in breast cancer, may exacerbate psychological conditions due to their systemic effects on estrogen levels, which are closely linked to mood regulation. Depressive symptomatology and anxiety frequently manifest in patients undergoing long-term endocrine therapy, highlighting how particular treatments can trigger or exacerbate mental health issues62. While pharmacovigilance data cannot establish causality or guide clinical decision-making directly, the observed associations warrant further investigation. In particular, the integration of psychological assessment into oncology care may help contextualize ADR reports and improve patient-centered management. Our analysis encountered several inherent limitations attributable to employing a pharmacovigilance database. The most notable limitation pertains to underreporting, a common issue in pharmacovigilance; however, this did not compromise the results or the significance of the disproportionality analysis63. Moreover, the probability that a suspected adverse event is drug-related may vary among cases. While the disproportionality analysis of spontaneous reports serves as a valuable instrument for detecting safety signals, it possesses limitations, including potential data quality issues due to incomplete information and the challenge of establishing definitive causal relationships between drugs and ADRs64. As previously mentioned, the infrequent co-reporting of antidepressants within pharmacovigilance data for oncology patients limits our ability to distinguish between trait depressive symptomatology and state depressive symptomatology. In our sample of cancer patients prescribed antidepressants, it is likely that they can be divided into two distinct subgroups. Our study population likely includes both patients with pre-existing (trait) depressive symptomatology and those with reactive (state) depressive symptomatology triggered by cancer diagnosis or treatment. These subgroups may differ in how ADRs are perceived and reported, which could influence the observed signals. Further research is needed to explore these differences. Therefore, the disproportionality analysis merely provides a rudimentary indication of signal strength, serving predominantly to generate hypotheses regarding novel potential ADRs65,66. Although the dataset included reports from individuals aged 0–17 years, this subgroup represented less than 3% of the total population. Given known differences in AE reporting between pediatric and adult populations, age category was included as an adjustment variable in all multivariable analyses to minimize bias. Another limitation of this study is the potential for confounding by indication, as antidepressants may be prescribed for conditions other than depressive symptomatology, such as anxiety, insomnia, or neuropathic pain. While this proxy enables the identification of a subset of patients likely experiencing depressive symptoms, it does not allow for precise classification. These underlying conditions may independently influence ADR reporting patterns, complicating the interpretation of associations. Additionally, incomplete reporting of concomitant medications in pharmacovigilance databases may obscure interactions or cumulative effects that contribute to ADRs. While we adjusted for age and sex to reduce bias, residual confounding from these clinical factors cannot be excluded. VigiBase® does not systematically capture clinical variables such as cancer stage, performance status, comorbidities, or psychosocial context, which may influence both the likelihood of antidepressant prescription and the reporting of adverse drug reactions. Additionally, the indication for antidepressant use (e.g., depression vs. anxiety or neuropathic pain) is not consistently documented, further complicating interpretation. These factors may confound the observed associations and should be considered when interpreting the results. In our study, ADRs were reported following the introduction of antidepressants; however, we cannot entirely dismiss the possibility that some antidepressants were initiated post-occurrence of prior ADRs. Additionally, treating depressive symptomatology tendentially implies symptom alleviation, potentially diminishing the reported seriousness of ADRs and attenuating signal for ADRs. This study elucidates a significant safety signal linking depressive symptomatology in oncology patients with the manifestation of ADRs and serious ADRs. Given the reliance on pharmacovigilance data, a definitive causal relationship between depressive symptomatology and the onset of ADRs cannot be established; however, the findings underscore the necessity for heightened clinical vigilance among patients with depressive symptomatology or those at increased risk for depressive symptomatology, as the effective management of these conditions may profoundly enhance both treatment efficacy and patient quality of life. Although not directly assessed in our study, previous research suggests that even ADRs considered non-serious may contribute to psychological distress in vulnerable patients, potentially affecting treatment adherence and perceived quality of life67. These events warrant meticulous monitoring and management. Additional prospective research is imperative to validate the signals identified in this analysis and to elucidate optimal strategies for the treatment and prevention of depressive symptomatology in oncology patients. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Acknowledgements Basile Chretien gratefully acknowledges the support of the THERS Make New Standards Program for the Next Generation Researchers (RESEARDENT), Tokai National Higher Education and Research System, Nagoya University [Authorized Number: L250015]. Abbreviations Adverse Drug Reaction Adverse Event Anatomical Therapeutic Chemical Confidence Interval Common Terminology Criteria for Adverse Events Immune Checkpoint Inhibitor Immune-related Adverse Drug Reaction Individual Case Safety Report Medical Dictionary for Regulatory Activities Odds Ratio Pharmacovigilance Randomized Controlled Trial REporting Accuracy and Data USability in Pharmacovigilance Reporting Odds Ratio World Health Organization Author contributions Basile Chretien and Jean-Matthieu L’Orphélin contributed to the study design, performed the research data analysis, interpretation of results, and manuscript writing. Charles Dolladille, Kazuki Nishida, Branko Aleksic, and Joachim Alexandre critically reviewed the manuscript for important intellectual content. All authors approved the final version of the manuscript. Funding The authors received no specific funding for this work. Data availability The data analyzed in this study were obtained from VigiBase®, maintained by Uppsala Monitoring Centre (UMC). Access to VigiBase® is restricted and subject to specific terms and conditions. Thus, data cannot be obtained from the corresponding author. However, data requests may be directed to UMC, subject to their data access policies and agreements. Declarations Competing interests The authors declare no competing interests. Consent for publication Not applicable, as this study does not include identifiable individual data. Ethics approval and consent to participate This study is based on anonymized pharmacovigilance data from VigiBase®. Since the data do not contain personally identifiable information and were collected for regulatory purposes, ethical approval from an Institutional Review Board (IRB) was not required, in accordance with local and international ethical guidelines. This study was conducted in compliance with the Declaration of Helsinki. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Data Availability Statement The data analyzed in this study were obtained from VigiBase®, maintained by Uppsala Monitoring Centre (UMC). Access to VigiBase® is restricted and subject to specific terms and conditions. Thus, data cannot be obtained from the corresponding author. However, data requests may be directed to UMC, subject to their data access policies and agreements.