A predictive model combining contrast-enhanced ultrasound and Shamblin classification for risk stratification of internal carotid artery resection in carotid body tumor surgery

Introduction

Carotid body tumors (CBTs) are rare paragangliomas originating from chemoreceptor cells at the carotid bifurcation (1). Surgical resection is the primary treatment for CBTs, for which management of the internal carotid artery (ICA) represents the principal challenge (2-4). Preservation of the ICA is critical due to its essential role in cerebral perfusion; however, resection carries a high risk of severe complications, including massive bleeding, severe cranial nerve injury, and even ischemic stroke (5-7). When resection is unavoidable, complex vascular reconstruction is imperative to maintain adequate cerebral blood flow (2,3,8).

Preoperative planning has traditionally relied on the imaging-based Shamblin classification system, which categorizes CBTs into three grades (I–III) based on the extent of circumferential encasement of the adjacent ICA (9). While the Shamblin classification has proven useful in predicting surgical complications and outcomes, including hemorrhage, cranial nerve injury, and the need for ICA resection (9-12), it fails to address the critical determinant of ICA preservation: vascular wall infiltration.

The critical intraoperative determinant of ICA preservation is the integrity of the avascular plane between the tumor and the arterial adventitia, commonly referred to as the surgical “white line” (4,10,12-14). The presence of the plane allows for safe tumor dissection and the preservation of the ICA; while its absence renders the tumor inseparable, thereby necessitating the resection and subsequent reconstruction of the ICA (15,16). Consequently, the key preoperative question shifts from assessing the degree to which the artery is surrounded to determining whether the artery can be dissected free. As an anatomic measure, the conventional Shamblin classification cannot reliably answer this question. For example, tumors that partially encase the ICA may require resection if infiltration is present, while extensively encased tumors without infiltration can be safely dissected.

Recent studies have shown that adventitial infiltration, rather than mere encasement, is the critical determinant for ICA resection (4,9,17). However, as noted by Luna-Ortiz et al. (13), the depth of infiltration is frequently only apparent intraoperatively, making reliable preoperative prediction a significant challenge. Therefore, improving preoperative planning requires an imaging marker that can reliably indicate the extent of adventitial infiltration.

Contrast-enhanced ultrasound (CEUS) offers a potential solution. As a dynamic modality that visualizes real-time microvascular perfusion (18-21), it can often delineate the interface between the tumor and the ICA, which may correspond to the “white line”. Conventional computed tomography angiography (CTA) can reveal the interface between the CBT and ICA to a certain extent; however, its relatively low resolution and lack of dynamic imaging limit its sensitivity for detecting this interface (4). Further, CEUS avoids the radiation exposure and iodinated contrast agent required for CTA, making it suitable for a broader patient population.

Therefore, this study aimed to evaluate the tumor-ICA interface on CEUS (the CEUS interface) as a novel predictor of ICA resection. We also sought to develop a combined predictive model to provide a practical preoperative risk-stratification tool for accurately determining the risk of ICA resection in patients with CBTs. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2849/rc).

Methods

Study population and data collection

This prospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Peking Union Medical College Hospital (No. K6557), and informed consent was obtained from all individual participants. Consecutive patients scheduled for surgical resection of CBTs at our tertiary institution were enrolled in the study between December 2022 and March 2025. The inclusion criteria were as follows: (I) clinical and imaging findings suggestive of a lateral neck mass scheduled for surgical excision; (II) completion of a preoperative CEUS examination; (III) intraoperative confirmation of tumor location at the carotid bifurcation; and (IV) a definitive postoperative pathological diagnosis of paraganglioma. The exclusion criteria were as follows: (I) a history of prior surgery, radiotherapy, or embolization therapy in the affected neck region; (II) incomplete clinical data; and/or (III) preoperative CEUS images of non-diagnostic quality.

During the study period, 71 suspected CBTs in 66 patients were assessed for eligibility. Twelve lesions were excluded: seven due to incomplete clinical data (five under observation without surgery and two lost to follow-up) and five due to non-diagnostic CEUS image quality (four with severe motion artifacts and one with poor contrast enhancement). All five lesions with non-diagnostic CEUS images underwent uneventful surgery with ICA preservation. Ultimately, 54 patients with 59 lesions were included in the final analysis. Demographic information, including age, gender, family history, residence, and clinical presentation, was systematically collected.

Imaging protocol

All 59 lesions underwent conventional ultrasound followed by CEUS, which was performed by a radiologist with 10 years of CEUS experience using an IU22 or EPIQ Elite system (Philips Medical Systems, Bothell, WA, USA). Patients were positioned supine with the head turned contralateral to the tumor and instructed to breathe quietly to minimize motion. For each lesion, grayscale and color Doppler images were acquired in both axial and sagittal planes. CEUS was then performed on the axial plane showing the maximal ICA encasement, using a low mechanical index (0.06) to prevent microbubble destruction. The ultrasound contrast agent SonoVue (Bracco, Italy) was diluted in 5 mL of saline and agitated until a homogeneous suspension was achieved; 1.2 mL of this suspension was subsequently injected into the median cubital vein. The ultrasound contrast mode was activated upon injection, and continuous scanning was maintained for 120 seconds, with all examinations saved as cine loops.

All patients underwent head-neck CTA before surgery to evaluate the morphology, anatomy, and surrounding neurovascular structures of the CBTs, either at our institution or elsewhere. Of these, 39 lesions were evaluated at our institution on the 320-row detector CT scanner (Aquilion ONE Genesis Edition, Canon Medical Systems). Patients were positioned supine with their arms at their sides, and scanning extended from the aortic arch to the skull. A bolus of 55 mL of iodinated contrast agent (Ultravist, Bayer, 370 mgI/mL) followed by 25 mL of saline was administered intravenously at 4 mL/s via the right median cubital vein. Arterial phase acquisition was triggered using bolus tracking (threshold 180 HU at the thoracic aorta), with a delayed phase obtained 90 s after contrast injection.

Imaging analysis

Tumor size was measured as the maximum diameter on grayscale ultrasound. The Shamblin grade was assigned according to the extent of ICA encasement by the tumor, based on established criteria in the literature: Grade I indicates no encasement, Grade II partial encasement, and Grade III complete encasement (9). The CEUS interface was evaluated on axial images on CEUS and classified as either present or absent (Figure 1). A present interface was defined by a continuous, distinct hypo-enhanced boundary separating the tumor parenchyma from the ICA (Figure 1A), whereas an absent interface was characterized by the loss of this boundary (Figure 1D).

Figure 1 Representative preoperative images and intraoperative outcomes of CBTs. (A,B) Axial CEUS (A) and CTA (B) images of a left-sided CBT showing a preserved tumor-ICA interface (arrows). (C) Intraoperative photograph showing complete tumor dissection without ICA resection. (D,E) Axial CEUS (D) and CTA (E) images of a right‑sided CBT showing loss of the interface (arrows). (F) Intraoperative photograph showing tumor resection together with the involved ICA segment (arrow). CBT, carotid body tumor; CEUS, contrast‑enhanced ultrasound; CTA, computed tomography angiography; ICA, internal carotid artery.

To compare the performance of CEUS and CTA in revealing the tumor-ICA interface, we also collected all available CTA images of 39 lesions at our institution. In this subgroup, the tumor-ICA interface (the CTA interface) was assessed similarly on axial CTA images, focusing on the presence or loss of a distinct fat plane or tissue boundary between the tumor and the ICA (Figure 1B,1E) (4,12).

The CEUS and CTA interfaces were independently evaluated by two radiologists who had received standardized training and were blinded to the surgical records.

Surgical protocol

All surgeries were performed by the same team following a standardized resection protocol (22,23). All patients underwent CBT surgery under general anesthesia via a longitudinal excision along the anterior margin of the sternocleidomastoid muscle. Following dissection of the subcutaneous tissues, the common carotid artery, ICA, external carotid artery (ECA), and jugular vein were exposed and controlled with vessel loops, with the cranial nerves carefully preserved. After careful identification and ligation of the feeding vessels, tumors were removed from carotid arteries along the sub-adventitia plane using a combination of blunt and sharp dissection. In cases with severe vascular attachment, the ECA was sometimes ligated to reduce the risk of massive bleeding and facilitate tumor excision from the posterior aspect of the carotid bifurcation. In cases with severe ICA encasement, ICA rupture or even resection of the involved ICA or carotid bifurcation was sometimes required; in such cases, ICA repair with or without patching, ICA bypass using an autologous vein graft, commonly the great saphenous vein, or stent implantation was performed to restore cerebral flow and prevent ischemic stroke. In cases with CBTs extending close to the skull base, the mastoid process and part of the external auditory canal were removed to expose the intrapetrous segment of the ICA for vascular anastomosis and complete tumor excision.

The primary outcome of this study was ICA management, which was classified as ICA preservation or ICA resection according to the surgical procedure. ICA preservation was defined as the successful dissection of the tumor from the artery with the patient’s own ICA remaining in place, with or without vascular repair. Conversely, ICA resection referred to cases in which the tumor was tightly adherent to and inseparable from the artery, necessitating segmental resection of the involved ICA followed by reconstruction. Representative intraoperative images are shown in Figure 1.

Statistical analysis

The statistical analyses were performed using SPSS (version 29.0.2.0) and R (version 4.3.1). All the tests were two-tailed, and a P value <0.05 was considered statistically significant. Data normality was assessed using the Shapiro-Wilk test. Normally distributed continuous variables are presented as mean ± standard deviation and were compared between groups using the independent t-test. Non-normally distributed variables are expressed as median (interquartile range) and were compared using the Mann-Whitney U test. Categorical variables are reported as frequencies (percentages), with between-group comparisons performed using Fisher’s exact test.

Univariate analyses were performed for age, tumor size, Shamblin grade, and CEUS interface. Shamblin grade was categorized as Grade I/II and Grade III in accordance with previous studies and clinical practice (10,13,16,24). Variables with P<0.05 in the univariate analysis were included in a multivariate logistic regression model. To account for the limited number of events, Firth’s penalized-likelihood regression was employed to obtain robust estimates (25), and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated.

To assess predictive performance, models were built based on each independent predictor separately, as well as a model combining all predictors. The predictive efficacy of each model was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC), with pairwise comparisons conducted using the DeLong’s test. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were also calculated. Predicted probabilities from the combined model were then used to stratify lesions into distinct risk groups.

In the subgroup with both CEUS and CTA available (n=39), McNemar’s test was used to compare the diagnostic agreement between the two modalities for identifying lesions requiring ICA resection. Inter-observer agreement for the CEUS interface and CTA interface was evaluated using Cohen’s kappa coefficient, which classifies agreement levels as slight (0.01–0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), and near-perfect (0.81–0.99) (26).

Results

Baseline patient characteristics

Of the 71 lesions in the 66 patients initially assessed for eligibility, 12 were excluded due to incomplete clinical data (n=7) or non-diagnostic CEUS image quality (n=5), resulting in a final cohort of 54 patients with 59 CBT lesions. The baseline demographic data and lesion characteristics are detailed in Table 1. The median age of the patients was 38.0 years (interquartile range, 33.0–55.0 years), and 26 (48.1%) were male and 28 (51.9%) were female. A family history of CBT was reported in 3 patients (5.6%), and 9 patients (16.7%) were long-term residents at altitudes of 2,500 m or higher. Symptoms were present in 13 patients (24.1%), of which the most common was syncope (9.3%). Tumors were left-sided in 28 cases (51.9%), right-sided in 21 (38.9%), and bilateral in 5 (9.3%), with a mean size of 3.6±1.2 cm. Preoperative color Doppler ultrasound assessment of the feeding artery origin showed that the tumors received blood supply from both the ICA and ECA in 28 lesions (47.5%), exclusively from the ECA in 27 lesions (45.8%), and exclusively from the ICA in 4 lesions (6.8%). According to the Shamblin classification, 27 lesions (45.8%) were grade I, 20 (33.9%) grade II, and 12 (20.3%) grade III. Intraoperatively, 47 lesions (79.6%) were resected with ICA preservation, while 12 (20.3%) required ICA resection and reconstruction. Among the 47 lesions with ICA preservation, 12 (25.5%) required vascular repair. Among the 12 patients who underwent ICA resection, 11 (91.7%) received autologous great saphenous vein interposition grafting, and 1 (8.3%) underwent stent implantation. In one case with tumor extension to the skull base, a complementary mastoidectomy was performed to expose the intrapetrous segment of the ICA.

Predictors of intraoperative ICA resection

Table 2 presents the univariate and multivariable analyses of predictors for intraoperative ICA resection. The univariate analyses showed that tumor size, Shamblin grade, and the CEUS interface were all significantly associated with ICA resection (all P<0.001). Multivariable analysis identified two independent predictors: Shamblin grade (OR 11.92, 95% CI: 1.02–314.23, P=0.049) and CEUS interface (OR 40.46, 95% CI: 4.55–1,190.09, P<0.001). The wide CIs are likely attributable to the limited number of ICA resection events (n=12). Based on these two factors, we developed a combined logistic regression model to estimate the probability of ICA resection, formulated as follows:

P=ex(1+ex)[1]

where P represents the predicted probability of ICA resection, and the linear predictor x is calculated as:

x=−3.453+4.072×(CEUSinterface)+2.640×(Shamblingrade)[2]

In this model, two binary variables were defined: the CEUS interface was coded as 1 if absent and 0 if present; and the Shamblin grade was coded as 1 for grade III and 0 for grades I/II.

Predictive model performance

The predictive performance of the combined model (Eq. [1]), along with the models based on single predictors, is summarized in Table 3, while the ROC curves are presented in Figure 2. The combined model (Model 3) exhibited the best diagnostic performance, achieving an AUC of 0.949 (95% CI: 0.860–1.000), a sensitivity of 91.7%, and an NPV of 97.7%. The model based on the CEUS interface (Model 1) achieved the highest accuracy (94.9%), specificity (97.9%), and PPV (90.9%), with an AUC of 0.906 (95% CI: 0.794–1.000). The model based on the Shamblin classification (Model 2) yielded an AUC of 0.843 (95% CI: 0.710–0.976) and an accuracy of 89.8%, a sensitivity of 75%, and a specificity of 93.6%. Pairwise comparisons conducted by DeLong’s test revealed that the differences in AUC between any two of the three models were not statistically significant (range, 0.071–0.416), which may be attributable to the limited number of ICA resection events.

Figure 2 ROC curves of the three predictive models for ICA resection in CBT surgery. The red curve (Model 1) represents the CEUS interface-based model (AUC =0.906, 95% CI: 0.794–1.000). The blue curve (Model 2) represents the Shamblin classification-based model (AUC =0.843, 95% CI: 0.710–0.976). The gray curve (Model 3) represents the combined model integrating both predictors (AUC =0.949, 95% CI: 0.860–1.000). AUC, area under the curve; CBT, carotid body tumor; CEUS, contrast-enhanced ultrasound; CI, confidence interval; ICA, internal carotid artery; ROC, receiver operating characteristic.

Risk prediction based on the combined model

Based on the combined model of the CEUS interface and Shamblin grade, the predicted probabilities (P) derived from Eq. [1] were used to stratify the patients into four distinct risk tiers (Table 4). The likelihood of needing ICA resection was 3.1% for lesions with an interface and Shamblin grade I/II, 96.3% for those without an interface and Shamblin grade III, 64.9% for those without an interface and Shamblin grade I/II, and 30.7% for those with an interface and Shamblin grade III. Figure 3 demonstrates a practical application of the risk-stratification system: a tumor was classified as Shamblin grade II with no CEUS interface; during surgery, the tumor was densely adherent to the ICA with no discernible avascular plane, requiring ICA resection and reconstruction.

Figure 3 A representative case highlighting the preoperative predictive role of the tumor-ICA interface in a 38‑year‑old man with a left‑sided CBT. (A,B) Axial grayscale and power Doppler ultrasound images showing a mass partially encasing the ICA, corresponding to Shamblin grade II. (C) Axial CEUS image showing loss of the tumor-ICA interface (arrows). (D) Axial CTA image showing loss of the fat plane between the tumor and the ICA (arrow). Intraoperatively, the tumor was densely adherent to the ICA with no discernible avascular plane, requiring ICA resection and reconstruction. CBT, carotid body tumor; CEUS, contrast-enhanced ultrasound; CTA, computed tomography angiography; ICA, internal carotid artery.

Comparison of diagnostic agreement for the tumor-ICA interface: CEUS versus CTA

In the subgroup (n=39), McNemar’s test revealed no statistically significant difference between CEUS and CTA in identifying lesions requiring ICA resection (P=0.453). Specifically, CEUS correctly identified five lesions requiring ICA resection that were not detected by CTA. Conversely, CTA successfully identified two lesions requiring ICA resection that were not detected by CEUS. Both imaging modalities failed to accurately classify one lesion. Inter-observer agreement for the CEUS interface was near-perfect (κ=0.838), and that for the CTA interface was substantial (κ=0.760). However, this subgroup analysis was limited by its small sample size and should be interpreted as hypothesis-generating only.

Discussion

This was the first study to systematically evaluate the CEUS interface as a novel predictor of ICA resection in CBT surgery. We found that loss of the CEUS interface was a strong independent predictor (adjusted OR =40.46), outperforming the conventional Shamblin classification based on encasement alone (adjusted OR =11.92). The observed interface may correspond to the preservation of the surgical avascular plane between the tumor and arterial adventitia (4,10,13,14). Its loss on imaging may be associated with adventitial infiltration, which may hinder safe tumor dissection and necessitate ICA resection.

The CEUS interface was assessed qualitatively (present vs. absent), as this binary distinction directly addresses the key surgical question of whether a dissectible plane exists. This approach aligns with intraoperative decision-making processes and provides an intuitive clinical tool. Quantitative perfusion analysis was not conducted because the tumor-ICA interface is a thin, linear boundary that is prone to motion artifacts from patient breathing or probe movement, making region-of-interest placement unreliable and inconsistent across observers.

Our findings corroborate and extend previous imaging findings. The importance of a clearly defined lucent plane between the tumor and the ICA was first noted in an earlier angiographic study (12). Further, a recent study employing CTA found that the tumor-ICA interface is an independent predictor of ICA resection (4). Our study extends this concept to CEUS, demonstrating diagnostic agreement comparable to that of CTA in paired analyses (P=0.453). These findings suggest that CEUS may serve as a promising non-invasive alternative for preoperative assessment, particularly when CTA is contraindicated or suboptimal. However, this comparison was based on a small subgroup (n=39) and therefore remains preliminary. Larger prospective studies with adequate sample sizes are needed to directly compare the diagnostic accuracy of CEUS and CTA in predicting ICA resection. The near-perfect inter-observer agreement (κ=0.838) for assessing this CEUS interface further supports the reproducibility and clinical utility of this qualitative CEUS assessment.

The principal clinical advance of our work lies in integrating the CEUS interface with the Shamblin grade. The combined model showed promising predictive capability (AUC =0.949), compared with the models based on the CEUS interface alone (AUC =0.906) or the Shamblin grade alone (AUC =0.843). This model was translated into a pragmatic four-tier risk-stratification system for ICA resection. Specifically, when the CEUS interface was absent and the Shamblin grade was III, the probability of ICA resection reached 96.3%, warranting preoperative preparation for vascular reconstruction and ensuring the availability of appropriate graft materials. When the interface was absent and the Shamblin grade was I/II, the probability was 64.9%, warranting preoperative preparation for possible vascular reconstruction and thorough patient counseling. When the interface was present and the Shamblin grade was III, the probability was 30.7%, warranting standard dissection with careful intraoperative vigilance and preparedness to adjust the surgical strategy if the sub-adventitial plane proves indistinct. When the CEUS interface was present and the Shamblin grade was I/II, the estimated probability was only 3.1%, warranting standard surgical dissection without special preoperative vascular preparation. This stratification provides surgeons with a graded preoperative alert system to facilitate individualized surgical planning, ranging from complete tumor resection to preparation of a vascular reconstruction team and necessary graft materials. Additionally, the provision of preoperative education about the surgical procedure may improve patient counseling and better help manage expectations.

The observed association between loss of the CEUS interface and the need for ICA resection suggests that this imaging finding may reflect adventitial tumor infiltration, which is known to preclude safe dissection along the sub-adventitial plane. However, histopathological confirmation of this relationship is currently lacking. Future studies should aim to establish the histopathological correlation between the loss of the CEUS interface and adventitial tumor infiltration, as this would provide a stronger biological basis for its use as a predictor of ICA resection.

This study had several limitations. First, the limited number of ICA resection events resulted in wide CIs for the adjusted ORs. This reflects statistical uncertainty due to the sample size, rather than any inherent instability of the predictors. Second, this study was conducted at a single tertiary center by a specialized vascular team, and no internal or external validation was performed. Therefore, the reported AUC of 0.949 may be optimistic and requires confirmation in independent cohorts. Future studies should include both internal and external validation in multicenter settings to assess model generalizability. Finally, genetic testing was not systematically performed in this study; only patients with bilateral tumors, young age, or a positive family history were referred for genetic counseling. Therefore, future studies should incorporate routine genetic screening to better elucidate the relationship between genetic status and imaging features.

Conclusions

Although our model is exploratory and hypothesis-generating, the tumor-ICA interface on CEUS is a novel, reproducible imaging marker that may predict the need for ICA resection in CBT surgery. Its integration with the Shamblin classification system yields a highly accurate predictive model, which has been translated into a practical, four-tier risk-stratification tool for preoperative planning. Future studies should aim to establish the histopathological correlation of this interface and explore whether quantitative CEUS parameters can provide further refinement to the predictive model.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2849/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2849/dss

Funding: This work was supported by the National High Level Hospital Clinical Research Funding (Nos. 2022-PUMCH-B-064 and 2022-PUMCH-B-100) and the Special Research Fund for Central Universities, Peking Union Medical College (No. 3332024001).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2849/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Peking Union Medical College Hospital (No. K6557) and informed consent was taken from all individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Lin EP, Chin BB, Fishbein L, Moritani T, Montoya SP, Ellika S, Newlands S. Head and Neck Paragangliomas: An Update on the Molecular Classification, State-of-the-Art Imaging, and Management Recommendations. Radiol Imaging Cancer 2022;4:e210088. [Crossref] [PubMed]
2. Plukker JT, Brongers EP, Vermey A, Krikke A, van den Dungen JJ. Outcome of surgical treatment for carotid body paraganglioma. Br J Surg 2001;88:1382-6. [Crossref] [PubMed]
3. Gu G, Wu X, Ji L, Liu Z, Li F, Liu B, Liu C, Ye W, Chen Y, Shao J, Zeng R, Song X, Guan H, Zheng Y. Proposed modification to the Shamblin’s classification of carotid body tumors: A single-center retrospective experience of 116 tumors. Eur J Surg Oncol 2021;47:1953-60. [Crossref] [PubMed]
4. Jasper A, Mammen S, Gowri MS, Keshava SN, Selvaraj D. Imaging criteria to predict Shamblin group in carotid body tumors - revisited. Diagn Interv Radiol 2021;27:354-9. [Crossref] [PubMed]
5. van Asperen de Boer FR, Terpstra JL, Vink M. Diagnosis, treatment and operative complications of carotid body tumours. Br J Surg 1981;68:433-8. [Crossref] [PubMed]
6. Amato B, Bianco T, Compagna R, Siano M, Esposito G, Buffone G, Serra R, de Franciscis S. Surgical resection of carotid body paragangliomas: 10 years of experience. Am J Surg 2014;207:293-8. [Crossref] [PubMed]
7. Li FD, Gao ZQ, Ren HL, Liu CW, Song XJ, Li YF, Zheng YH. Pre-reconstruction of cervical-to-petrous internal carotid artery: An improved technique for treatment of vascular lesions involving internal carotid artery at the lateral skull base. Head Neck 2016;38:E1562-7. [Crossref] [PubMed]
8. Jiang X, Fang G, Guo D, Xu X, Chen B, Jiang J, Dong Z, Fu W. Surgical Management of Carotid Body Tumor and Risk Factors of Postoperative Cranial Nerve Injury. World J Surg 2020;44:4254-60. [Crossref] [PubMed]
9. Shamblin WR. ReMine WH, Sheps SG, Harrison EG Jr. Carotid body tumor (chemodectoma). Clinicopathologic analysis of ninety cases. Am J Surg 1971;122:732-9.
10. Kim GY, Lawrence PF, Moridzadeh RS, Zimmerman K, Munoz A, Luna-Ortiz K, et al. New predictors of complications in carotid body tumor resection. J Vasc Surg 2017;65:1673-9. [Crossref] [PubMed]
11. Pellitteri PK, Rinaldo A, Myssiorek D, Gary Jackson C, Bradley PJ, Devaney KO, Shaha AR, Netterville JL, Manni JJ, Ferlito A. Paragangliomas of the head and neck. Oral Oncol 2004;40:563-75. [Crossref] [PubMed]
12. Knight TT Jr, Gonzalez JA, Rary JM, Rush DS. Current concepts for the surgical management of carotid body tumor. Am J Surg 2006;191:104-10. [Crossref] [PubMed]
13. Luna-Ortiz K, Rascon-Ortiz M, Villavicencio-Valencia V, Herrera-Gomez A. Does Shamblin’s classification predict postoperative morbidity in carotid body tumors? A proposal to modify Shamblin’s classification. Eur Arch Otorhinolaryngol 2006;263:171-5. Erratum in: Eur Arch Otorhinolaryngol 2006;263:1161.
14. Gordon-Taylor G. On carotid tumours. Br J Surg 1940;28:163-72.
15. Smith JJ, Passman MA, Dattilo JB, Guzman RJ, Naslund TC, Netterville JL. Carotid body tumor resection: does the need for vascular reconstruction worsen outcome? Ann Vasc Surg 2006;20:435-9. [Crossref] [PubMed]
16. Alanezi T, Alomran F, Koussayer S, Abdulrahim O, Dahman M, Alsuhaibani E, Alokaili R, Al-Omran M. Preoperative radiological features in predicting complications of carotid body tumor resection. J Vasc Surg 2025;81:665-671.e2. [Crossref] [PubMed]
17. Ozawa H. Current management of carotid body tumors. Auris Nasus Larynx 2024;51:501-6. [Crossref] [PubMed]
18. Dietrich CF, Averkiou MA, Correas JM, Lassau N, Leen E, Piscaglia F. An EFSUMB introduction into Dynamic Contrast-Enhanced Ultrasound (DCE-US) for quantification of tumour perfusion. Ultraschall Med 2012;33:344-51. [Crossref] [PubMed]
19. Clevert DA, D’Anastasi M, Jung EM. Contrast-enhanced ultrasound and microcirculation: efficiency through dynamics--current developments. Clin Hemorheol Microcirc 2013;53:171-86. [Crossref] [PubMed]
20. Zhang C, Wu J, Ma S, Zheng X. Contrast-enhanced ultrasound for the prediction of intraoperative blood loss in patients undergoing resection for carotid body paraganglioma. Quant Imaging Med Surg 2025;15:2766-73. [Crossref] [PubMed]
21. Li X, Staub D, Rafailidis V, Al-Natour M, Kalva S, Partovi S. Contrast-enhanced ultrasound of the abdominal aorta - current status and future perspectives. Vasa 2019;48:115-25. [Crossref] [PubMed]
22. Gu G, Wang Y, Liu B, Chen Y, Shao J, Li F, Wu X, Cui L, Lu X, Liu C, Guan H, Gao Z, Feng G, Zheng Y. Distinct features of malignant carotid body tumors and surgical techniques for challengeable lesions: a case series of 11 patients. Eur Arch Otorhinolaryngol 2020;277:853-61. [Crossref] [PubMed]
23. Gao J, Gu G, Zeng R, Chen Y, Ye W, Zheng Y. Incidence and risk factors of cerebral hyper-perfusion syndrome like symptoms after resection of carotid body tumours. Eur J Surg Oncol 2025;51:110153. [Crossref] [PubMed]
24. Metheetrairut C, Chotikavanich C, Keskool P, Suphaphongs N. Carotid body tumor: a 25-year experience. Eur Arch Otorhinolaryngol 2016;273:2171-9. [Crossref] [PubMed]
25. Firth D. Bias reduction of maximum likelihood estimates. Biometrika 1993;80:27-38.
26. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.