Pulmonary artery floating sign: a novel predictor of fatal haemoptysis in pulmonary mucormycosis and its early intervention strategy Authors: Mingjing Yu, Faming Jiang, Li Guo, Hongxia Wu, Xiang Tong, Hao Wang, Lijuan Gao, Yina Li, Jiangang Wang, Hong Fan, Ye Wang Published: 2026 ABSTRACT AbstractBackgroundMassive haemoptysis is a life-threatening complication of pulmonary mucormycosis, often resulting from invasive destruction of pulmonary arteries. Early predictors and preventive strategies remain poorly defined.MethodsWe retrospectively enrolled all hospitalised patients with pulmonary mucormycosis regardless of haemoptysis between 2019 and 2024. Patients were stratified into non-massive and massive haemoptysis groups. Clinical and radiological variables were compared between groups, and univariable and multivariable logistic regression analyses were performed. Given the limited number of events and separation, Firth’s penalised logistic regression and reduced multivariable models were used to identify robust predictors of massive haemoptysis. Additionally, we explored the feasibility of prophylactic pulmonary artery embolisation (PPAE) in a subset of high-risk patients to evaluate its potential role in preventing fatal haemorrhage.ResultsFifty-seven patients with pulmonary mucormycosis were enrolled, and 22 (38.6%) patients had massive haemoptysis, with a mortality rate of 72.7% (16/22). Cavity (68.2% vs 34.3%, p=0.026) and central necrosis (100% vs 74.3%, p=0.009) were significantly more common in the massive haemoptysis group. The pulmonary artery floating sign (PAFS) was observed in 81.8% (18/22) of the massive haemoptysis group versus 31.4% (11/35) of the non-massive haemoptysis group (p<0.001), and was the only independent predictor (adjusted OR 6.20, 95% CI 1.68 to 28.49). Five high-risk patients, including three with floating sign, underwent PPAE; only one subsequently died of bleeding.ConclusionThe PAFS is a novel and independent predictor of massive haemoptysis in pulmonary mucormycosis. Its early identification may support early risk stratification and inform consideration of PPAE in selected high-risk patients. WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS Pulmonary artery floating sign predicts massive haemoptysis in mucormycosis, supporting prophylactic pulmonary artery embolisation (PPAE) feasibility in high-risk patients. Pulmonary artery floating sign predicts massive haemoptysis in mucormycosis, supporting prophylactic pulmonary artery embolisation (PPAE) feasibility in high-risk patients. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY PPAE enabled by early floating sign detection may prevent fatal haemorrhage and improve outcomes. PPAE enabled by early floating sign detection may prevent fatal haemorrhage and improve outcomes. Introduction Pulmonary mucormycosis is a life-threatening opportunistic infection that primarily affects immunocompromised individuals, such as those with haematological malignancies, uncontrolled diabetes or prolonged immunosuppressive therapy.13 Characterised by its aggressive angioinvasive behaviour, mucormycosis frequently invades vascular structures, leading to thrombosis, tissue necrosis and rapid clinical deterioration, with a high mortality rate despite antifungal treatment and surgical intervention.4 5 Massive haemoptysis is among the most devastating complications of pulmonary mucormycosis, often occurring abruptly and progressing rapidly to fatal outcomes.6 Invasive angioinvasion of the pulmonary artery by fungal elements is a key pathological feature of mucormycosis.7 Pulmonary artery aneurysms and pseudoaneurysms have been repeatedly reported in case series and imaging studies as hallmark radiological features associated with life-threatening haemoptysis in mucormycosis.810 These lesions reflect advanced arterial wall disruption and serve as visible indicators of an imminent bleeding risk. However, aneurysm/pseudoaneurysm formation typically occurs at a late stage of disease progression, after substantial vascular injury has already taken place. And the detection of aneurysm/pseudoaneurysm requires enhanced chest CT. Many clinical cases experience catastrophic bleeding without any detectable aneurysm or pseudoaneurysm on imaging—a finding that has been reported both in retrospective analyses and autopsy series.1113 The presentation of haemoptysis is often abrupt, profuse and rapidly fatal, with insufficient time for effective rescue,2 which is distinct from bronchial artery bleeding (usually gradual and responsive to intervention) seen in chronic structural lung diseases like bronchiectasis. Therefore, we believe the bleeding most likely originates from pulmonary artery rupture, even in the absence of visible arterial disruption on CT angiography (CTA). This disconnect between radiological findings and clinical events highlights a critical unmet need: the identification of earlier and more sensitive imaging markers capable of predicting massive haemoptysis before irreversible vascular rupture occurs. In this study, we retrospectively evaluated the imaging characteristics of patients with pulmonary mucormycosis to identify risk factors for massive haemoptysis. We particularly focused on a novel imaging feature—the pulmonary artery floating sign (PAFS)—which may represent a pre-rupture state of vascular compromise. Furthermore, we assessed the feasibility and short-term outcomes of prophylactic pulmonary artery embolisation (PPAE) in patients deemed to be at high risk, aiming to inform early interventional strategies. Methods Study design and setting This was a retrospective observational study conducted at West China Hospital, Sichuan University—a tertiary academic medical centre in Southwest China—between January 2019 and December 2024. The study protocol was approved by the Institutional Review Board (No. 2025-265), and the requirement for informed consent was waived due to the retrospective nature of the study. Patient selection and diagnostic criteria Patients were eligible for inclusion if they met the following criteria: (1) Age ≥18 years; (2) Clinical, microbiological or pathological diagnosis of pulmonary mucormycosis regardless of haemoptysis; (3) Underwent chest CT; (4) Complete clinical and imaging data available for review. The diagnosis of pulmonary mucormycosis was established according to the 2019 European Organisation for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria, integrating host factors, typical radiological features (reversed halo sign) and microbiological or histopathological confirmation, includes: broad, non-septate hyphae with right-angle branching in lung or bronchial tissue; positive fungal culture for Mucorales from sputum, bronchoalveolar lavage fluid (BALF) or biopsy specimens or detection of Mucorales DNA by next-generation sequencing in BALF or respiratory samples. Grouping and clinical definitions Patients were stratified into two groups based on haemoptysis severity during hospitalisation: Massive haemoptysis group: bleeding ≥300 mL in 24 hours, or accompanied by respiratory failure, haemodynamic instability or death.14 Non-massive haemoptysis group: no haemoptysis or mild/moderate bleeding <300 mL/day, without clinical deterioration or need for emergency intervention. Radiological assessment All included patients underwent chest CT scanning using institutional protocols. Two board-certified thoracic interventional radiologists (>5 years’ experience), blinded to clinical outcomes, independently reviewed the CT images. In the event of a disagreement between the two evaluating physicians (particularly PAFS), a third radiologist was consulted to make the final determination. Radiologic features assessed: (1) Lesion morphology: cavity, necrosis, reversed halo sign; (2) Distribution: single versus multiple lesions, lobar involvement; (3) Vascular involvement, including: pulmonary artery aneurysm/pseudoaneurysm: focal dilation ≥1.5× adjacent segment, intact wall; (4) Cavity: defined as a gas-filled space within a pulmonary lesion or consolidation, with a well-demarcated wall, visible on CT as a low-attenuation area indicating tissue breakdown and airway communication. Cavities were considered significant when the internal airspace exceeded 5 mm in diameter and was not attributable to pre-existing structural lung disease (eg, emphysema). Mass with central necrosis: defined as central non-enhancing low-attenuation areas within a mass or consolidation on contrast-enhanced CT, representing devitalised tissue without contrast uptake. Necrosis was considered present when such areas involved more than 30% of the lesion volume. PAFS: refers to one or more pulmonary artery branches that traverse, either centrally or eccentrically, through a necrotic or cavitary region of lung parenchyma, without visible contact with the surrounding cavity wall on contrast-enhanced CT or non-contrast CT. These arteries appear to be suspended within the air- or fluid-filled space, giving the visual impression of ‘dead branches floating on water’.15 For patients who did not undergo contrast-enhanced chest CT, pulmonary artery aneurysms or pseudoaneurysms could not be reliably assessed and were therefore excluded from the multivariable analysis. Prophylactic pulmonary artery embolisation (PPAE) Patients in the massive haemoptysis group were evaluated for interventional management based on radiologic and clinical criteria. We selected patients who had not yet developed massive haemoptysis but exhibited high-risk imaging features, including pulmonary artery aneurysms or the proposed floating sign. In select cases, PPAE was performed by experienced vascular interventionalists. Procedure details include: (1) Access: right femoral vein puncture, 6-Fr sheath placement; (2) Catheterisation: Swan-Ganz or pigtail catheter for pulmonary angiography; (3) Target selection: based on CT/angiographic findings (floating sign, irregular vessel contour, contrast leakage); (4) Embolic materials: Nester coils (Cook Medical, Bloomington, IN, USA) with diameters ranging from 4 to 10 mm and lengths from 2 to 14 cm according to vessel size; (5) Monitoring: all patients observed for 48–72 hours postprocedure for signs of pulmonary infarction, rebleeding or right ventricular decompensation. Data collection and variables The following data were collected, includes: (1) Demographics: age, sex, comorbidities (eg, diabetes, malignancy, immunosuppression); (2) Clinical outcomes: development of massive haemoptysis, in-hospital mortality, need for surgical or bronchoscopic intervention; (3) Radiological data: lesion characteristics, vascular findings, presence of floating sign or aneurysm; (4) Treatment modalities and clinical outcomes: the use of antifungal therapy, classified as monotherapy or combination therapy (eg, amphotericin B with Isavuconazole); (5) PPAE details: indication, timing, technical success, complications and 30-day outcome. Statistical analysis Continuous variables are presented as mean±SD or median with IQR, as appropriate, and categorical variables as counts and percentages. Between-group comparisons were performed using Student’s t-test or Mann–Whitney U test for continuous variables and the χ2 test or Fisher’s exact test for categorical variables, as appropriate. Univariable logistic regression analyses were first performed to screen potential predictors of massive haemoptysis. Variables showing statistical significance in univariable analyses and those considered clinically relevant were considered for multivariable modelling. The results of the univariable analyses are presented in online supplemental table S1. Given the limited number of events and the presence of separation, Firth’s penalised likelihood logistic regression was applied to reduce small-sample bias and obtain reliable estimates. Because central necrosis is a prerequisite and structural component for identifying PAFS, necrosis was not included simultaneously with PAFS in the main multivariable models to avoid structural dependence. Reduced multivariable models were constructed as sensitivity analyses to minimise overfitting. Mortality was initially analysed as a binary outcome using Fisher’s exact test and reduced Firth penalised logistic regression models. Given that not all patients underwent contrast-enhanced CT, baseline characteristics and haemoptysis outcomes were compared between patients with and without contrast-enhanced CT to assess potential selection bias (online supplemental table S3). Time-to-event analyses were performed to evaluate the association between the PAFS and mortality. The index time (time zero) was defined as the date of the CT examination on which PAFS was assessed. Overall survival was calculated from the index CT date to death or the end of follow-up. Patients who were alive at the end of follow-up were censored at their last follow-up date. Kaplan–Meier survival curves were generated stratified by PAFS status and compared using the log-rank test. Patients with missing information on the date of death or last follow-up were excluded from the survival analysis. To further explore the relationship between PAFS and mortality while accounting for major clinical confounders, Cox proportional hazards regression models were constructed. Given the limited sample size and number of events, a parsimonious adjustment strategy was adopted. Massive haemoptysis, a clinically dominant determinant of short-term mortality, was included as an adjustment variable. HRs with 95% CIs were reported. All statistical analyses were conducted using R (V.4.4.3) and SPSS (V.27.0). A two-sided p value<0.05 was considered statistically significant. Patient and public involvement Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this research. Results Patient characteristics Online supplemental figure S1 illustrates the flow of patients through the study. A total of 57 hospitalised patients with pulmonary mucormycosis were included in the final analysis and baseline characteristics of the study participants are summarised in table 1. The median age was 52.3 years (IQR): 44.0–62.3), and the majority were male (n=45, 78.9%). Table 1Baseline clinical and radiological characteristics of the study populationVariablesMassive haemoptysis(n=22)Non-massive haemoptysis(n=35)P valueGender (male)21 (95.5%)24 (66.7%) 0.019 Age (years)55.86±14.6950.23±15.060.169Comorbidities Diabetes14 (63.6%)24 (68.6%)0.923 COVID-19/influenza1 (5%)4 (11.1%)0.639 Malignant disease3 (15%)6 (17.2%)1 Allogeneic HSCT2 (9%)3 (8.6%)1 Renal dysfunction3 (15%)12 (34.3%)0.124Radiological information Aneurysms2 (9.1%)1 (2.9%)0.553 Lesion (single)16 (72.7%)20 (57.1%)0.365 Lesion size (cm)8.3±3.47.5±4.00.430 Involving the entire lobe11 (50%)11 (31.4%)0.262 Cavity15 (68.2%)12 (34.3%) 0.026 Mass with central necrosis22 (100%)26 (74.3%) 0.009 Reversed halo sign10 (45.5%)12 (34.3%)0.573 Pulmonary artery floating sign18 (81.8%)11 (31.4%) <0.001 Antifungal therapy17 (77.3%)34 (97.1%) Monotherapy-amphotericin B4 (18.2%)7 (20%)1 Monotherapy-triazole10 (45.5%)22 (62.9%)0.763 Amphotericin B+triazole3 (13.6%)5 (14.3%)1Survival status (survival)6 (27.3%)17 (48.6%)0.348Fisher’s exact test was used for variables with expected counts<5; χ2 test was used otherwise; p value not reported due to early death before treatment initiation. Bold indicates statistical significance (p<0.05).HSCT, haematopoietic stem cell transplantation. Baseline clinical and radiological characteristics of the study population Fisher’s exact test was used for variables with expected counts<5; χ2 test was used otherwise; p value not reported due to early death before treatment initiation. Bold indicates statistical significance (p<0.05). HSCT, haematopoietic stem cell transplantation. The most common comorbidity was uncontrolled diabetes mellitus, affecting 66.7% of patients (n=38), followed by renal dysfunction (26.3%, n=15) and malignant disease (15.8%, n=9). A minority had a history of COVID-19 or influenza (8.8%, n=5), or had undergone allogeneic haematopoietic stem cell transplantation (8.8%, n=5). 51 patients (89.5%) received systemic antifungal therapy, and 37 (64.9%) underwent contrast-enhanced CT imaging during their hospital stay. Based on the presence and severity of haemoptysis, patients were categorised into two groups: the massive haemoptysis group (n=22, 38.6%) and the non-massive haemoptysis group (n=35, 61.4%), which included patients with either mild haemoptysis or no haemoptysis. Antifungal therapy was administered to 17 of 22 patients (77.3%) in the massive haemoptysis group and 34 patients (97.1%) in the non-massive group. Several patients in the massive haemoptysis group died before antifungal therapy could be initiated. In the massive haemoptysis group, patients who did not receive antifungal therapy died before treatment initiation. In the non-massive haemoptysis group, one patient who did not receive antifungal treatment died of disease progression prior to diagnosis. Regarding the treatment regimens, the proportions of patients receiving monotherapy with amphotericin B were similar between the groups (18.2% vs 20.0%, p=1.000). Likewise, no significant differences were observed in the use of triazole monotherapy (45.5% vs 62.9%, p=0.763) or combination therapy with amphotericin B and triazoles (13.6% vs 14.3%, p=1.000). Survival was observed in 6 of 22 patients (27.3%) in the massive haemoptysis group and in 17 of 35 patients (48.6%) in the non-massive haemoptysis group, with no statistically significant difference between groups (p=0.348). Radiological features and group comparisons Contrast-enhanced CT scans were reviewed in 37 patients. Aneurysms were identified in only 3 patients (5.3%): two in the massive haemoptysis group (9.1%) and one in the non-massive haemoptysis group (2.9%). In contrast, the PAFS was a more prevalent and discriminative feature. It was observed in 81.8% of patients in the massive haemoptysis group (18/22) and 31.4% of patients in the non-massive haemoptysis group (11/35, p<0.001). Inter-rater agreement for the floating sign between two interventional radiologists was excellent, with Cohen’s κ=0.836 (95% CI 0.693 to 0.979, p<0.001). Other notable radiologic findings included: Cavity: 68.2% versus 34.3% (in the massive haemoptysis group vs that in the non-massive haemoptysis group) (p=0.026); Necrosis: 100% versus 74.3% (p=0.009); Reversed halo sign: 45.5% versus 34.3% (p=0.573). Typical radiological features of pulmonary mucormycosis are provided in figure 1. Typical radiological features of pulmonary mucormycosis. (a) Pulmonary cavity, (b) reversed halo sign, (c) pulmonary artery aneurysms (white arrow), (d) mass with central necrosis, (e,f) pulmonary artery floating sign within necrotic and cavitary lesions (white arrows). Lesions involving an entire pulmonary lobe were more common in the massive haemoptysis group (50.0% vs 31.4%), but this difference did not reach statistical significance. The mean lesion size was larger in the massive haemoptysis group (8.3±3.4 cm vs 7.5±4.0 cm), though also not statistically significant (p=0.430). In multivariable analyses using Firth penalised logistic regression, the PAFS was the only radiological feature independently associated with massive haemoptysis (table 2). Other imaging findings, including cavitation and necrosis, did not retain statistical significance in the multivariable model. Importantly, the association between PAFS and massive haemoptysis remained robust across reduced multivariable models constructed to minimise overfitting (table 2; online supplemental tables S2 and S4). Table 2Multivariate Firth logistic regression analysis of risk factors for massive haemoptysisVariablesFirth logistic regressionOR (95% CI)P valueGender (male)0.47 (0.04 to 6.36)0.535Cavity0.18 (0.001 to 1.95)0.181Mass with central necrosis11.40 (0.46 to 1897.66)0.136Pulmonary artery floating sign26.66 (2.22 to 3849.83)0.006Due to the presence of complete separation and small sample size, we used Firth’s penalised likelihood logistic regression to reduce bias and obtain stable estimates. Multivariate Firth logistic regression analysis of risk factors for massive haemoptysis Due to the presence of complete separation and small sample size, we used Firth’s penalised likelihood logistic regression to reduce bias and obtain stable estimates. Association between pulmonary artery floating sign and mortality To further evaluate the prognostic relevance of the PAFS, a time-to-event analysis was performed. Using the date of the index CT examination as time zero, Kaplan–Meier survival curves were generated stratified by PAFS status. Patients alive at the end of follow-up were censored at their last follow-up date. Among the 57 patients included in the cohort, 53 had complete time-to-event data and were eligible for survival analysis, while 4 patients were excluded due to missing information on the date of death or last follow-up. Kaplan–Meier analysis demonstrated significantly worse overall survival in patients with PAFS compared with those without PAFS (log-rank p=0.0368; online supplemental figure S2). Patients with PAFS showed an early and sustained separation of survival curves, indicating a higher risk of mortality over time. In a Cox proportional hazards model adjusting for massive haemoptysis, massive haemoptysis remained a strong independent predictor of mortality (HR 4.26, 95% CI 1.70 to 10.68; p=0.002), whereas the association between PAFS and mortality was attenuated and no longer statistically significant (HR 1.23, 95% CI 0.49 to 3.07, online supplemental table S5). These findings suggest that the prognostic impact of PAFS on mortality may be largely mediated through its association with massive haemoptysis. Prophylactic pulmonary artery embolisation (PPAE) Five patients with PAFSs (n=3) and aneurysms (n=2) underwent PPAE based on imaging features (table 3). Table 3Characteristics and procedural information of five patients undergoing PPAEPatientAgeGenderComorbiditiesRadiological featuresBAEPrognosis134FemaleDMANSurvived248MaleDMPAFS, CYDeceased344MaleDMPAFS, CNSurvived434MaleDMAYSurvived569MaleCOVID-19, DMPAFS, CNSurvivedA, aneurysm; BAE, bronchial artery embolization; C, cavity; DM, diabetes mellitus; N, no; PAFS, pulmonary artery floating sign; PPAE, prophylactic pulmonary artery embolisation; Y, yes. Characteristics and procedural information of five patients undergoing PPAE A, aneurysm; BAE, bronchial artery embolization; C, cavity; DM, diabetes mellitus; N, no; PAFS, pulmonary artery floating sign; PPAE, prophylactic pulmonary artery embolisation; Y, yes. Technical success was achieved in all five cases, using nester coils, sizes ranging from 6 to 10 mm in diameter and 2–4 cm in length) to occlude the segmental or subsegmental pulmonary arteries implicated by imaging and angiography (figure 2). No procedure-related complications were reported. Three of the five patients also underwent concomitant bronchial artery embolisation, due to enlargement and tortuosity of bronchial arteries on CTA. During a median follow-up period of 30 days, four patients (80%) survived without recurrent haemoptysis. One patient, despite embolisation of a pseudoaneurysm and two bronchial artery-pulmonary arteries, died from rebleeding. No clinical signs of progressive right heart strain, hypoxia or other embolisation-related adverse events were reported in surviving patients. Image of patient undergoing prophylactic pulmonary artery embolisation (PPAE). (a) Enhanced chest CT reveals a concerning pulmonary artery floating sign (white arrow) and cavity. (b) Digital subtraction angiography shows the floating artery with the cavity (white arrow). (c) Four floating pulmonary artery branches were embolised using coils. (d,e) Subsequent treatment included intravenous administration of antifungal agents, along with bronchoscopic instillation for left airway management. Discussion In this retrospective study of patients with pulmonary mucormycosis, we identified the PAFS as a novel radiologic feature strongly associated with massive haemoptysis. This association remained consistent across multivariable and reduced models designed to minimise overfitting. Additionally, our preliminary evaluation of PPAE in high-risk patients demonstrated procedural feasibility and suggested a potential role in haemorrhage prevention. However, this observation should be interpreted as exploratory given the very small number of treated patients. These findings may help establish an early intervention strategy for one of the deadliest complications of pulmonary mucormycosis. Importantly, the PAFS demonstrated excellent interobserver agreement in our cohort (Cohen’s κ=0.836), suggesting that this imaging feature can be reproducibly identified in routine clinical practice by experienced radiologists. The PAFS, as defined in this study, refers to a segment of the pulmonary artery suspended within a necrotic cavity and partially detached from adjacent bronchi or lung parenchyma. Previous studies have reported this imaging feature in conditions such as lymphoma and hydatid disease.15 16 We hypothesise that this sign may reflect fungal angioinvasion—a hallmark of mucormycosis pathophysiology —leading to vascular wall degradation. We therefore propose the PAFS as a novel imaging finding characterised by one or more pulmonary artery branches suspended within a necrotic or cavitary lesion, often without obvious mural contact, creating a visual impression akin to floating twigs. This sign, to our knowledge, has not been formally described in prior radiological literature. Its presence may imply advanced tissue necrosis with exposure of vascular structures, serving as a potential imaging marker for increased risk of massive haemoptysis.15 17 Unlike pulmonary aneurysms or pseudoaneurysms, which involve vessel wall dilation due to contained haemorrhage, the floating sign could represent a pre-rupture state where the artery, deprived of surrounding tissue support, becomes highly prone to sudden rupture.18 An additional high-risk factor is that the floating artery lies within a cavity formed by necrotic material draining into the bronchus (indicating bronchial communication). In such cases, arterial rupture may result in rapid, massive haemorrhage, leading to fatal outcomes. Aneurysms and pseudoaneurysms are widely recognised as radiologic hallmarks of impending haemorrhage in mucormycosis, tuberculosis and cavitary lung malignancies.4 17 However, they are often identified only in the late stages of disease, when the arterial wall has already been extensively destroyed.12 In our study, aneurysms or pseudoaneurysms were rare (5.3%), and their presence was not significantly associated with haemoptysis in multivariate analysis. In contrast, the PAFS was more prevalent (50.9%) and more strongly correlated with bleeding risk. This suggests that the floating sign may either precede pseudoaneurysm formation or represent an alternative morphologic manifestation of arterial wall destabilisation. It is possible that some aneurysms or pseudoaneurysms develop after a short-lived floating phase, while others progress directly to rupture without significant dilation. The absence of visible wall discontinuity on CTA should not be taken as evidence of vascular integrity, particularly in patients with this radiologic feature. These findings have direct implications for clinical management. First, they support the routine use of CTA in patients with suspected pulmonary mucormycosis, particularly those with cavitary or necrotic lesions.6 Detection of a floating artery should prompt urgent multidisciplinary review and consideration of prophylactic embolisation before bleeding occurs. Second, in our experience, patients who underwent PPAE targeting the floating artery tolerated the procedure well and remained free from recurrent haemoptysis during short-term follow-up. Importantly, embolisation was technically feasible even when the affected arteries were embedded within necrotic lung, and no infarction-related complications were observed. Third, the identification and occlusion of high-risk vascular segments may facilitate subsequent fibreoptic bronchoscopy, allowing safer lavage and targeted antifungal administration. In the post-PPAE setting, bronchoscopy can also be used for clot removal or diagnostic sampling with lower risk of provoked bleeding. Taken together, these results suggest that the floating sign may serve as a clinically relevant imaging marker that could inform the timing of vascular assessment and interventional decision-making in selected patients. The application of this strategy may extend beyond mucormycosis. Cavitary lung abscesses and necrotic lung tumours may also exhibit floating arteries, and these patients have been reported to suffer from sudden, fatal haemoptysis in the absence of pseudoaneurysm.18 If validated in future studies, this imaging sign may represent a more general radiological indicator of vascular destabilisation within cavitary lung lesions. Integrating the floating sign into risk algorithms across a spectrum of cavitary lung diseases may help stratify haemorrhagic risk and select patients for early interventional radiology consultation. Several limitations merit consideration. First, this was a single-centre retrospective study with a limited sample size, which may constrain the generalisability of our findings. Second, contrast-enhanced CTA was not available for all patients, potentially underestimating the incidence of aneurysms and pseudoaneurysms or vascular invasion. Lastly, our study lacked long-term follow-up, and the durability of PPAE in preventing delayed bleeding remains to be established. To confirm these observations, prospective multicentre studies with larger cohorts are needed. Future work should focus on developing standardised radiologic criteria for the floating sign and evaluating its predictive value in different fungal and non-fungal cavitary lung diseases. Importantly, a controlled study comparing prophylactic embolisation versus conservative management in floating sign-positive patients could provide robust evidence to guide clinical decision-making. Conclusion The PAFS represents a novel radiologic feature associated with an increased risk of massive haemoptysis in pulmonary mucormycosis. Its presence may indicate severe vascular involvement and identify patients at particularly high risk of life-threatening bleeding. Recognition of this feature may facilitate earlier risk stratification and multidisciplinary decision-making. Prospective multicentre studies are needed to confirm these findings and to clarify whether early interventional strategies can improve clinical outcomes. Supplementary material