QPPV.COM

Evaluating the Fluoroquinolone–Aortic Aneurysm and Dissection Signal: From Signal Detection to Regulatory Action

A detailed evaluation of how the fluoroquinolone–aortic aneurysm and dissection signal evolved from an epidemiological concern into regulatory action, including the evidence that supported the signal and later studies that challenged its causal interpretation.

Audio Lesson 10 min
Knowledge Assessment Test your understanding of this article. Take the assessment →

Evaluating the Fluoroquinolone–Aortic Aneurysm and Dissection Signal: From Signal Detection to Regulatory Action

Introduction

A pharmacovigilance signal is not a conclusion.

It is an indication that an association between a medicinal product and an event may represent a previously unrecognised or insufficiently characterised risk and therefore requires further evaluation. The distinction is important because the evidence that identifies a signal is often substantially weaker than the evidence required to establish causality.

The history of the association between systemic fluoroquinolones and aortic aneurysm or aortic dissection provides a useful example.

The concern emerged from epidemiological evidence suggesting that fluoroquinolone exposure was associated with an increased risk of aortic aneurysm and dissection. The European pharmacovigilance system subsequently evaluated the association, and the Pharmacovigilance Risk Assessment Committee (PRAC) recommended changes to product information. Later studies, however, raised important questions about confounding by infection, indication and differential diagnostic surveillance.

The resulting regulatory history illustrates an important principle for signal management:

A signal can be sufficiently credible to justify regulatory action even when the available evidence does not establish a simple or completely quantified causal relationship.

It also demonstrates why signal evaluation should not stop when an initial regulatory conclusion has been reached. New epidemiological evidence can change the interpretation of the association, even when the underlying safety concern remains relevant to clinical decision-making.

This article reconstructs the evaluation of the fluoroquinolone–aortic aneurysm and dissection signal using publicly available regulatory documents and published epidemiological evidence.

The purpose is not to reproduce the PRAC assessment or to make a new regulatory determination. It is to examine the evidence as a pharmacovigilance professional would: identifying the signal, examining the supporting evidence, testing alternative explanations, following the regulatory reasoning and assessing how the product information evolved.

The Signal

What Was the Safety Concern?

The signal concerned an association between systemic and inhaled fluoroquinolones and:

Aortic aneurysm and aortic dissection are clinically serious conditions. Both can be life-threatening, particularly when rupture occurs.

The initial pharmacovigilance question was therefore not whether fluoroquinolones caused every case of aortic aneurysm or dissection.

The relevant question was narrower:

Was there sufficient evidence that exposure to fluoroquinolones increased the risk of these events to justify further regulatory assessment and possible risk-minimisation measures?

That distinction is fundamental to signal management.

Why the Signal Was Plausible

Biological Considerations

A biological hypothesis existed for an association between fluoroquinolone exposure and connective-tissue-related pathology.

Fluoroquinolones had already been associated with tendon disorders, including tendonitis and tendon rupture. Tendons and the aortic wall both contain substantial extracellular matrix components, including collagen.

Non-clinical work also provided a potential mechanistic basis for concern about effects on connective tissue.

However, biological plausibility is supportive rather than determinative.

A plausible mechanism cannot establish that an epidemiological association is causal. It can, however, increase the credibility of a signal and make further investigation appropriate.

The Early Epidemiological Evidence

The early signal was strengthened by observational epidemiological studies.

A 2017 systematic review and meta-analysis identified two observational studies and found statistically significant associations between current fluoroquinolone use and both aortic dissection and aortic aneurysm. The pooled odds ratio was 2.79 for aortic dissection and 2.25 for aortic aneurysm. The authors described the evidence as suggesting a small but significant increased risk, while also recognising that the number of studies was limited. [1]

This was important signal-generating evidence.

It was not, however, equivalent to randomised evidence.

The studies were observational, and aortic aneurysm and dissection are strongly associated with age, cardiovascular disease and other patient characteristics. Fluoroquinolones are also prescribed in particular clinical circumstances, and the underlying infection may itself influence the risk of the outcome.

Consequently, several alternative explanations needed to be considered.

The 2016 PRAC Signal

The association had already reached the European pharmacovigilance system before the major 2018 review.

In September 2016, PRAC listed:

Fluoroquinolones for systemic use — aortic aneurysm and dissection

as a safety signal with EPITT number 18651.

At that stage, the recommendation was routine pharmacovigilance rather than an immediate product-information change. [2]

This is an important point.

The presence of a signal in the PRAC recommendations does not mean that PRAC has concluded that the association is causal.

In this case, the regulatory action at that stage was consistent with continued surveillance rather than immediate amendment of the product information.

The signal therefore illustrates an intermediate stage of the pharmacovigilance process:

signal identified → evidence monitored → further evidence accumulated → regulatory reassessment.

What Changed the Assessment?

The evidence base continued to develop.

A major contribution came from the Swedish nationwide cohort study published in 2018.

The Swedish Cohort Study

Pasternak and colleagues compared fluoroquinolone treatment episodes with amoxicillin treatment episodes using nationwide Swedish registry data.

The study was designed to address confounding by indication by using an active antibiotic comparator rather than comparing fluoroquinolone users with untreated individuals.

The investigators followed patients for 60 days after treatment initiation and assessed first diagnoses of aortic aneurysm or dissection. [3]

This study was particularly relevant to signal evaluation because the comparator strategy attempted to answer a more clinically meaningful question:

Among patients requiring antibiotic treatment, is the risk different in those receiving a fluoroquinolone compared with those receiving another antibiotic?

That is generally more informative than simply comparing exposed patients with an unexposed population.

The study reported an increased risk associated with fluoroquinolone treatment compared with amoxicillin.

The findings therefore strengthened the epidemiological basis of the signal.

At the same time, the study did not eliminate all possible sources of bias.

The 2018 PRAC Assessment

In September 2018, PRAC formally reassessed the fluoroquinolone–aortic aneurysm and dissection signal.

The PRAC recommendation explicitly considered:

The studies cited by PRAC included work by Lee and colleagues, Daneman and colleagues, and Pasternak and colleagues, together with non-clinical evidence from LeMaire and colleagues. [4]

PRAC concluded that the evidence justified amendment of the product information.

The recommendation was not simply:

“Fluoroquinolones cause aortic aneurysm.”

Instead, the regulatory wording was framed around epidemiological evidence reporting an increased risk, particularly in older patients, together with reported cases and a precautionary benefit-risk assessment.

This distinction is important.

Regulatory product information often deliberately uses language that communicates a clinically important association without claiming a level of causal certainty that the evidence cannot support.

What Was Added to Product Information?

The PRAC recommendation resulted in a variation to the product information for systemic and inhaled fluoroquinolones.

The proposed wording stated that epidemiological studies reported an increased risk of aortic aneurysm and dissection following fluoroquinolone exposure, particularly in older people.

The product information therefore moved from the earlier absence of a specific warning to an explicit risk statement.

This represents a major pharmacovigilance transition:

epidemiological signal → regulatory evaluation → risk communication.

The product information did not need to establish an exact attributable risk for every patient before a warning could be justified.

The regulatory decision was based on the combination of:

The Evidence Was Not Uniform

The history becomes more interesting after the regulatory action.

A good signal evaluation must actively search for evidence that contradicts the initial hypothesis.

Simply collecting studies that support the signal creates confirmation bias.

Several subsequent studies raised important concerns about confounding.

Confounding by Infection

The Infection May Be Part of the Causal Story

One of the most important developments was recognition that infection itself may be associated with aortic aneurysm or dissection.

This creates a major epidemiological problem.

Suppose a patient develops an infection and receives a fluoroquinolone.

The patient subsequently develops an aortic event.

There are at least three broad possibilities:

  1. the fluoroquinolone contributed to the event;
  2. the infection contributed to the event;
  3. another patient characteristic contributed to both the choice of antibiotic and the event.

If the analysis does not adequately separate these possibilities, an association between the antibiotic and the event may be exaggerated.

This is classic confounding by indication.

The 2020 Infection-Adjusted Analysis

Dong and colleagues conducted a large nationwide nested case-control study specifically examining the role of infection and fluoroquinolone exposure.

The study included 28,948 cases and 289,480 matched controls from a population of more than 21 million adults.

The investigators found that indicated infections themselves were associated with an increased risk of aortic aneurysm or dissection.

The adjusted odds ratio for any indicated infection was 1.73.

More importantly for the drug-safety question, among patients with indicated infections, fluoroquinolone use was not associated with an increased risk compared with:

The reported odds ratios were 1.01 and 0.88 respectively. [5]

This substantially complicated the interpretation of the earlier signal.

Why This Study Matters

The result does not prove that fluoroquinolones have no causal relationship with aortic disease.

It demonstrates something more specific:

When the analysis better accounts for the underlying infection and compares fluoroquinolones with antibiotics used for similar indications, the previously observed association can substantially diminish or disappear.

That is an important distinction.

A signal evaluation should therefore not ask only:

"Is there an association?"

It should also ask:

"Does the association survive increasingly appropriate attempts to remove plausible sources of bias?"

This is one of the most important lessons from the case.

Surveillance Bias

Another concern is differential surveillance.

Aortic aneurysms are frequently detected incidentally through imaging performed for other clinical reasons.

Patients receiving different antibiotics may have different patterns of imaging depending on:

If patients receiving one antibiotic are more likely to undergo imaging, previously undiagnosed aneurysms may be detected more frequently.

The resulting difference in observed event rates may then be partly attributable to detection rather than drug toxicity.

This is surveillance bias.

The 2020 US Cohort Study

Gopalakrishnan and colleagues examined this issue using two large propensity-score-matched cohorts.

The study included patients aged 50 years or older with either:

Fluoroquinolone users were compared with active antibiotic comparators.

The findings differed according to indication.

Among patients treated for pneumonia, an increased rate of aortic aneurysm or dissection was observed compared with azithromycin.

Among patients treated for urinary tract infection, no increased rate was observed compared with trimethoprim-sulfamethoxazole.

The authors also performed an analysis designed to address differential imaging.

When baseline imaging was required, the association with amoxicillin as comparator was substantially attenuated, with a hazard ratio of 1.13 and a 95% confidence interval of 0.96 to 1.33. [6]

This is highly informative for pharmacovigilance interpretation.

An association that weakens after adjustment for a plausible source of detection bias deserves a different level of causal confidence than an association that remains robust.

The Importance of Active Comparators

The fluoroquinolone signal also demonstrates why active-comparator studies are valuable.

Consider three possible comparisons:

Comparison 1

Fluoroquinolone users versus people who did not receive antibiotics.

This comparison may be heavily confounded by the underlying infection.

Comparison 2

Fluoroquinolone users versus users of another antibiotic.

This is usually more informative because both groups have an indication for treatment.

Comparison 3

Fluoroquinolone users versus users of another antibiotic for the same infection, with adjustment for severity and other important risk factors.

This can provide substantially stronger control of confounding.

The progression from the first type of comparison toward the third illustrates how an apparently strong pharmacovigilance association can change when epidemiological design improves.

A Signal Can Remain Important Despite Causal Uncertainty

The conflicting evidence does not make the original signal irrelevant.

There is an important distinction between:

causal certainty

and

regulatory relevance.

A regulator may reasonably conclude that a potential serious risk warrants communication or precaution even when the exact magnitude of the causal effect remains uncertain.

This is particularly relevant when:

The regulatory decision therefore cannot be reduced to whether one study produces a statistically significant result.

Evolution of the Product Information

The product information history provides a useful illustration of how signal evaluation translates into regulatory communication.

Stage Evidence / Regulatory Development Product-information implication
2016 PRAC listed aortic aneurysm and dissection as a fluoroquinolone signal Routine pharmacovigilance
2017 Meta-analysis supported an association with aneurysm and dissection Evidence base strengthened
2018 Further epidemiological and non-clinical evidence reviewed by PRAC Product-information amendment recommended
2018 PRAC recommended wording describing increased epidemiological risk, particularly in older people Specific warning introduced
2019 EU-wide regulatory outcome implemented following the broader fluoroquinolone review Restrictions and warnings strengthened
2020 Further PRAC product-information wording incorporated aortic aneurysm/dissection and valve regurgitation/incompetence Risk communication expanded
2020 onward Additional observational evidence highlighted confounding and uncertainty Scientific interpretation became more nuanced
2023–2024 EMA continued to reinforce broader fluoroquinolone risk-minimisation measures Clinical-use restrictions and warnings remained important

The important observation is that product information does not represent a static scientific conclusion.

It is the current regulatory expression of the accumulated evidence and the resulting benefit-risk judgment.

The 2020 Product-Information Expansion

In 2020, PRAC recommended further product-information wording for systemic and inhaled fluoroquinolones covering:

The wording referred to epidemiological studies reporting increased risk, particularly in older patients, and also noted reports of aortic aneurysm and dissection, including fatal ruptures. [7]

This illustrates another important aspect of signal management.

A safety issue may evolve from a single event pair into a broader mechanistic or clinical safety domain.

The evaluation therefore needs to remain open to related outcomes rather than treating the original MedDRA concept as an isolated endpoint.

Evidence Supporting the Signal

The evidence supporting a causal interpretation included several elements.

Epidemiological consistency in early studies

Multiple observational studies reported increased risk estimates.

Temporal relationship

Several analyses focused on relatively short periods following exposure, consistent with the hypothesis of an acute or short-latency effect.

Biological plausibility

Potential effects on connective tissue provided a mechanistic rationale.

Seriousness

Aortic aneurysm and dissection can result in catastrophic outcomes.

Population vulnerability

Older patients and patients with underlying aortic risk factors may have a different baseline risk.

Regulatory convergence

The issue was considered sufficiently credible by the European regulatory system to justify product-information changes and risk-minimisation measures.

Taken together, these factors explain why the signal warranted regulatory action.

Evidence Against a Simple Causal Interpretation

The counter-evidence is equally important.

Confounding by infection

Infection itself may increase the risk of aortic events.

Confounding by indication

Patients receiving fluoroquinolones may differ systematically from patients receiving other antibiotics.

Surveillance bias

Differences in diagnostic imaging can influence detection of previously unrecognised aneurysms.

Active-comparator analyses

Some studies found no increased risk when fluoroquinolones were compared with other antibiotics used for similar indications.

Heterogeneity

Different studies produced different estimates depending on:

These limitations mean that the epidemiological literature does not support a simplistic statement that every observed association represents a direct drug effect.

What Does the Totality of Evidence Support?

A reasonable pharmacovigilance interpretation is more nuanced than either:

"Fluoroquinolones cause aortic aneurysm and dissection."

or:

"The signal was disproven."

The evidence supports a more cautious conclusion.

An association between fluoroquinolone exposure and aortic aneurysm or dissection was sufficiently credible and clinically important to trigger regulatory assessment and product-information changes.

However, subsequent epidemiological studies demonstrated that the magnitude of the association is sensitive to study design and to adjustment for infection, indication and surveillance.

Consequently, the available evidence does not establish a simple, uniformly quantified causal effect applicable to all fluoroquinolone-treated patients.

This distinction is central to a scientifically defensible signal evaluation.

How a Pharmacovigilance Team Could Evaluate the Signal

The case provides a practical framework for evaluating a similar signal.

Step 1: Define the signal precisely

The evaluation should specify:

"Aortic disease with fluoroquinolones" is too broad for rigorous evaluation.

Step 2: Establish the background risk

Before evaluating the drug, determine:

Without a background-risk model, the observed number of cases can be misleading.

Step 3: Review individual cases

For individual case reports, assess:

Individual cases rarely establish population-level causality, but they can identify patterns that inform the epidemiological hypothesis.

Step 4: Review epidemiological evidence

Do not simply count positive studies.

Assess:

A large study with poor comparator selection may provide less useful causal information than a smaller but better-controlled study.

Step 5: Examine contradictory evidence

The evaluation should deliberately search for:

A signal evaluation that does not attempt to falsify its own hypothesis is incomplete.

Step 6: Evaluate biological plausibility

Determine whether:

Mechanistic evidence should be used as one component of the assessment rather than as proof of causality.

Step 7: Assess clinical importance

The seriousness of the event matters.

Even a relatively small potential increase in risk can be clinically important when:

Step 8: Consider regulatory precedent

Review:

The regulatory history provides important context but should not substitute for independent scientific evaluation.

What Should the Signal Conclusion Say?

A useful signal conclusion should separate several dimensions.

Association

Is there evidence of an epidemiological association?

Causality

How strongly does the evidence support a causal relationship?

Clinical importance

How serious would the potential risk be?

Population

Which patients may be particularly vulnerable?

Uncertainty

What important limitations remain?

Regulatory relevance

Does the evidence justify:

These are related but distinct questions.

Why "Signal Confirmed" Is Often Too Crude

A binary conclusion such as:

signal confirmed

can obscure the actual scientific reasoning.

For example, a more informative conclusion might be:

The available evidence supports a clinically relevant association between exposure and the event, but substantial uncertainty remains regarding the magnitude and causal contribution of the medicinal product because of confounding by indication and infection. The evidence is nevertheless sufficient to support precautionary risk communication in patients with relevant baseline risk factors.

That conclusion communicates considerably more information than a binary label.

It also makes clear which part of the evidence is strong and which part remains uncertain.

The Regulatory Threshold Is Not the Same as the Scientific Threshold

Another important lesson from the fluoroquinolone history is that different decisions require different evidentiary thresholds.

The threshold for:

investigating a signal

is relatively low.

The threshold for:

communicating a potential risk

may be higher.

The threshold for:

restricting use

may be higher still.

And the evidence required to make a definitive causal statement may be higher again.

These decisions should not be conflated.

A regulator does not need absolute proof of causality before taking a proportionate precautionary measure when the potential harm is serious and alternatives exist.

Risk Factors and Clinical Context

The regulatory wording also illustrates why signal evaluation should move beyond the crude exposure-event pair.

The risk associated with a medicine is rarely uniform across every patient.

For aortic aneurysm and dissection, relevant considerations include:

The clinical value of a signal evaluation therefore lies partly in determining whether the potential risk can be concentrated into a population where targeted risk minimisation is feasible.

Lessons for QPPV and Signal Governance

The case has several implications for the QPPV.

The QPPV should challenge the first interpretation

A positive disproportionality analysis or epidemiological association is not the end of the evaluation.

The QPPV should expect competing hypotheses

If infection, indication or surveillance could explain the association, those possibilities should be actively investigated.

The QPPV should distinguish evidence from conclusion

The assessment should clearly separate:

The QPPV should understand study design

Pharmacovigilance oversight increasingly requires the ability to recognise important epidemiological limitations.

The QPPV should monitor whether the conclusion remains valid

A signal conclusion made in 2018 may need reconsideration when substantial new evidence becomes available.

The QPPV should connect signal evaluation with risk minimisation

The ultimate objective is not simply to classify signals.

It is to ensure that emerging risks are understood and managed appropriately.

What This Case Teaches About Signal Management

The fluoroquinolone example demonstrates that signal evaluation is iterative.

The sequence was approximately:

initial epidemiological concern

↓

pharmacovigilance signal

↓

additional epidemiological evidence

↓

regulatory assessment

↓

product-information change

↓

new epidemiological evidence

↓

reassessment of confounding and causal interpretation

↓

continued risk communication and risk minimisation

This is how a mature pharmacovigilance system should operate.

The conclusion is not frozen at the moment the signal first appears.

Conclusion

The fluoroquinolone–aortic aneurysm and dissection signal is a useful example of why pharmacovigilance signal evaluation requires more than identifying an association.

The initial evidence was sufficiently concerning to justify regulatory attention. PRAC subsequently evaluated epidemiological and non-clinical evidence and recommended product-information changes.

Later studies, however, demonstrated that the observed association was sensitive to important epidemiological issues, particularly confounding by infection, indication and surveillance.

The appropriate interpretation is therefore neither that the signal was simply "proven" nor that later conflicting studies automatically "disproved" it.

The stronger conclusion is that a clinically serious potential association was sufficiently credible to warrant regulatory risk communication, while the precise magnitude and causal contribution of fluoroquinolone exposure remained subject to important uncertainty.

This distinction is at the heart of good signal management.

A high-quality signal evaluation should answer not only:

"What evidence supports the signal?"

but also:

"What evidence would make the signal less convincing?"

and:

"Does the conclusion remain appropriate when the strongest alternative explanations are tested?"

That is the difference between signal detection and signal evaluation.

Key Takeaways

References

  1. Singh S, Nautiyal A. Aortic Dissection and Aortic Aneurysms Associated with Fluoroquinolones: A Systematic Review and Meta-Analysis. American Journal of Medicine. 2017;130(12):1449-1457. doi:10.1016/j.amjmed.2017.06.029.

  2. European Medicines Agency. PRAC recommendations on safety signals adopted at the PRAC meeting of 26–29 September 2016. EPITT No. 18651: Fluoroquinolones for systemic use – aortic aneurysm and dissection. EMA/PRAC/634818/2016.

  3. Pasternak B, Inghammar M, Svanström H. Fluoroquinolone use and risk of aortic aneurysm and dissection: nationwide cohort study. BMJ. 2018;360:k678. doi:10.1136/bmj.k678.

  4. European Medicines Agency. PRAC recommendations on signals adopted at the PRAC meeting of 3–6 September 2018. EPITT No. 18651: Fluoroquinolones for systemic and inhaled use – aortic aneurysm and dissection. EMA/PRAC/414645/2018.

  5. Dong Y-H, et al. Association of Infections and Use of Fluoroquinolones With the Risk of Aortic Aneurysm or Aortic Dissection. JAMA Internal Medicine. 2020. doi:10.1001/jamainternmed.2020.4199.

  6. Gopalakrishnan C, et al. Association of Fluoroquinolones With the Risk of Aortic Aneurysm or Aortic Dissection. JAMA Internal Medicine. 2020. doi:10.1001/jamainternmed.2020.4199.

  7. European Medicines Agency. New product information wording – extracts from PRAC recommendations on signals adopted at the 31 August–3 September 2020 PRAC meeting. Fluoroquinolones for systemic and inhalation formulations – heart valve regurgitation, cervical artery dissection, and aortic aneurysm and dissection. EPITT No. 19522.

  8. Rawla P, El Helou ML, Vellipuram AR. Fluoroquinolones and the Risk of Aortic Aneurysm or Aortic Dissection: A Systematic Review and Meta-Analysis. Current Vascular Pharmacology. 2019;17(1):3-10. doi:10.2174/1871525717666190402121958.

  9. Noman AT, Qazi AH, Alqasrawi M, et al. Fluoroquinolones and the risk of aortopathy: A systematic review and meta-analysis. International Journal of Cardiology. 2019;274:299-302. doi:10.1016/j.ijcard.2018.09.067.

  10. Gopalakrishnan C, et al. Association Between Fluoroquinolone Use and the Risk of Aortic Aneurysm or Aortic Dissection. JAMA Internal Medicine. 2020.

  11. European Medicines Agency. Quinolone- and fluoroquinolone-containing medicinal products – referral. Article 31 referral; PRAC recommendation and subsequent regulatory outcome.

  12. European Medicines Agency. Fluoroquinolone antibiotics: reminder of measures to reduce the risk of long-lasting, disabling and potentially irreversible side effects. 2023.

  13. European Medicines Agency. Evaluation and monitoring of medicines: highlights. EMA Annual Report 2024. Safety monitoring and subsequent product-information changes for fluoroquinolones.

  14. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP), Module IX – Signal management.

Revision History

Related Articles

Historical Signal Evaluation: Fluoroquinolones and Aortic Aneurysm or Dissection

A detailed evaluation of the fluoroquinolone–aortic aneurysm or dissection signal, examining epidemiological evidence, biological plausibility, confounding, susceptible populations, regulatory assessment and changes to product information.

Evaluating the Statin–Rhabdomyolysis Signal: From Spontaneous Reports to Dose, Drug Interactions and Causal Assessment

How the statin–rhabdomyolysis signal was evaluated, why raw reporting numbers can mislead, and how dose, pharmacokinetics, drug interactions and patient factors determine the clinical interpretation of muscle toxicity.

How to Read an EMA Referral From a Pharmacovigilance Perspective

How to reconstruct the pharmacovigilance reasoning behind an EMA referral, from the initial safety concern and evidence assessment through PRAC recommendations, regulatory decisions and implementation in product information.