Introduction
Juvenile nasopharyngeal angiofibroma (JNA) is a rare benign tumour that accounts for less than 0.05 per cent of head and neck tumours. Recent epidemiological studies indicate an incidence of approximately 1 in 5,000 to 1 in 50,000 males, with the highest prevalence observed in adolescents from Mediterranean and Middle Eastern regions. It typically affects teenage males between 14 and 25 years of age, with peak incidence occurring between 15 and 17 years. Contemporary research suggests a slight increase in diagnosis rates over the past decade, possibly due to improved imaging techniques and increased awareness among clinicians. Based on their clinical and radiological features, these tumours originate from the sphenopalatine foramen, extending into the sinonasal tract and the nearby foramina and fissures of the skull base. Histologically, JNA is non-encapsulated, highly vascularised and composed of stromal cells and numerous inflammatory cells. The high androgen receptor expression in JNA may explain the prevalence in adolescent males.Reference Snyderman, Pant, Carrau and Gardner1–Reference Bakshi and Bhattacharjee3
Modern treatment approaches require proper pre-operative staging, imaging and angiographic evaluation which are crucial for predicting tumour prognosis and planning the surgical approach. Surgical resection remains the gold standard treatment for JNA. The complexity of surgical management has evolved significantly with advances in endoscopic techniques and imaging technologies.
Different surgical approaches have been described in the literature based on tumour size, location, vascularization and stage. These approaches may create a dilemma for the surgeon regarding choosing an open or endoscopic approach, which can be complicated by massive bleeding. The tumour’s vascular plexus lacks the normal muscular layer in the vessel wall, which explains its tendency towards spontaneous bleeding or bleeding when carefully manipulated during surgery. In cases where surgical risk is deemed excessive, radiotherapy, hormonal therapy and embolisation are suggested as alternative or adjunctive treatments, particularly in cases of intracranial invasion and high operative risk.Reference Makhasana, Kulkarni, Vaze and Shroff4–Reference Siniluoto, Luotonen, Tikkakoski, Leinonen and Jokinen6
To address the challenge of intra-operative bleeding, many techniques have been reported for reducing blood loss during endoscopic resection. Among these interventions, embolisation of the tumour feeding blood vessels has gained significant prominence. Embolisation is a minimally invasive procedure that involves the selective occlusion of tumour blood supply through various approaches. There are different embolisation techniques, including trans-arterial, percutaneous and direct tumoural puncture to inject the embolic agent for the occlusion of the tumour’s feeding blood vessels.Reference Elhammady, Johnson, Peterson and Aziz-Sultan7, Reference Maroda, Beckmann, Sheyn, Elijovich, Michael and DiNitto8
The historical development of embolisation for JNA began when Roberson et al. (1972) were the first to report pre-operative endovascular embolisation in JNA. Since then, the role of embolisation has been supported by some authors to reduce intra-operative blood loss, the risk of tumour recurrence and other complications. Moreover, some studies showed that, in high-stage tumours, there is a significant reduction in the amount of blood loss during surgery among patients who underwent embolisation compared to those who did not.Reference Roberson, Biller, Sessions and Ogura9–Reference Moulin, Chagnaud, Gras, Gueguen, Dessi and Gaubert11
Given the growing body of evidence supporting pre-operative embolisation, the purpose of the current study was to systematically review the available literature and carry out a meta-analysis of the existing studies that compared the surgical outcomes of patients undergoing JNA surgical resection with and without pre-operative endovascular embolisation.
Currently, pre-operative endovascular embolisation is considered the standard of care for JNA resection.Reference Giorgianni, Molinaro, Agosti, Terrana, Vizzari and Arosio12 In this meta-analysis, non-embolised patients were used as the control group.
Materials and methods
Population, intervention, comparison (PICO) outcome framework
Population: The study population included patients diagnosed with JNA.
Intervention: The intervention investigated was pre-operative endovascular embolisation.
Comparison: The control group included patients diagnosed with JNA who did not undergo pre-operative endovascular embolisation.
Outcomes: Outcome measure was intra-operative blood loss.
Study design: Most study designs consisted of case series, while the remaining studies were randomised controlled trials.
Literature search strategy and Inclusion criteria
We conducted a comprehensive search for articles published from 1972 through December 2024 in the following electronic databases: PubMed, Scopus and Google Scholar. The following Medical Subject Heading (MeSH) terms were used “Juvenile nasopharyngeal angiofibroma” or “Juvenile angiofibroma” or “angiofibroma” and “skull base” or “tumor” and “preoperative” or “endovascular” or “embolization”. Additionally, we checked the reference list of each obtained article for additional potentially relevant studies.
For our meta-analysis, we included studies published since 1972 examining patients with JNA undergoing surgical resection that compared pre-operative embolisation with no pre-operative embolisation. We excluded studies that lacked a control group or an embolised group, reviews, meta-analyses, letters, comments and studies that were not written in English (Figure 1).
Flowchart showing the process of study selection for the meta-analysis.
Figure 1 Long description
The flowchart outlines the study selection process for a meta-analysis. It begins with 'Records identified through electronic database searching (n = 489).' An arrow leads to 'Records after duplicates removed (n = 209).' Next, 'Records excluded based on title or abstract review (n = 144)' is shown, leading to 'Records screened (n = 65).' From here, 'Records excluded (n = 45)' is detailed with subpoints: 'No control group (n = 36)' and 'No embolization group (n = 9).' This leads to 'Full text articles assessed for eligibility (n = 20).' An arrow then points to 'Full text articles excluded (n = 3).' Finally, the process concludes with 'Studies included in the meta analysis (n = 17).' The flowchart is organized into sections labeled 'Identification,' 'Screening,' 'Eligibility,' and 'Included,' indicating the sequential steps in the study selection process.
Data extraction
We extracted descriptive information from the studies included in this meta-analysis (Table 1), as well as estimated mean intra-operative blood loss, operative time, hospital stay and recurrence data (eTable 1).Reference Siniluoto, Luotonen, Tikkakoski, Leinonen and Jokinen6, Reference Li, Qian, Shan and Wang10, Reference Moulin, Chagnaud, Gras, Gueguen, Dessi and Gaubert11, Reference Pletcher, Newton, Dedo and Norman13–Reference Ardesan, Ardehali and Doustmohammadian26
Descriptive information of the studies included in the meta-analysis
Table 1 Long description
The table summarizes 17 studies by author and year, listing study design, sample size, participant age, sex distribution, and counts of embolised versus non-embolised patients. Study designs vary widely, including case series, case–control studies, retrospective studies, cross-sectional study, and two randomised controlled trials. Sample sizes range from 8 to 74 participants, with most studies reporting all-male samples; two studies do not provide sex information. Mean ages, when reported, span from about 12.6 to 23.8 years, and several studies include a standard deviation while others do not report age. Embolised patient counts range from 4 to 32, while non-embolised counts range from 2 to 42; several studies have equal group sizes, but others are notably unbalanced, such as Ardehali 2010 with far more non-embolised than embolised patients. The largest study is Tan 2017 with 74 participants, reporting 32 embolised and 42 non-embolised. Interpretation is limited by missing demographic details in some rows and by heterogeneity in study design and group balance across studies.
Abbreviations: F = female; M = male; NP = not provided; SD = standard deviation.
Statistical analysis
The statistical analysis was performed using the IBM SPSS Statistics 28 software. We evaluated the differences in intra-operative blood loss, operative time, hospital stay and recurrence between embolised and non-embolised patients using meta-regression analysis. Cochran’s Q statistic was used to assess the level of statistical heterogeneity among the included studies. The Q statistic follows a chi-squared distribution with K−1 degrees of freedom.Reference Kulinskaya and Dollinger27
We conducted an additional analysis investigating the impact of tumour stage on intra-operative blood loss using meta-regression analysis.
Risk of bias assessment
To determine the risk of bias in each study, we used the Risk Of Bias In Non-randomised Studies - of Interventions, (ROBIN1) tool to create a visual representation of bias across these four domains, which is shown in eFigure 1.
Results
We found 489 citations through our initial search, from which we removed 280 duplicates. This left us with 209 relevant studies, of which 144 were excluded as irrelevant based on their title or abstract, leaving 65 studies to be screened for essential data. We assessed 20 full-text articles for eligibility and eventually included 17 studies in our final analysis, which were reported in accordance with Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (Figure 1). Our meta-analysis included 17 studies with a total of 524 patients.Reference Siniluoto, Luotonen, Tikkakoski, Leinonen and Jokinen6, Reference Li, Qian, Shan and Wang10, Reference Moulin, Chagnaud, Gras, Gueguen, Dessi and Gaubert11, Reference Pletcher, Newton, Dedo and Norman13–Reference Ardesan, Ardehali and Doustmohammadian26 Most study designs consisted of case series, while the remaining studies were randomised controlled trials.
Study descriptive information, design, sample size and embolisation prevalence are reported in Table 1.
The analysis resulted in a high level of heterogeneity among the studies (Q = 55.74). The heterogeneity can be explained by the small sample sizes of the studies and insufficient information regarding the methodologies for collecting blood loss data during surgery (Figure 2, eFigure 2).
Forest plot of estimated intra-operative blood loss. overall effect size -685 ml; 95 per cent confidence interval -1013 ml to -357 ml; p < 0.001. The p-value indicates the significance between the differences in the embolised and non-embolised results.
Figure 2 Long description
The forest plot illustrates estimated intra-operative blood loss effect sizes across multiple studies. The x-axis is labeled 'Effect Size (ml)' ranging from -1000 to 1000. Each study is represented by a square marker, with horizontal lines indicating the confidence intervals. The pooled random-effects estimate is shown as a diamond, with a value of -685 ml and a 95 percent confidence interval from -1013 ml to -357 ml. Most study estimates are negative, indicating reduced blood loss, with some confidence intervals crossing zero. The plot includes a note on heterogeneity, with a Q value of 55.74, suggesting variability among study results. The significance level is less than 0.001, highlighting differences between embolised and non-embolised groups.
We identified a pattern indicating a more substantial effect of embolisation in patients with higher tumour stages. The stage-specific analysis revealed its own heterogeneity patterns. The heterogeneity (Q = 20.21) observed in the results can be attributed to insufficient data on intra-operative blood loss for each stage, leading to small sample sizes. Moreover, the different JNA tumour staging systems used across studies, including Radkowski, Chandler, Andrew and Fisch, may also contribute to the observed heterogeneity.
Primary outcomes
A comparison of the results for embolised and non-embolised patients revealed a significant difference in estimated intra-operative blood loss (overall effect size -685 ml; 95 per cent confidence interval -1013 ml to -357 ml; p < 0.001) (Figure 2). Other outcomes, such as operative time, hospital stay and recurrence, did not reach statistical significance (Figure 3). We could not analyse other factors reported in the studies due to small sample sizes.
Forest plot of operative time (A: overall effect size -8.51 minutes; 95 per cent confidence interval -66.45 to 49.43 minutes; p = 0.77), hospital stay (B: overall effect size -1.15 days; 95% confidence interval -4.84 to 2.55 days; p = 0.54) and recurrence (C: overall effect size -0.89 recurrence; 95% confidence interval 0.32 to 2.45; p = 0.82).
Figure 3 Long description
A composite figure with three labeled parts: A, B and C. Each part is titled Forest Plot and includes a study list and a forest plot with blue squares and horizontal confidence interval lines, plus an overall diamond. The y-axis in each part is a list of study names under the header Study. A. The x-axis is an effect size scale for Mean difference, with tick labels minus 200, minus 100, 0, 100 and 200. A vertical dashed reference line is at 0. The table columns are Mean difference, Standard error, Weight and p value. The Overall row reports mean difference minus 8.51 minutes, with 95 per cent confidence interval minus 66.45 to 49.43 minutes and p value equals 0.77. B. The x-axis is an effect size scale for Mean difference, with tick labels minus 10, minus 5, 0, 5 and 10. A vertical dashed reference line is at 0. The table columns are Mean difference, Standard error, Weight and p value. The Overall row reports mean difference minus 1.15 days, with 95 per cent confidence interval minus 4.84 to 2.55 days and p value equals 0.54. C. The x-axis is an effect size scale for Odds ratio, with tick labels 0.01, 0.1, 1, 10, 100 and 1000. A vertical dashed reference line is at 1. The table columns are Odds ratio, Lower, Upper, p value and Weight. The Overall row reports odds ratio 0.89 recurrence, with 95 per cent confidence interval 0.32 to 2.45 and p value equals 0.82. Across A and B, the overall confidence intervals span the 0 reference line. In C, the overall confidence interval spans the 1 reference line.
Stage-specific analysis
Analysis of intra-operative blood loss by tumour staging revealed that embolisation significantly reduced intra-operative blood loss in patients with stage III more than in patients with stage I and II tumours (eTable 2). We noticed a trend among embolised patients in which higher stages were associated with decreased effect size, indicating lower intra-operative blood loss compared to non-embolised patients (Figure 4). This finding did not reach statistical significance, with a p-value of 0.065 (95 per cent confidence interval -1.221 to 0.044). Additionally, among patients with higher-stage tumours, intra-operative blood loss in patients who underwent embolisation was lower than in those who did not.
Bubble plot of the reduction in intra-operative blood loss regarding tumour staging.
Figure 4 Long description
Bubble Plot Moderator: stage Legend: Primary studies Meta-regression prediction line 95 percent confidence intervals The horizontal axis shows stage from 1 to 3. The vertical axis shows Predicted Value of Effect Size from negative 3 to 1. A meta-regression prediction line slopes downward from stage 1 to stage 3. A shaded band around the line shows 95 percent confidence intervals. Bubble size meaning is not labeled in the chart. Primary studies data points: Stage 1: four points at approximately 0.2, 0.0, negative 0.3 and negative 0.6. Stage 2: six points at approximately 0.6, 0.0, negative 0.4, negative 0.7, negative 1.8 and negative 0.2. Stage 3: four points at approximately negative 0.2, negative 0.7, negative 1.4 and negative 2.2.
Discussion
Primary findings and context
Since JNA is a rare benign tumour that accounts for less than 0.05 per cent of head and neck tumours, we conducted a systematic review and meta-analysis to reach evidence-based conclusions concerning pre-operative embolisation. The rarity of this condition necessitates systematic synthesis of available evidence to guide clinical practice. Due to its high vascularity and complex anatomical location, removing this tumour could result in significant bleeding. Therefore, surgeons seek to reduce intra-operative blood loss to prevent complications and recurrence.
Mechanisms and indications of embolisation
Embolisation serves multiple therapeutic purposes beyond simple blood loss reduction. It can serve as a method to block of the vascular supply to JNA tumours and is also indicated for (I) managing arterial feeders that cannot be reached during surgery; (II) decreasing surgical complications by reducing intra-operative blood loss; (III) reducing operative time; (IV) increasing the chances of complete surgical removal; (V) decreasing the risk of damage to surrounding healthy tissue; (VI) relieving persistent pain; (VII) reducing recurrence of the tumour; and (VIII) providing better visibility of the surgical site while decreasing overall surgical complications.Reference Gupta, Purkayastha, Bodhey, Kapilamoorthy and Kesavadas28 However, in this systematic review, we found that some of the parameters outlined above were not mentioned in the included studies and could not be measured due to small sample sizes.
Risk–benefit profile
Pre-operative embolisation of JNA remains controversial regarding its beneficial outcomes. The intervention presents both significant advantages and potential complications that must be carefully weighed. Advantages relate to embolisation as a useful intervention that reduces intra-operative blood loss, may facilitate complete surgical resection and improves overall treatment outcomes. Conversely, pre-operative embolisation may result in rare complications such as recurrence of bleeding, cranial nerve injury, nasal synechia, cavernous sinus damage, vision disturbance, low-grade fever, nausea, local oedema, trismus, temporary reduction in facial sensation, reduced lacrimation and thrombosis.
Evidence synthesis and comparison with previous studies
This meta-analysis demonstrated a significant difference in intra-operative blood loss between embolised and non-embolised patients (overall effect size -685 ml; p < 0.001; Figure 2). This finding is supported by the recent meta-analysis conducted by Diaz et al. in 2023,Reference Diaz, Wang, Bujnowski, Arimoto, Armstrong and Cyberski29 which confirmed that our results from 17 studies provide substantial statistical and clinical evidence of a notable decrease in blood loss through pre-operative embolisation, regardless of the technique employed. This consistency across independent meta-analyses strengthens the evidence base for embolisation’s effectiveness. Thus, our meta-analysis provides statistically significant evidence from 17 studies demonstrating a reduction in intra-operative blood loss.
In contrast to blood loss outcomes, for other evaluated factors, including operative time, hospital stay and recurrence, the results yielded no significant difference between embolised and non-embolised patients (Figure 3). The lack of significant differences in these secondary outcomes may reflect the multifactorial nature of surgical recovery and the complex interplay of variables affecting these endpoints.
Recurrence patterns and long-term outcomes
Some studies, including that of Roberson et al. (1979), have raised concerns that pre-operative embolisation could obscure tumour margins and increase recurrence risk.Reference Roberson, Price, Davis and Gulati14 However, our meta-analysis found no significant difference in recurrence rates between embolised and non-embolised patients (p = 0.82). We initially expected that recurrence would be significantly lower in embolised patients, which we attributed to improved bleeding control and reduced intra-operative blood loss during surgery. The absence of increased recurrence risk is reassuring and contradicts early concerns about embolisation potentially compromising surgical outcomes. These factors can lead to a clearer surgical field and more complete tumour resection, potentially reducing the risk of future tumour recurrence. Modern intra-operative imaging techniques may help mitigate this issue. Future prospective studies focusing on recurrence patterns post-embolisation are warranted.
Stage-specific considerations
When embolised and non-embolised patients were compared by tumour stage, we noticed a trend indicating that embolised tumours at higher stages corresponded to lower intra-operative blood loss, although this did not reach statistical significance. This stage-specific pattern has important clinical implications for treatment planning. Although larger tumours often develop collateral circulation from the internal carotid artery or contralateral vessels, the observed reduction in blood loss may stem from selective embolisation of primary feeders (e.g., the internal maxillary artery) combined with surgical ligation of secondary feeders intra-operatively. Additionally, high-stage tumours may develop more chaotic, less efficient neovascularisation patterns, which respond well to embolisation. This pathophysiological understanding suggests that embolisation may be particularly beneficial in advanced cases where traditional surgical control of bleeding is most challenging.
Study limitations
The current meta-analysis has several limitations that must be considered when interpreting these findings. These include: a limited number of studies available for inclusion, the absence of standard deviation data in some studies that precluded their inclusion and a lack of detailed information on evaluated factors, such as the percentage and types of complications, in all included studies. Furthermore, studies used different staging systems, and in some studies, the staging system was not reported. The heterogeneity in staging systems represents a fundamental challenge in JNA research and limits our ability to provide definitive stage-specific recommendations.
Another limitation of this meta-analysis is the variability and insufficient information regarding surgical techniques reported across the included studies. While some studies employed an open surgical approach, others utilised an endoscopic technique and a few described a combined approach. Additionally, the lack of detailed reporting on specific methods, such as coblation, bipolar electrocautery or microdebrider use, limits our ability to evaluate their impact on intra-operative outcomes. This technical heterogeneity may explain some of the observed variation in outcomes and represents an important area for future standardisation efforts. These inconsistencies in surgical techniques may contribute to variations in measured intra-operative blood loss and hinder comprehensive comparison of clinical outcomes.
Novel contributions and clinical significance
Regarding tumour stage, our meta-analysis is, to our knowledge, the first to compare embolised and non-embolised patients while analysing outcomes by tumour stage. Unlike the published systematic review by Diaz et al. in 2023, the novelty of our study lies in providing stage-specific outcomes, offering more clinically actionable insights. A recent systematic review by Kothari et al. (2023),Reference Kothari, Linker, Tham, Maroda, McElfresh and Fastenberg30 concluded that the evidence base is insufficient to support definitive expert recommendations regarding pre-operative embolisation in JNA. Our findings and the conclusions of Kothari et al. are concordant in underscoring the need for further high-quality prospective research, particularly with regard to stage-specific outcomes.Reference Diaz, Wang, Bujnowski, Arimoto, Armstrong and Cyberski29 Our findings suggest that embolised patients with higher-stage tumours may experience lower intra-operative blood loss compared to non-embolised patients with tumours of the same stage, presenting near-significant evidence of this effect. Additionally, our discussion introduces a novel pathophysiological explanation for the observed reduction in blood loss, even in advanced tumours following embolisation.
Clinical implications
The findings of this meta-analysis have several important clinical implications for the management of JNA patients. The significant reduction in intra-operative blood loss (685 ml) supports the routine use of pre-operative embolisation, particularly in advanced-stage tumours where bleeding risk is highest. Patients and families can be informed that embolisation significantly reduces surgical bleeding without increasing recurrence risk or major complications. The improved bleeding control may allow surgeons to consider more complex endoscopic approaches that might otherwise be contraindicated due to bleeding concerns. Reduced blood loss may decrease the need for blood products and intensive monitoring, potentially reducing healthcare costs and resource utilisation. Although the trend towards greater benefit in higher-stage tumours did not reach statistical significance (p = 0.065), this near-significant finding—supported by biological rationale—suggests that embolisation warrants serious consideration in advanced cases, pending confirmation in a larger prospective study. Lastly, the heterogeneity observed highlights the need for standardised staging systems, surgical techniques and outcome reporting to improve future evidence synthesis.
Conclusion
Our findings provide consistent and statistically significant evidence that pre-operative embolisation significantly reduces intra-operative blood loss during JNA resection overall. While a near-significant trend (p = 0.065) suggests a potentially greater benefit in higher-stage tumours, it did not reach statistical significance and should be interpreted with caution given the limited sample sizes and the paucity of high-quality comparative data in this population. These results support the continued use of embolisation as a standard adjunct for JNA surgery, especially in advanced cases. The clinical benefit is substantial and clinically meaningful, with an average reduction of 685 ml of intra-operative blood loss. Additionally, the shift towards endoscopic techniques highlights the importance of future research comparing outcomes between surgical approaches. Our study emphasises the need for ongoing research to validate these findings and further investigate the impact of embolisation on surgical outcomes, particularly in high-stage tumours. Future studies should focus on standardising staging systems, developing consensus guidelines for embolisation techniques and conducting larger prospective trials to further refine patient selection criteria.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0022215126104885.
Data availability statement
All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author.
Funding sources
None.
Competing interests
All authors have no conflict of interest.
Declaration of generative artificial intelligence (AI) and AI-assisted technologies in the writing process
During the preparation of this work, the authors used ChatGPT in order to correct grammar and readability. After using this tool, the authors reviewed and edited the content as needed and takes full responsibility for the content of the publication.
Open access
