Multimodal investigation of the association between bevacizumab and pulmonary arterial hypertension

Abstract

Background

Pulmonary arterial hypertension (PAH) is a rare and severe condition that may be associated with exposure to certain drugs. Bevacizumab, a monoclonal antibody targeting vascular endothelial growth factor, has been classified as having a possible association with PAH, but data remain scarce. This study aimed to assess the potential link between bevacizumab exposure and PAH through complementary clinical, medico-administrative and pharmacovigilance databases.

Methods

We performed a multimodal investigation combining: 1) case identification from the French Pulmonary Hypertension (PH) Registry, 2) a population-based cohort analysis using the French National Health Data System (Système National des Données de Santé (SNDS)) and 3) a disproportionality analysis in the World Health Organization pharmacovigilance database (VigiBase). Clinical, haemodynamic and exposure data were reviewed in the French PH Registry; prevalence of PAH in patients exposed to bevacizumab was computed in the SNDS; and disproportionality analyses (information component (IC) values) were calculated in VigiBase.

Results

Seven cases of PAH temporally associated with bevacizumab were identified in the French PH Registry, most displaying predisposing conditions (BMPR2 mutation, portal hypertension or COPD). In the SNDS cohort of 9964 PAH patients, 30 had prior bevacizumab exposure, with no increased incidence compared to the general population. In VigiBase, 21 PAH cases were reported with bevacizumab, yielding a modest disproportionality signal only under broad definitions (IC 0.4 (95% credibility interval 0.1–0.6)).

Conclusions

Across complementary data sources, there was no evidence of a causal association between exposure to bevacizumab and the development of PAH in adults. However, the effect of bevacizumab as a secondary trigger in predisposed individuals cannot be definitely or completely excluded.

Shareable abstract

Exposure to bevacizumab was not found to be associated with an increased risk of PAH across clinical, population-based and pharmacovigilance data in adults, although a potential secondary trigger role in predisposed individuals cannot be ruled out https://bit.ly/4dGpUFG

Introduction

Pulmonary arterial hypertension (PAH) is a rare form of pulmonary hypertension (PH), characterised by remodelling of the three layers of the distal pulmonary vasculature (i.e. vessels with a calibre of 50–500 μm) which involves uncontrolled growth of endothelial and smooth muscle cells leading to an increase in pulmonary vascular resistance (PVR) [1]. PAH may be idiopathic, heritable, associated with various diseases or associated with drugs and toxins use [2]. In 2004, the US Food and Drug Administration granted approval for bevacizumab for the first-line treatment of metastatic colorectal cancer before seeing its indications extended to many other advanced or metastatic cancers [3]. Bevacizumab is a recombinant humanised monoclonal IgG antibody that selectively binds to and neutralises the biological activity of human vascular endothelial growth factor (VEGF). The drug demonstrates high affinity for circulating VEGF [4], neutralising the biological activity of both VEGF receptor (VEGFR)-1 and -2. This mechanism can result in the reduction of tumour vascularisation but may, on the other hand, induce proliferative pulmonary vasculopathy [5]. Indeed, in the well-established Sugen/hypoxia model of PH, inhibition of VEGFR signalling by the angiogenesis inhibitor SU5416 is thought to contribute to endothelial dysfunction and vascular remodelling, mimicking key features of the human disease [6, 7]. Moreover, pathogenic variants in the KDR gene have recently been established as a cause of heritable PAH, typically associated with a distinct clinical phenotype characterised by markedly reduced diffusing capacity of the lung for carbon monoxide (D_LCO) and coexistent parenchymal lung abnormalities. The KDR gene encodes for VEGFR-2, the main receptor for VEGF-A, which plays a central role in the development and maintenance of the pulmonary vasculature.

In a large cohort of pre-term infants treated with intravitreal bevacizumab for retinopathy, bevacizumab was associated with a clinically important increase of PAH-approved drug use, suggesting that bevacizumab may be associated with PAH in this particular population [8]. The literature also reports several case series describing the development of PH in patients exposed to bevacizumab [9–11]. Based on these reported cases and the underlying pathophysiological rationale, bevacizumab was classified as a drug with a “possible association” with PAH in the 7th World Symposium on Pulmonary Hypertension (WSPH) (Barcelona, 2024) [2]. However, the available data remain limited and do not allow a definitive conclusion regarding a causal relationship between bevacizumab exposure and the onset of PAH.

The aim of this study was to investigate the reliability of the association between bevacizumab and PAH in adult patients and, secondarily, to describe the characteristics of the identified cases. Therefore, we combined data from the French PH Registry, the French National Health Data System (Système National des Données de Santé (SNDS)) and the World Health Organization (WHO) pharmacovigilance database (VigiBase) to explore this association.

Methods

Analysis of the French PH Registry

Identification of cases

We reviewed all cases of PH associated with exposure to bevacizumab recorded in the French PH Registry (including cases from the PH reference centre at Hospital Bicêtre, University of Paris-Saclay, Paris and 28 other PH expert centres) over 21 years (1 May 2004 to 1 May 2025). Inclusion criteria were a confirmed diagnosis of PH by right heart catheterisation (RHC), age over 18 years and a bevacizumab exposure, apart from a context of hereditary haemorrhagic telangiectasia. PH diagnostic thresholds used were based on the haemodynamic definition in use at the time of inclusion in the registry [12–14]. The date of PH diagnosis was defined as the date of the initial RHC. A comprehensive aetiological evaluation of PH was conducted in accordance with the European Society of Cardiology/European Respiratory Society guidelines [13, 14].

The registry was established in accordance with French bioethics laws (Commission Nationale de l'Informatique et des Libertés number 842063), and all patients gave their formal informed consent.

Clinical, functional, radiological and haemodynamic evaluation

Specific data concerning exposure to bevacizumab were collected: dosage and duration of exposure, reason for introduction and combination with other drugs associated with PAH or not. We collected demographic data, New York Heart Association Functional Class (NYHA FC), N-terminal pro-brain natriuretic peptide, 6-min walk distance (6MWD) and pulmonary function data, including D_LCO, of patients at inclusion. Haemodynamic parameters recorded during RHC were also collected, including mean pulmonary arterial pressure (mPAP), pulmonary arterial wedge pressure (PAWP), right atrial pressure (RAP), cardiac output and mixed venous oxygen saturation. PVR was calculated as (mPAP−PAWP)/cardiac output. When available, these parameters were also collected at first follow-up as well as PAH treatments. Vital status was assessed using the national open-access database matchID (https://deces.matchid.io).

Analysis of the French national healthcare database (SNDS)

The SNDS database is a nationwide database of claims and hospital care, with almost 20 years of history. We first identified all dispensations of bevacizumab in the French population, using French Presentation Identification Code (CIP) and Common Dispensing Unit (UCD) codes, between 1 January 2009 and 31 December 2024. Then we estimated the prevalence of PAH after exposure to bevacizumab using a validated algorithm [15]. Criteria of identification were hospitalisation with a specific PAH diagnosis code (International Classification of Diseases, 10th Revision: I27.0), a RHC procedure within 1 year around this hospitalisation and a dispensation of specific PAH treatment 6 months before or ever after hospitalisation or RHC. The date of diagnosis was defined as the date of the first hospitalisation, the first RHC or the first specific treatment dispensation, whichever came first. For comparison we estimated the prevalence of PAH in the entire French population using the same algorithm for PAH identification. The total number of person-years was determined from the SNDS data, from the time beneficiaries entered the database until their date of death or 1 January 2025. Characteristics of PAH patients exposed to bevacizumab were also compared with non-exposed PAH patients. All code-based identification algorithms (e.g. PAH, COPD and systemic scleroderma) used in this study are available in supplementary tables S1–S3.

Analysis of the WHO pharmacovigilance database (VigiBase)

We reported the present disproportionality analysis following the READUS-PV recommendations [16] (https://readus-statement.org) (supplementary table S4). The protocol of the study was pre-registered in OSF (Open Science Framework: https://osf.io/mjgvz).

We first extracted all Individual Case Safety Reports (ICSRs) of PAH associated with a VEGF inhibitor in VigiBase (bevacizumab, ranibizumab, ramucirumab, aflibercept, sorafenib, sunitinib, pazopanib, vandetanib, cabozantinib, axitinib, ponatinib, regorafenib, nintedanib and lenvatinib [17]).

Cases were identified through a prespecified collection of MedDRA (Medical Dictionary for Regulatory Activities) terms related to PAH: cor pulmonale, cor pulmonale acute, cor pulmonale chronic, PAH, PAP abnormal, PAP increased, pulmonary artery dilatation, pulmonary artery wall hypertrophy, PH, pulmonary hypertensive crisis, pulmonary valve incompetence, PVR abnormality, pulmonary veno-occlusive disease and vascular resistance pulmonary increased.

We considered only deduplicated cases, drugs considered as suspect or interacting and we performed the last update of the analysis on 2 October 2025.

Statistical methods

For analysis of the French PH Registry, all quantitative variables are presented as median and interquartile range (IQR), and qualitative variables as number and frequency. Non-parametric tests were used, notably the Mann–Whitney U-test or Wilcoxon rank test for before/after comparisons. Results with a two-sided p-value <0.05 were considered statistically significant. All graphs, calculations and statistical analyses were performed using Prism version 10.2.1 for Mac (GraphPad, San Diego, CA, USA).

For the analysis of the WHO pharmacovigilance database (VigiBase), we used the “Bayesian confidence propagation neural network method” developed by the Uppsala Monitoring Centre research team, which displays the best sensitivity and specificity and fewer fluctuations for rare events [18, 19]. The analyses were performed using the whole database as a comparator group, after excluding ICSRs relating solely to vaccines in order to reduce masking bias associated with COVID-19 vaccines [20]; indeed, at the time of analysis, around 10 million cases were solely related to vaccines, resulting in a considerable increase in the number of expected cases and a decrease in sensitivity. A signal of disproportionate reporting was deemed significant if the lower boundary of the 95% credibility interval (95% CrI) of the information component (IC₀₂₅) was >0 [18, 21].

We performed disproportionality analyses using two definitions of PAH and using the whole database (excluding vaccines) as a reference/comparator group: 1) narrow definition (MedDRA Preferred Term “Pulmonary arterial hypertension”) and 2) broad definition (collection of MedDRA terms related to PAH, as described above).

Some sensitivity analyses were performed using the broad definition. 1) Excluding competitors: we excluded drug (i.e. benfluorex, fenfluramine, dexfenfluramine and dasatinib [22]) competitors to limit potential masking effect and co-medication bias. 2) Excluding cases reporting PAH with PAH-approved drugs: we excluded reports with drugs used in PAH to exclude misclassified cases and patients with pre-existing PAH using broad WHO Drug Standardised Drug Groupings: “Drugs for pulmonary arterial hypertension (PAH)” [23]. 3) Modified comparator group: we restricted the comparator group to all drugs belonging to the Anatomical Therapeutic Chemical third level of the drug of interest to assess confounding by indication. 4) Healthcare professionals only: we restricted the analysis to cases reported by healthcare professionals only (physician, pharmacist and other health professional according to WHO category). 5) Adjusted for gender and age: we adjusted by stratification the IC estimates on age and sex to limit confounding by these factors.

Results

Patients from the French PH Registry

We identified seven cases of PAH in a context of bevacizumab exposure: four females and three males. One additional patient was diagnosed with chronic thromboembolic PH and was therefore excluded from the analysis. The median (IQR) age at PAH diagnosis, in those seven patients, was 61.3 (58.7–68.6) years. The median (IQR) duration of bevacizumab exposure was 15.2 (2.9–17.4) months. The indications for bevacizumab were hepatocellular carcinoma (n=2), ovarian cancer (n=3), breast cancer (n=1) and colorectal cancer (n=1). None of the patients had been treated with drugs known to be associated with PAH.

One patient developed PAH after bevacizumab exposure during follow-up in the DELPHI-2 screening programme [24], which is dedicated to apparently healthy individuals carrying a BMPR2 mutation and therefore at risk of developing incident heritable PAH. Two other patients had cirrhosis with portal hypertension and one had mild COPD (i.e. mild obstructive impairment without significant diffusion limitation (forced expiratory volume in 1 s (FEV₁)/forced vital capacity ratio <0.7 and FEV₁ 57% predicted with mild hypoxaemia)). Clinical, functional, biological and haemodynamic characteristics are summarised in table 1. The median (IQR) 6MWD was 351 (289–378) m. Haemodynamic studies showed severe pre-capillary PH with median (IQR) mPAP 47.0 (44.5–49) mmHg, RAP 5.0 (3.5–10.5) mmHg, cardiac index 2.6 (2.1–3.0) L·min⁻¹·m⁻² and PVR 8.5 (7.0–10.1) WU.

TABLE 1.

Characteristics of the seven identified patients with pulmonary arterial hypertension (PAH) and bevacizumab exposure in the French Pulmonary Hypertension (PH) Registry

	Case 1	Case 2	Case 3	Case 4	Case 5	Case 6	Case 7
Age at PAH diagnosis, years	57	74	76	61	56	60	63
Bevacizumab indication	Hepatocellular carcinoma	Hepatocellular carcinoma	Ovarian cancer	Ovarian cancer	Ovarian cancer	Breast cancer	Colorectal cancer
Duration of treatment, months	0.7	3.3	18.6	15.2	16.2	65.9	2.6
Other condition associated with or risk factor of PH	Cirrhosis	Cirrhosis		Mild COPD		BMPR2 mutation
NYHA FC	III	II	IV	III	III	III	IV
6MWD, m	358	354		270	385	420	120
mPAP, mmHg	42	64	50	46	47	48	43
PAWP, mmHg	10	8	5	12	8	7	11
Cardiac index, L·min−1·m−2	5.2	3.2	1.0	2.6	2.4	2.8	1.8
PVR, WU	3.4	9.6	27.9	6.5	10.5	8.5	7.4
NT-proBNP, ng·L−1	272	232				528
BNP, ng·L−1				68	303		397
DLCO, % pred	Unknown	56	Unknown	>80	54	46	74
PAH-approved drugs	None	PDE5i	ERA+ prostacyclin	PDE5i+ERA	PDE5i+ERA+ prostacyclin	PDE5i+ERA	PDE5i+ERA
First re-evaluation#
NYHA FC		II		II	III	III
6MWD, m		440		323	370	448
mPAP, mmHg	60	38			45	43
Cardiac index, L·min−1·m−2	6.9	3.6			2.4	4.7
PVR, WU	2.3	4.3			10.1	4
Vital status	Died of neoplasm	Alive	Died of right heart failure	Died	Died	Died	Died of neoplasm

NYHA FC: New York Heart Association Functional Class; 6MWD: 6-min walk distance; mPAP: mean pulmonary arterial pressure; PAWP: pulmonary arterial wedge pressure; PVR: pulmonary vascular resistance; NT-proBNP: N-terminal pro-brain natriuretic peptide; BNP: brain natriuretic peptide; D_LCO: diffusing capacity of the lung for carbon monoxide; PDE5i: phosphodiesterase 5 inhibitor; ERA: endothelin receptor antagonist. ^#: first re-evaluation after a median (interquartile range) of 8.0 (7.2–9.2) months.

In all patients, bevacizumab was discontinued following PAH diagnosis. Six of them were subsequently treated with PAH-approved drugs: monotherapy (n=1), dual therapy (n=4) or triple therapy (n=1) (table 1). Re-evaluation after a median (IQR) of 8.0 (7.2–9.2) months showed marked improvement in clinical, functional, biological and haemodynamic parameters (figure 1).

FIGURE 1.

Evolution between diagnosis of pulmonary arterial hypertension (visit 0) and the first re-evaluation (visit 1). a) Dyspnoea evaluated by New York Heart Association Functional Class (NYHA FC). b) 6-min walk distance (6MWD). c) Cardiac biomarkers (brain natriuretic peptide (BNP) or N-terminal pro-BNP). d) Pulmonary vascular resistance (PVR). The first re-evaluation occurred after a median (interquartile range) of 8.0 (7.2–9.2) months.

Analysis of the French national healthcare database (SNDS)

Among the 9964 patients identified as the PAH cohort, 30 had received at least one bevacizumab dispensation prior to the date of PAH diagnosis. The median (IQR) age at PAH diagnosis was 65.5 (60.3–71.3) years; 53% were women. A history of cancer preceding bevacizumab exposure was identified in 83% of these patients. The median (IQR) duration of bevacizumab exposure was 4.8 (1.0–13.7) months. Among PAH patients with exposure to bevacizumab, 20 (66.7%) had a previous diagnosis of COPD, while 2215 (22.3%) had COPD in the group of patients with PAH without exposure to bevacizumab. Additional demographic and clinical characteristics, including comorbidities and cancer type, for bevacizumab-exposed versus unexposed patients in the PAH cohort are detailed in table 2. In the SNDS database, 12.51 patients per 100 000 exposed to bevacizumab developed PAH, which is similar to the prevalence of PAH in the general population (11.64 per 100 000) (table 3).

TABLE 2.

Description of the French National Health Data System (SNDS) pulmonary arterial hypertension (PAH) cohort, with and without bevacizumab dispensation before PAH diagnosis

	PAH with exposure to bevacizumab (n=30 (0.3%))	PAH without exposure to bevacizumab (n=9934 (99.7%))
Female	16 (53.3)	4996 (51.9)
Age at PAH diagnosis, years	65.5 (60.3–71.3)	67 (57–74)
PAH-associated conditions
Systemic scleroderma	0 (0.0)	1124 (11.3)
Portal hypertension	3 (10.0)	592 (6.0)
Congenital heart disease	1 (3.3)	367 (3.7)
HIV positive	0 (0.0)	273 (2.7)
Comorbidities
Hypertension	29 (96.7)	8609 (86.7)
Heart failure	10 (33.3)	3764 (37.9)
Diabetes	13 (43.3)	3310 (33.3)
COPD	20 (66.7)	2215 (22.3)
Venous thromboembolic events	4 (13.3)	1402 (14.1)
Chronic dialysis	0 (0.0)	175 (1.8)
Sickle cell disease	2 (6.7)	35 (0.4)
Cancer	20 (66.7)	1736 (17.5)
Digestive organs	12 (40.0)	804 (8.1)
Breast	4 (13.3)	328 (3.3)
Respiratory organs	4 (13.3)	612 (6.2)
Female genital organs	3 (10.0)	130 (1.3)
Urinary tract	2 (6.7)	355 (3.6)
Male genital organs	1 (3.3)	302 (3.0)
Duration of exposure, months	4.8 (1.0–13.7)
Time between last cancer diagnosis and first exposure to bevacizumab, years	1.6 (0.6–3.7)

Data are presented as n (%) or median (interquartile range).

TABLE 3.

Prevalence of pulmonary arterial hypertension (PAH) in the general French general population and in patients exposed to bevacizumab

	Individuals, n	PAH, n	Prevalence of PAH per 100 000
General population	85 600 842	9964	11.64
Bevacizumab cohort	239 783	30	12.51

Disproportionality analysis from the WHO pharmacovigilance database (VigiBase)

Among the 33 million ICSRs recorded in VigiBase as of September 2025, 106 574 cases were associated with bevacizumab as either a suspect or interacting drug. We identified 21 cases of PAH associated with bevacizumab exposure and 152 cases of PH using broader definitions. Nearly all PAH cases (95.2%) were reported by healthcare professionals, and 16 (76.2%) originated from France. Among these, 12 (57.1%) cases were female. The median (IQR) age was 61.5 (49.8–67.2) years and the median (IQR) time to PAH onset was 384 (187–672) days. Case characteristics for other VEGF inhibitors are presented in supplementary table S5. No cases were identified for vandetanib, axitinib, ramucirumab and aflibercept.

Bevacizumab was associated with a significant signal of disproportionate reporting only when using the broad definition of PH (IC 0.4 (95% CrI 0.1–0.6)) (figure 2) and in three out of five sensitivity analyses (supplementary figure S1). A significant signal of disproportionate reporting was found for nintedanib and ponatinib in both the specific analysis (IC 1.7 (95% CrI 1.2–2.1) and IC 1.8 (95% CrI 0.9–2.4), respectively) and the broad analysis (IC 2.7 (95% CrI 2.5–2.9) and IC 2.1 (95% CrI 1.7–2.5), respectively).

FIGURE 2.

Result of disproportionality analysis (information component (IC) with 95% credibility interval (95% CrI)) in the VigiBase database for pulmonary arterial hypertension (PAH) reported with vascular endothelial growth factor inhibitors using a narrow definition (MedDRA Preferred Term “Pulmonary arterial hypertension”) and a broad definition (collection of terms related to PAH; see Methods).

Discussion

This article is the first, to the best of our knowledge, to summarise the available clinical evidence for a potential association between bevacizumab and PAH in adults. Collectively, these results do not establish a relationship between bevacizumab exposure and PAH in adult patients. Indeed, the majority of patients exposed to bevacizumab in the French PH Registry had other identified predisposing conditions for PH (i.e. BMPR2 mutation, portal hypertension and mild COPD), pharmacovigilance disproportionality analyses are inconsistent and we did not find an increased prevalence of PAH in patients exposed to bevacizumab compared to the whole French population using the French national healthcare database (SNDS). Some patients may also have PH associated with thrombotic microangiopathy, which may be difficult to diagnose with certainty [25]. However, the potential role of bevacizumab in children and as a secondary trigger of PH remains to be further explored.

In 2020, an Austrian team described for the first time a proliferative pulmonary vasculopathy induced by bevacizumab, in a translational study. In kdr-deleted mice (i.e. coding for VEGFR-2), a mild PH phenotype (significant increase in right ventricular systolic pressure, pulmonary vascular thickness and vessel occlusions with proliferative cells) was observed when mice were exposed to a normoxic environment and a severe PH phenotype was observed when exposed to hypoxia [26]. In this model and in 34 patients, interruption of VEGF-A signalling by bevacizumab led to a significant increase in serum VEGF levels [26]. Moreover, in three patients that had pulmonary metastasectomy associated with colorectal cancer treated with bevacizumab, proliferative pulmonary vascular alterations were observed. This translational study was the first to describe evidence of PH in kdr-knockout rodents exposed to hypoxia and vascular remodelling and in humans treated with bevacizumab [26]. In case series where haemodynamic assessment was described, pre-capillary PH of similar severity to that described in the French PH Registry was observed, with, in all cases, a poor prognosis. Diagnosis of portopulmonary hypertension was systematically considered in patients with PH in a context of liver disease. In Cases 1 and 2, portopulmonary hypertension was indeed retained due to liver disease, alcohol consumption and arguments for portal hypertension on liver echography.

In patients from the SNDS database who had been exposed to bevacizumab, most of them had COPD as a comorbidity. This represented the main difference compared with PAH patients who had not been exposed to bevacizumab and is probably explained by the association of tobacco with several cancers treated by bevacizumab.

Disproportionality analysis from the VigiBase database found no safety signal for bevacizumab using the narrow definition and in some sensitivity analyses despite the widespread use of bevacizumab and prolonged monitoring of its safety. Nintedanib had a stronger signal but probably biased by confounding by indication due to the development of PH associated with pulmonary fibrosis. Ponatinib also displayed a strong signal of disproportionate reporting and has already been identified by the WSPH as a possible cause of PAH [22, 27].

This study does not suggest a robust link between PAH and exposure to bevacizumab in adults. However, we cannot rule out that bevacizumab may act as a second effector to trigger PAH in predisposed patients. Indeed, most of the patients identified in the French PH Registry had a predisposing condition. In another population of patients with a predisposition to PH (i.e. pre-term infants, with incomplete lung development), more infants who received bevacizumab were treated for PH but, after adjustment for year of treatment and hospital, this was not statistically significant [8]. Intravitreal bevacizumab also led to lower serum VEGF levels for up to 8 weeks after injection, supporting the systemic passage [8, 28, 29]. However, the role of serum VEGF levels in the development of PAH remains to be further explored, given the discrepancies found with the results of Winter et al. [26]. It is noteworthy that in the only randomised study comparing anti-VEGF with laser photocoagulation, four infants died (out of seven deaths) in the bevacizumab group due to “low oxygen” or “respiratory failure” [30]. In intravitreal bevacizumab-treated infants, out of 15 patients, eight (53.3%) had PH following bevacizumab therapy in contrast to zero infants (out of 17), when treated with laser therapy [31].

Our study has limitations. Regarding the French PH Registry, collected data are retrospective and some patients may not have been identified in the registry due to the limited accessibility of RHC in patients with metastatic neoplasia due to their fragility or poor prognosis, which may have led to underdiagnosis. Exposure data to all PAH-associated drugs are not systematically recorded and may have also led to bias. Analysis of the French national healthcare database (SNDS) allows coverage of all of the French population with systematic recording of drug exposures and hospitalisations, and reliable estimation of events incidence and quantification of drug-related risk. However, the algorithm used to identify PAH, although validated, does not contain haemodynamic data, and in-hospital drugs that are not specifically reimbursed may not be available.

Disproportionality analysis from the WHO pharmacovigilance database (VigiBase) enables worldwide coverage and a very flexible system of case reporting and quantification of strength of the signal of disproportionate reporting. However, no clinical, haemodynamic or genetic data are available. In a previous study we found 75% of PAH confirmed by RHC for French cases [32]; however, given we do not have access to case narratives in other countries, performances are unknown in other countries and heterogeneity between countries in PAH coding practices has not yet been evaluated. Moreover, the number of cases reported in pharmacovigilance databases and disproportionality estimates can be influenced by several biases such as selective reporting, confounding bias or competitions between drugs and/or events [33, 34].

In conclusion, we have not identified a signal between exposure to bevacizumab and the development of PAH in adult inpatients in the general population. Bevacizumab alone is probably not sufficient to cause de novo PAH, but it is necessary to continue collecting data on these patients and to remain vigilant, as bevacizumab could act as a second trigger in patients with a risk factor for developing PH.

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material

00209-2026.SUPPLEMENT