The efficacy of azithromycin in the treatment of influenza A infection: A systematic review

1. Introduction

Influenza is a communicable respiratory viral infection affecting the upper respiratory tract and occasionally the lower tract, causing pneumonia (Boktor, Hafner et al. 2023, Moghadami 2017). In addition, Influenza is highly infectious with an incubation period of 1 – 3 days and, in healthy individuals, is usually a self-limiting disease with recovery usually within 2 to 7 days (Bhatt, Singh et al. 2021). The enveloped Influenza viruses are four distinct strains that are composed of single-stranded RNA chains and have high mutation rates and antigenic variances. Annually, Influenza types A and B infect individuals throughout the winter season, also known as the epidemic season (Boktor, Hafner et al. 2023). Four influenza pandemics have arisen since the early 1900s, namely in 1918 (Spanish Influenza H1N1), 1957 (Asian Influenza H2N2), 1968 (Hong Kong Influenza H3N2), and 2009 (H1N1) (Harrington, Kackos et al. 2021).

Influenza virus infections persistently lead to high levels of morbidity and mortality and impose a financial strain on healthcare systems and societies. The World Health Organization (WHO) estimated the annual occurrence of Influenza A infections to be about 3-5 million cases, with a death toll of 290,000-650,000 individuals globally (Mostafa 2023). While the influenza virus can potentially infect all age groups, a higher risk of contracting the infection is reported in adults aged 18-64 years. Moreover, life-threatening cases of illness are more evident among pregnant women associated with perinatal mortality, prematurity, smaller neonatal size and lower birth weight (Pierce, Kurinczuk et al. 2011), young children under 5 years of age, elderly patients aged 65 years and above (de Courville, Cadarette et al. 2022), and those with underlying health conditions such as respiratory or cardiac disease, chronic neurological conditions, and immunosuppression (Pebody, Andrews et al. 2013). According to the Centers for Disease Control and Prevention (CDC), fatalities within the elderly population accounted for 70% - 85% of the total Influenza-related mortality (Centers for Disease Control and Prevention 2024). Additionally, a study published in 2017 stated that influenza-related lower respiratory tract infection (LRTI) accounted for an estimated 145,000 deaths, 9,459,000 hospitalizations, and 81,536,000 days of hospitalization (Troeger, Blacker et al. 2019). Also, an earlier study reported that 65% of the economic impact attributed to vaccine-preventable illnesses is caused by Influenza infection (Ozawa, Portnoy et al. 2016).

The manifestations of seasonal Influenza can be noticed across a spectrum, ranging from asymptomatic infection or moderate upper respiratory symptoms (dry cough and nasal congestion for 3-7 days), with or without fever, to more significant complications, including but not limited to secondary bacterial infections, pneumonia, bronchiolitis, croup, myocarditis, myositis, rhabdomyolysis, encephalopathy, encephalitis, renal failure, respiratory failure, acute respiratory distress syndrome, and septic shock (Tyrrell, Allen et al. 2021). Intensive care unit (ICU) admission among influenza patients is correlated with pre-existing illnesses such as obstructive/central sleep apnoea syndrome, myocardial infarction, and body mass index (BMI) > 30, where a higher incidence of pulmonary co-infection is described (Beumer, Koch et al. 2019).

The development of influenza complications and the mortality rates are contributed by the infection-associated immune reactions. According to Jiang and Zhang (2023), overactivation of the innate immune system due to the influenza virus could yield a state of excessive cytokines release and aggressive proinflammatory effects that are responsible for mortality and morbidity (Jiang, Zhang 2023). Additionally, it was found that increased systemic levels of pro-inflammatory cytokines constituted a hallmark in severe cases of pandemic (H1N1) 2009” (Bermejo-Martin, Ortiz de Lejarazu et al. 2009, To, Hung et al. 2010). This serves as a potential target in the management of Influenza by medications that show immunomodulatory effects such as Macrolides (Viasus, Paño-Pardo et al. 2011).

As stated in the latest World Health Organization (WHO) recommendations, the current guidelines for the treatment of influenza A include antivirals and vaccination (World Health Organization 2021). Antiviral medications, specifically neuraminidase inhibitors such as oseltamivir, zanamivir, and peramivir, reduce the symptoms’ duration and severity, decrease the risk of complications, and enhance the treatment outcomes. Given that influenza is a vaccine-preventable disease, influenza vaccines stand as the most efficient approach to managing the symptoms of seasonal influenza, particularly among vulnerable groups and healthcare workers. It is mandatory to regularly update the vaccines to align with the repetitive evolution of circulating viruses (de Courville, Cadarette et al. 2022). Additionally, other alternatives for influenza treatment, such as corticosteroids, macrolides, and passive immunotherapy, are not favoured according to WHO due to lack of evidence (World Health Organization 2021).

The recent viral pandemics repositioned macrolides for their potential use as anti-influenza agents. Macrolides is a group of naturally derived compounds consisting of a lactone ring with attached deoxy sugars. Some of the macrolides, such as azithromycin (AZ), possess antibiotic activity through binding to the bacterial 50S ribosomal subunit, which results in the interruption of bacterial protein synthesis (Arsic, Barber et al. 2018). This action is mainly bacteriostatic, and it can also become bactericidal at high concentrations (Zarogoulidis, Papanas et al. 2012). Various animal studies proposed a potential positive effect of AZ against viral infections. Dat Huu Tran et al. (2019) showed that a single administration of AZ intranasally to mice infected with A(H1N1)pdm09 virus had effectively minimized viral load in the lungs and alleviated the infection-induced hypothermia (Tran, Sugamata et al. 2019). Other macrolides showed a favourable impact on preventing lung disease and viral proliferation in mice by their ability to control the release of neutrophil myeloperoxidase (which has been implicated in the pathogenesis of severe influenza-induced pneumonia), pro-inflammatory cytokines, and interferon-α production in the lung (Sugamata, Sugawara et al. 2014).

The WHO-based recommendation against the use of macrolides reflects the scattered few studies on human subjects or clinical trials about the effect of macrolides on viral infections, especially influenza. Furthermore, the WHO guidelines were generally based on a study with very low-quality evidence randomized control trial (World Health Organization 2021). Here, we aim to test the evidence for the recommendation of the usage of AZ in the treatment of influenza, particularly in light of recent observations that the use of AZ was reported despite the WHO’s unfavourable recommendations.

2. Methods

This systematic review is reported per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to test the evidence of AZ use in the treatment regimen of Influenza A infections (Figure 1). The PRISMA checklist is also included in the extended data depository. The systematic review was not registered. To compile a comprehensive dataset, a thorough search strategy was implemented across multiple electronic databases, including PubMed, ProQuest, Scopus, and Embase (August 2023). The chosen databases provided a broad scope, allowing us to capture a diverse range of studies relevant to our review topic.

Figure 1. PRISMA chart outlining the review process.

The search strategy employed a combination of keywords, including “azithromycin”, and “influenza A”. Only studies about the Influenza “A” strain were included, and all studies focused on other strains were excluded. We deliberately imposed no limitations on participant age or recruitment location, encompassing community, outpatient, and hospital settings. To enhance precision, additional filters such as “full-text”, “research article”, “scholarly journals”, and “peer-reviewed” were applied. We restricted results to publications between 2007 and 2023 and set the language criterion to English publications specifically. The deliberate choice aimed to eliminate potential language barriers and ensure a uniform screening process.

The systematic review methodology involved a two-stage deduplication process, excluding 1544 redundant entries out of 6145 obtained titles. The initial phase of deduplication utilized Excel and yielded an initial pool of 5141 articles (1004 deduplicated automatically). Subsequently, manual deduplication was performed which reduced this pool to 4601 articles. This meticulous approach ensured the dataset’s integrity and reliability for comprehensive analysis.

To manage the extensive dataset, a systematic approach to screening was adopted. The six co-authors were organized into three pairs; each assigned the responsibility of independently screening titles and abstracts for all the search-returned publications. This is to determine the eligibility of studies for data retrieval by checking if they met our study criteria. Any conflicts or disagreements during the screening phase were resolved through consultation with a third opinion. After excluding 4549 articles by title and abstract review, 52 reports were sought for retrieval and full-text review. The same pairs reviewed the full text of the included studies, and the required data was extracted from each. The full-text review resulted in the further exclusion of 47 studies and 5 records for further steps. The manual search in the excluded review resulted in another 2 articles qualified for the next step of data extraction and evidence testing bringing the total included studies to 7.

Data Extraction: A structured data extraction form (Google form) was used to collect data elements of each study and extracted them into a Microsoft Excel (Microsoft® Excel® for Microsoft 365 MSO (Version 2210 Build 16.0.15726.20188) 64-bit) worksheet (the worksheet is provided in an alternative free format; Google Sheets, refer to the Data availability statement). This form encompassed key elements such as study design (i.e., prospective, retrospective, case-control, cohort study, etc …), country of conduct, sample size, publication year, study duration, primary outcomes and measurement method, secondary outcomes (if any), research question, and the journal of publication. Each divided pair independently extracted study data from included publications, and the results were then compared. Two studies were excluded after full-text review as they failed to report AZ specifically in the treatment. A third party resolved differences. Data sources were carefully reviewed, and the extracted data was used in our analysis.

Assessment of bias: After completing data extraction, we noticed that all our included articles lay under the study design of either randomized controlled trials, cohort studies, or case-report series. Thus, an Evidence Appraisal Tool called “LEGEND” (Let Evidence Guide Every New Decision), which was developed by Cincinnati Children’s Hospital, and was used to assess the quality of the data. The authors of the forms have made these forms available for use given proper referencing is followed: “refer to data repository documents for permission” (Cincinnati Children’s Organization 2019). The authors have contacted the authors and obtained permission to use them. This tool was selected for its suitability in accommodating varied study designs. It facilitated the evaluation of validity, reliability, and applicability, culminating in the determination of Evidence Levels for each study.

Themes were generated based on the selected studies’ patient criteria, treatment regimens, and clinical outcomes. Table 1 summarizes the included studies and evidence evaluations.

3. Results

4. Discussion

In this systematic review, we provided further evidence about the use of AZ in the treatment of influenza A viral infection in hospitalized or critically ill patients. Based on the limited reports with insufficient evidence, a recommendation for the use of AZ in the treatment regimen was not possible despite the associated reduction in inflammatory cytokines by AZ administration and the absence of significant adverse effects or mortality in hospital-admitted or ICU patients. While these findings support the current guidelines for the treatment of influenza A viral infection, the evidence behind these guidelines and treatment characteristics were provided in this report.

The use of AZ in Influenza A is mechanistically well grounded and indirectly supported by prior experiences with other viral pneumonia, chronic pulmonary diseases (COPD, Asthma, Chronic Cough, and Bronchiectasis), and inflammatory disorders (Bronchiolitis obliterans) (Smith, Du Rand et al. 2020). Yet, the empirical practice of AZ treatment for Influenza A has not been substantiated by good-quality clinical data. In addition, the current guidelines recommend not administering a macrolide antibiotic for the treatment of influenza and justify that by the “insufficient evidence for benefit rather than evidence for harm”, denoting low-quality evidence. The guidelines were based on a few studies (1 RCT and 1 observational study) (World Health Organization 2021). Despite, maybe even because of, these limitations, new articles have been published recently. A critical appraisal of the currently available evidence is valuable especially since the guidelines do not provide evidence about the macrolides’ safety and complications except for the nosocomial infection (World Health Organization 2021). Moreover, the guidelines do not focus on AZ only and include clarithromycin and other macrolides which necessitates contextualizing the ongoing trials and improving the set-up of future trials (Kanoh, Rubin 2010, Zimmermann, Ziesenitz et al. 2018).

Despite the pleiotropic effects of azithromycin, it is certainly not a monotherapy for influenza nor the most potent molecule. Targeted antiviral drugs will likely have a more robust effect on the viral load and they should be started as soon as possible (Muthuri, Venkatesan et al. 2016). However, the anti-inflammatory effects of AZ at targeted anti-IL1, anti-IL6, or steroids are stronger than anti-viral medications (Kanoh, Rubin 2010, Zimmermann, Ziesenitz et al. 2018). AZ is not a standalone therapy as highlighted in all the reviewed studies, but rather an adjunctive therapy for antiviral, anti-inflammatory, and in selected cases, antibiotic drugs. The combination of these medications depends on the patient’s presentation, immune status, and disease stage as it can help the disease prognosis and help prevent secondary bacterial complications (Ishaqui, Khan et al. 2020, Kakeya, Seki et al. 2014, Lee, N., Wong et al. 2017, Martin-Loeches, Bermejo-Martin et al. 2013, Viasus, Paño-Pardo et al. 2011). However, the decision to use drug combinations was not justified or explicitly detailed in these studies, which makes suggesting personalized recommendations for AZ use difficult.

The safety of AZ has always been in limbo as it has shown cardiovascular effects such as acute heart failure, specifically for what was seen recently in a COVID-19 cohort of patients where acute heart failure developed in patients with pre-existing cardiovascular diseases (Bergami, Manfrini et al. 2023, Juurlink 2014). Despite that, as seen in our results, the risk of cardiovascular adverse effects was low in patients who took AZ in combination therapy. In addition, the adverse effects noted were diarrhea (Kakeya, Seki et al. 2014) and dizziness (Lee, N., Wong et al. 2017), while other adverse effects were not significantly different from other treatment options devoid of AZ.

Furthermore, as proven in studies regarding acute respiratory distress syndrome, mechanical respiratory need improves significantly in patients who take AZ and even leads to quicker discontinuation of mechanical ventilation (Kawamura, Ichikado et al. 2018). This confirms our findings as we have concluded that the mechanical respiratory need was improved leading to a lesser hospital stay and faster disease resolution (Lee, N., Wong et al. 2017, Ishaqui, Khan et al. 2020).

Regarding viral load, AZ has direct and indirect antiviral activity in bronchial epithelial cells and other host cells (Gielen, Johnston et al. 2010). In addition to SARS-CoV-2, this has also been shown for influenza, rhinovirus, dengue, ebolavirus, parainfluenza virus, zika virus and enterovirus (Gyselinck, Janssens et al. 2021, Zeng, Meng et al. 2019, Damle, Vourvahis et al. 2020). Data collected here was suggestive of a decrease in the viral load yet inconclusive as only 2 out of the 7 articles mentioned a positive decrease in viral load and 1 of the articles mentioned a neutral relation where there was no change in viral load which highly indicates future focus on this matter. In connection to the anti-inflammatory role of AZ, it has a golden history of having a benefit in reducing cytokines released as evident in multiple COVID researches and animal models of viral infections (Fernandez, Elmore et al. 2004, Stellari, Sala et al. 2014, Zimmermann, Ziesenitz et al. 2018). These findings match our review results (Lee, N., Wong et al. 2017).

Influenza A infection symptom control with AZ-combination therapy has generally shown a positive effect and there are no incidence reports as far as the evidence reviewed that AZ caused a deterioration of an influenza A patient. However, failure to reach a significant relationship between AZ and symptom improvement in some reports prevented us from concluding this positive effect (Ishaqui, Khan et al. 2020, Kakeya, Seki et al. 2014, Lee, N., Wong et al. 2017, Shah, Tar-Ching et al. 2011). In related studies about COVID-19 infection, the time of initiation of the drugs had a significant effect on the resolution of symptoms (Siddiqi, Mehra 2020). This is unlike the studies under review where AZ was initiated for patients requiring hospital admission or critically ill patients requiring ICU. This strongly suggests that further studies are warranted to help guide the appropriate timing of initiation of macrolide therapy in influenza A to better observe its potential benefits.

The findings of this systematic review must be seen in light of some limitations. First, and most importantly, the number and the evidence quality of the published data about AZ in influenza A treatment are restricted to infer an evidence-based recommendation. The available studies, in addition to being few, did not follow the same methodology for sampling, study design, disease severity, treatment regimen, and primary and secondary outcomes. This renders data pooling and comparison for our review question impossible. Finally, it is important to isolate the use of AZ from other macrolides, which is not considered in the current guidelines (World Health Organization 2021). Most other macrolides such as clarithromycin have a variety of findings that either support or go against their use (Snow, Longobardo et al. 2022, Ishii, Komiya et al. 2012, Lee, C., Tai et al. 2021) while AZ generally has stronger molecular and clinical supportive evidence.

In conclusion, AZ shows a potential symptom improvement in influenza A patients, as well as pro-inflammatory laboratory indices such as cytokine release. Another positive aspect of AZ usage is the absence of complications associated with its administration to critically ill patients and acceptable prognosis. However, it is crucial to have more quality research on this matter that focuses on the indication of the use of AZ, and the timing of initiation, dosage, and duration of treatment while monitoring adverse events.