Sage Journals: Discover world-class research

Abstract

Level of Evidence

Level III – Retrospective Cohort Study.

Background

Musculoskeletal lesions commonly present as pain, masses, or incidental imaging findings. Accurate diagnosis is critical for treatment planning. Although open biopsy was historically considered the gold standard, core needle biopsy (CNB) has emerged as a minimally invasive and reliable diagnostic method with a favorable safety profile. The primary objective was to determine the diagnostic accuracy of image-guided core needle biopsy (CNB) for musculoskeletal lesions.

Methods

We conducted a retrospective analysis of 426 patients who underwent image-guided CNB for musculoskeletal lesions at a single tertiary referral center between January 2014 and December 2020. Patient demographics, biopsy characteristics, and diagnostic outcomes were analyzed. Diagnostic performance was evaluated using sensitivity, specificity, predictive values, and area under the ROC curve (AUC).

Results

Most patients were adults (82.4%) and male (56.6%). The lower limb was the most common biopsy region (43.2%), with the trunk (26.3%), upper limb (15.5%), and pelvis (14.3%) following in order of frequency. CNB demonstrated a sensitivity of 98.8%, specificity of 72.3%, negative predictive value of 99.0%, and overall accuracy of 82.3% (AUC = 0.855).

Conclusion

Image-guided CNB provides high diagnostic sensitivity and excellent negative predictive value for musculoskeletal lesions, with a very low complication rate. While it is effective as a first-line diagnostic tool, limitations remain in specificity and tumor subtype confirmation, particularly in heterogeneous or necrotic lesions. Standardized protocols and multidisciplinary interpretation are essential to optimize accuracy.

Keywords

core needle biopsy musculoskeletal accuracy

Introduction

Musculoskeletal tumors, including sarcomas, represent a diverse group of rare mesenchymal neoplasms that require accurate and timely diagnosis to guide treatment decisions. While these lesions may present as pain-inducing masses, benign soft tissue abnormalities, or incidental findings on imaging studies, distinguishing between benign and malignant pathologies is critical to avoid overtreatment or delayed management. Proper histopathological diagnosis is particularly crucial in the context of musculoskeletal oncology, where treatment strategies—including surgery, radiation, and systemic therapy—depend heavily on the tumor’s type, grade, and location.¹

Historically, open incisional biopsy was considered the gold standard for diagnosing musculoskeletal tumors. This technique offers high diagnostic accuracy, often reported between 91% and 99%, due to the ability to obtain larger and more representative tissue samples.^2–4 However, open biopsy carries several limitations, including the need for general anesthesia, the use of operating room resources, prolonged recovery, and higher risks of complications such as hematoma, infection, or tumor seeding along the biopsy tract.^5–9 Furthermore, open biopsy may delay diagnosis and increase healthcare costs, especially when performed inappropriately or in non-specialized centers. In response to these challenges, other methodologies have grown in popularity.^6,7 In contrast, image-guided core needle biopsy (CNB) has gained widespread adoption as a minimally invasive diagnostic alternative. CNB can be performed in radiology suites using modalities such as ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI), and often requires only local anesthesia. This allows for outpatient-based procedures with minimal patient burden, reduced complication rates, and lower overall costs.^4,6,7,9–13 Importantly, CNB can be used to obtain adequate tissue samples for histopathologic, immunohistochemical, and molecular analysis, providing sufficient diagnostic information in most cases. Nevertheless, certain potential negatives include the possibility of a decline in diagnostic accuracy as well as the possibility of a mistake occurring during the tumor sample.⁵

Numerous studies have reported diagnostic accuracy rates of CNB between 66 and 98%, depending on lesion location, imaging modality used, operator experience, and institutional protocols.^{10,11,14–24} Meta-analyses have demonstrated that CNB yields high diagnostic accuracy and outperforms fine-needle techniques in both benign and malignant bone tumors.^12,24 These results have led to CNB being recommended as a first-line biopsy method in many clinical guidelines, especially when performed in specialized musculoskeletal oncology centers. Despite its advantages, CNB is not without limitations. Concerns remain regarding sampling error, non-representative tissue acquisition, or inconclusive results due to necrotic or fibrotic lesions. In particular, certain anatomical locations or tumor subtypes may pose challenges to accurate diagnosis using CNB alone.¹⁰

Another important consideration is the complication profile of CNB compared to open biopsy. Studies have shown that CNB has significantly lower complication rates, often reported as 0–2%, whereas open biopsy complications can reach up to 16%.^4,6,7,9 Common minor complications include localized bleeding, hematoma, or transient pain, while major complications such as infection, neurovascular injury, or tumor seeding are exceedingly rare with proper technique. Additionally, the smaller incision used in CNB can often be incorporated into the definitive surgical approach, minimizing the impact on future procedures.^4,6,7,25

In clinical practice, the utility of CNB is also evident in its impact on time-to-diagnosis. With reduced need for operating room scheduling and hospital admission, CNB allows for faster turnaround from initial imaging to treatment planning.^11,17–21 This is particularly valuable in high-volume tertiary care settings, where resource optimization and patient throughput are priorities.^4,6,7

Despite the growing evidence supporting core needle biopsy, several gaps remain in the literature. Many prior studies are limited by small sample sizes or combine bone and soft tissue lesions, which may reduce generalizability. In addition, inconsistent reporting of nondiagnostic or inadequate samples can lead to overestimation of diagnostic accuracy. Regional data, particularly from Middle Eastern populations and healthcare systems that rely on centralized tertiary care models, are also scarce. Therefore, there is a need for large, single-center studies that comprehensively report the diagnostic accuracy and safety profile of image-guided CNB for musculoskeletal lesions.

Given these limitations, this study aims to evaluate the diagnostic accuracy, complication rate, and procedural characteristics of image-guided core needle biopsy for musculoskeletal lesions at a high-volume tertiary referral center. We present a retrospective analysis of 426 consecutive patients who underwent CNB for suspected bone or soft tissue tumors, using standardized radiologic guidance and pathology evaluation. Our primary objective was to determine the diagnostic sensitivity, specificity, and overall accuracy of CNB. Secondary objectives included analysis of complication rates, time to diagnosis, and the impact of factors such as lesion location, patient age, and tissue type (bone vs soft tissue) on diagnostic yield.

We hypothesize that CNB provides a high level of diagnostic accuracy comparable to open biopsy, with significantly lower complication rates and better logistical efficiency. By analyzing a large and diverse patient cohort, we aim to validate the use of image-guided CNB as a reliable diagnostic modality in musculoskeletal oncology and support its integration into diagnostic algorithms in tertiary care environments.

Materials and methods

This study was approved by the Institutional Review Board and conducted in accordance with the Declaration of Helsinki.²⁶ The potential benefits and risks of CNB were explained to all patients before the procedure, and the study adhered to STROBE guidelines for reporting observational research.²⁷

Study design

This retrospective study was performed between January 2014 and December 2020. All medical records from all patients who underwent CNB for a musculoskeletal mass were reviewed.

Data collection method

The data were collected retrospectively in a structured electronic data sheet. The data included patients’ demographics, operative-related data, diagnostic accuracy of CNB, and complications. All patients who underwent image-guided core needle biopsy for suspected bone or soft tissue lesions between January 2014 and December 2020 were eligible for inclusion. Patients were included if they had complete demographic, imaging, and histopathological data available. Patients were excluded if they had incomplete medical or pathology records; if they had undergone open or excisional biopsy as the initial diagnostic procedure; or if their CNB samples were nondiagnostic without subsequent confirmatory follow-up. Nondiagnostic CNB samples without a confirmed final diagnosis were excluded from accuracy calculations because diagnostic metrics require comparison against a verified reference standard. The average time from biopsy to pathology was 5.3 days, and there were no significant procedural complications observed. All image-guided core needle biopsies, whether targeting soft tissue or bone lesions, were performed under local anesthesia using standard aseptic technique.

Statistical analysis

Continuous normally distributed data were reported as mean and standard deviation (SD). Receiver operating characteristic (ROC) curve and area under ROC curve (AUC) were used to assess the diagnostic accuracy of core needle biopsy for musculoskeletal lesions. The sensitivity, specificity, positive predictive value (+PV), negative predictive value (-PV), positive likelihood ratio (+LR), and negative likelihood ratio (-LR) were calculated. All tests were considered significant when p < 0.05. Statistical analysis was performed using SPSS software version 25 for Windows (SPSS Inc., Chicago, IL, USA) and MedCalc software version 20.^28,29

Results

There were 426 patients in the study, most of whom were adults (82.4%) and men (56.6%). With a standard deviation of 23.2 years, the average age was 43.0 years. Core needle biopsy (CNB) was the most prevalent type of biopsy, making up 98.1% of all biopsies performed. An average of three biopsy passes was performed per patient. The mean duration from biopsy to pathology was 5.3 days, and no significant complications were related to the biopsies. The study participants’ demographic details and biopsy-related data are summarized in Table 1.

Table 1.

Patients’ demographic characteristics and biopsy-related information.

N	N%	Mean (SD)
Gender
Male	241	56.6%
Female	185	43.4%
Age in years		43.0 (23.2)
Age category
Pediatrics (<18)	75	17.6%
Adults (≥18)	351	82.4%
Type of biopsy
CNB	418	98.1%
Needle aspiration	2	0.5%
Excisional biopsy	6	1.4%
Number of biopsies		3.4 (1.3)
Complications
None	425	99.8%
Minimal bleeding	1	0.2%
Biopsy to pathology time in days		5.3 (5.2)

The lower limb (43.2%) was the most often biopsied region, with the trunk (26.3%), upper limb (15.5%), and pelvis (14.3%) following in order of frequency. In particular, the femur/thigh (24.6%), back/scapula/spine (23.1%), and groin/pelvis/pubic (11.7%) were the most frequently biopsied regions. The locations and sites of the biopsies that were done on the study participants are presented in Table 2.

Table 2.

Regions and sites of biopsies performed to the study participants.

Region of the mass		N	N%
Pelvis	Buttock/Gluteus	4	0.9%
	Hip	7	1.6%
	Groin/Pelvis/Pubic	50	11.7%
	Total	61	14.3%
Upper limb	Hand	5	1.2%
	Arm/Humerus	31	7.3%
	Shoulder/Axilla	15	3.5%
	Elbow/Forearm/Radius/Ulna	15	3.5%
	Total	66	15.5%
Lower limb	Ankle/Calcaneus/Foot	28	6.6%
	Femur/Thigh	105	24.6%
	Knee	17	4.0%
	Leg/Tibia/Fibula	34	8.0%
	Total	184	43.2%
Trunk	Rib/Sternum/Clavicle/Supraclavicular tissue	14	3.3%
	Back/Scapula/Spine	98	23.0%
	Total	112	26.3%
Others	Bone marrow	1	0.2%
	Skull	2	0.5%
	Total	3	0.7%

With a sensitivity of 98.8%, CNB could accurately diagnose 98.8% of patients with the intended ailment. With a specificity of 72.3%, CNB could accurately classify 72.3% of patients as not having the target condition. 68.4% of patients who tested positive for the target ailment had it, according to the positive predictive value of CNB. 99.0% of patients who tested negative for the target condition did not have it, according to the CNB’s negative predictive value of 99.0%. With an accuracy of 82.3%, CNB could properly identify the target condition in 82.3% of the patients. As evidenced by its area under the curve (AUC) of 0.855 (95% CI 0.819-0.892), CNB was a very reliable predictor of the intended condition. Figure 1 and Table 3 present the CNB diagnostic test accuracy metrics.

Figure 1.

Receiver Operating Characteristics (ROC) curve displaying diagnostic utility and Area Under the Curve (AUC) for CNB (AUC = 0.855).

Table 3.

CNB diagnostic test accuracy parameters.

Parameter	%	p-value
Sensitivity	98.8%
Specificity	72.3%
Positive predictive value	68.4%	<0.001*
Negative predictive value	99.0%
Accuracy	82.3%
Area under the curve (95% CI)	0.855 (0.819–0.892)

Accuracy was found to be slightly higher in males (83.8%) compared with females (80.3%), but the difference was not statistically significant (p = 0.546). Additionally, the accuracy of CNB was marginally greater in pediatric patients (86.5%) compared with adults (81.4%), although this difference was also not statistically significant (p = 0.19). Comparable diagnostic accuracy was observed between bone lesions (82.5%) and soft tissue lesions (81.9%) (p = 0.85). Table 4 compares the tissue type, age, and gender-specific diagnostic accuracy of CNB in biopsied patients.

Table 4.

Comparison of CNB diagnostic accuracy according to gender, age, and tissue type.

	Accuracy	AUC	p-value
Gender
Male	83.8%	0.865	0.546
Female	80.3%	0.843
Age category
Pediatrics (<18)	86.5%	0.9	0.19
Adults (≥18)	81.4%	0.845
Tissue type
Bone	82.5%	0.857	0.85
Soft tissue	81.9%	0.85

Specimen adequacy and repeat biopsies

Specimen adequacy was assessed for all CNB cases. A total of 11 biopsies (2.6%) were categorized as nondiagnostic or inconclusive. Among the nondiagnostic or inconclusive CNBs, seven patients (63.6%) required a repeat biopsy. Of these, 5 underwent repeat CNB and 2 proceeded directly to open biopsy. Definitive diagnoses were achieved in 3 of these repeat-biopsy cases.

Discussion

The results of this study provide convincing evidence regarding the diagnostic utility of percutaneous core needle biopsy (CNB) for musculoskeletal lesions. Our comprehensive assessment highlights the advantages of CNB as a minimally invasive and reliable alternative to open biopsy. In our cohort, CNB demonstrated an overall diagnostic accuracy of 82.3%, which is consistent with the literature and supports its clinical value.^{4,11,17–21,30–32} These findings contribute meaningfully to the growing body of evidence supporting the use of percutaneous techniques for obtaining accurate histopathologic diagnoses in a broad range of musculoskeletal tumors. Nonetheless, the modest specificity and positive predictive value indicate that CNB is more reliable for excluding malignancy than for precisely defining tumor subtype. This study is in alignment with findings by Siddiqi et al., who compared fine-needle aspiration with core needle biopsy and found CNB to be more effective in providing definitive tissue characterization. Similarly, previous studies have demonstrated that CNB provides dependable diagnostic information while minimizing patient morbidity.^33–35 However, certain sarcomas with necrosis, fibrosis, or heterogeneous architecture may be challenging to grade accurately using CNB alone. In such cases, an open biopsy or repeat CNB may still be necessary, and CNB findings should ideally be reviewed within a multidisciplinary team setting to support shared decision-making and ensure optimal diagnostic accuracy. Pediatric patients represented a relatively small proportion of our cohort (17.6%); therefore, the findings in children may not fully represent the broader pediatric population. An important aspect of this analysis was the comparison of complication rates between CNB and open biopsy methods. In our cohort, the complication rate associated with CNB was minimal, with only one minor bleeding episode (0.2%). This is substantially lower than the complication rates reported in some open biopsy series, which can range up to 16%.^4,6,7 These results reinforce the established notion that percutaneous approaches are safer and less invasive, offering a favorable safety profile that supports broader clinical adoption. Azabdaftari et al. emphasized the growing interest in CNB due to its safety and efficiency in the diagnosis of musculoskeletal lesions.³⁵ Reduced complication rates not only improve patient safety but also facilitate a smoother and faster recovery. This, in turn, minimizes the burden on healthcare systems and improves the overall patient experience. Another advantage of CNB observed in our study was the significantly shorter hospital stay. Patients undergoing CNB had an average hospital stay of 1.5 days, compared to longer admissions often associated with open biopsy procedures.³² This reduction in hospitalization enhances patient satisfaction and supports earlier recovery. These practical benefits align with the ongoing shift toward less invasive diagnostic strategies in orthopedic oncology. The quick recovery time associated with CNB also highlights its value in clinical workflows. Faster post-biopsy recovery facilitates earlier treatment planning, reduces delays, and may contribute to improved oncologic outcomes. Moreover, the use of CNB promotes more efficient utilization of operating rooms and inpatient beds, which is particularly beneficial in high-volume tertiary centers. With regard to lesion location, the lower limb was the most commonly biopsied region in our cohort. However, accuracy remained high across all anatomical sites, suggesting that CNB maintains its diagnostic utility regardless of biopsy site when performed in specialized centers. Despite the strengths of this study, some limitations should be acknowledged. Excluding nondiagnostic, unconfirmed biopsies likely increased the measured diagnostic accuracy and may not represent true clinical performance in all cases. The retrospective nature of the study may introduce selection bias, and the single-center design may limit the generalizability of the findings. Additionally, our patient population may not reflect the broader diversity of patients seen in other regions or healthcare systems. Future prospective, multicenter studies are needed to validate our findings and compare CNB directly with open biopsy under standardized protocols.

Conclusion

Footnotes

Acknowledgment

We would like to thank Suhair Ayman Altramsy for her significant contributions and for drawing the illustrations of the surgical procedure presented in the present study.

ORCID iD

Tariq Almunyif

Ethical considerations

All procedures performed in the study involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent for publication

Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

Author contributions

Each author took part in formulating the research idea, plan the study, and collect the data. All authors were responsible for assessing the quality of the included data, doing data analysis, and writing the manuscript. All authors agree to accept equal responsibility for accuracy of this paper and approve the publication of the final manuscript. We have no financial interest linked to this work.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets used in the present study are available from the corresponding author on reasonable request.*

References

Enneking

Spanier

Goodman

. A system for the surgical staging of musculoskeletal sarcoma. Clin Orthop Relat Res 1980; 153: 106–120.

Pouedras

Briand

Crenn

, et al. Non–image-guided core needle biopsies can be used safely to improve diagnostic efficiency for soft tissue tumors. Surg Oncol 2021; 37: 101518.

Seng

Png

Tan

. Accuracy of core needle biopsy for musculoskeletal tumours. J Orthop Surg 2013; 21(1): 92–95.

Skrzynski

Biermann

Montag

, et al. Diagnostic accuracy and charge-savings of outpatient core needle biopsy compared with open biopsy of musculoskeletal tumors. J Bone Joint Surg Am 1996; 78(5): 644–649.

Mitsuyoshi

Naito

Kawai

, et al. Accurate diagnosis of musculoskeletal lesions by core needle biopsy. J Surg Oncol 2006; 94(1): 21–27.

Mankin

Simon

. The hazards of the biopsy, revisited. Members of the musculoskeletal tumor society. J Bone Joint Surg Am 1996; 78(5): 656–663.

Mankin

Lange

Spanier

. The hazards of biopsy in patients with malignant primary bone and soft-tissue tumors. J Bone Joint Surg Am 1982; 64(8): 1121–1127.

Markel

Neumann

Steinau

. Appropriate techniques for musculoskeletal tumor biopsy. Orthop Rev 1994; 23(2): 176–180.

Mavrogenis

Altsitzioglou

Tsukamoto

, et al. Biopsy techniques for musculoskeletal tumors: basic principles and specialized techniques. Curr Oncol 2024; 31(2): 900–917.

10.

Goldsmith

Horwich

, et al.

Bone and soft-tissue lesions: what factors affect diagnostic yield of image-guided core-needle biopsy?

Radiology 2008; 248(3): 962–970.

11.

Shin

Amaral

Armstrong

, et al. Image-guided percutaneous biopsy of musculoskeletal lesions in children. Pediatr Radiol 2007; 37(4): 362–369.

12.

Arvind Kumar

Peddamadyam

Krishna

, et al. An image-guided percutaneous core needle biopsy for accurate assessment of musculoskeletal tumors: a targeted diagnosis. Cureus 2024; 16(5): e60757.

13.

Burke

ZDC

Lazarides

Gundavda

, et al. Open versus core needle biopsy in lower-extremity sarcoma: current practice patterns and patient outcomes. J Bone Joint Surg Am 2023; 105(Suppl 1): 57–64.

14.

Mills

Leake

Crawford

, et al. Concepts in musculoskeletal bone and soft tissue biopsy. Semin Muscoskel Radiol 2021; 25(6): 711–724.

15.

Daley

Reed

Peterson

. Strategies for biopsy of musculoskeletal tumors. Semin Roentgenol 2017; 52(4): 282–290.

16.

Szaro

Wong

Blain

, et al. The diagnostic value of ultrasound-guided percutaneous core needle biopsy of musculoskeletal soft tissue lesions. J Ultrason 2021; 21(84): 22–33.

17.

Hryhorczuk

Strouse

Biermann

. Accuracy of CT-guided percutaneous core needle biopsy for assessment of pediatric musculoskeletal lesions. Pediatr Radiol 2011; 41(7): 848–857.

18.

Heslin

Lewis

Woodruff

, et al. Core needle biopsy for diagnosis of extremity soft tissue sarcoma. Ann Surg Oncol 1997; 4(5): 425–431.

19.

Domanski

Akerman

Carlén

, et al. Core-needle biopsy performed by the cytopathologist: a technique to complement fine-needle aspiration of soft tissue and bone lesions. Cancer 2005; 105(4): 229–239.

20.

Rimondi

Staals

Errani

, et al. Percutaneous CT-guided biopsy of the spine: results of 430 biopsies. Eur Spine J 2008; 17(7): 975–981.

21.

De Marchi

Brach del Prever

Linari

, et al. Accuracy of core-needle biopsy after contrast-enhanced ultrasound in soft-tissue tumours. Eur Radiol 2010; 20(11): 2740–2748.

22.

Ogilvie

Torbert

Finstein

, et al. Clinical utility of percutaneous biopsies of musculoskeletal tumors. Clin Orthop Relat Res 2006; 450: 95–100.

23.

Virayavanich

Ringler

Chin

, et al. CT-guided biopsy of bone and soft-tissue lesions: role of on-site immediate cytologic evaluation. J Vasc Intervent Radiol 2011; 22(7): 1024–1030.

24.

Ferreira

FBMD

Puchnick

Garcia

, et al. Image-guided percutaneous needle biopsy for benign and malignant bone tumors: systematic review and meta-analysis. J Vasc Intervent Radiol 2023; 34(4): 623–632.

25.

Barrientos-Ruiz

Ortiz-Cruz

Serrano-Montilla

, et al.

Are biopsy tracts a concern for seeding and local recurrence in sarcomas?

Clin Orthop Relat Res 2017; 475(2): 511–518.

26.

Goodyear

Krleza-Jeric

Lemmens

. The declaration of Helsinki. BMJ 2007; 335(7621): 624–625.

27.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370(9596): 1453–1457.

28.

IBM Corp . IBM SPSS statistics for windows, version 25.0. IBM Corp; 2017.

29.

Schoonjans

. ROC curve analysis with MedCalc. MedCalc, 2018.

30.

Zhang

, et al. Analysis on diagnostic failure of US-guided core needle biopsy for soft tissue tumors. Res Diagn Interv Imaging 2023; 5: 100023.

31.

Adams

Potter

Pitcher

, et al. Office-based core needle biopsy of bone and soft tissue malignancies: an accurate alternative to open biopsy with infrequent complications. Clin Orthop Relat Res 2010; 468(10): 2774–2780.

32.

Qaisar

Ullah

Salman

, et al. Open vs percutaneous core needle biopsy for musculoskeletal malignancies. J Popul Ther Clin Pharmacol 2024; 31(1): 893–901.

33.

Siddiqi

Kim

Jede

, et al. Association of core needle biopsy tract resection with local recurrence in extremity soft tissue sarcoma. Skelet Radiol 2017; 46(4): 507–512.

34.

Hau

Kim

Kattapuram

, et al. Accuracy of CT-guided biopsies in 359 patients with musculoskeletal lesions. Skelet Radiol 2002; 31(6): 349–353.

35.

Azabdaftari

Goldberg

Wang

. Efficacy of on-site specimen adequacy evaluation of image-guided fine and core needle biopsies. Acta Cytol 2010; 54(2): 132–137.

Diagnostic accuracy and safety of core needle biopsy for musculoskeletal lesions: A retrospective cohort study from a tertiary referral center

Abstract

Level of Evidence

Background

Methods

Results

Conclusion

Keywords

Introduction

Materials and methods

Study design

Data collection method

Statistical analysis

Results

Specimen adequacy and repeat biopsies

Discussion

Conclusion

Footnotes

Acknowledgment

ORCID iD

Ethical considerations

Consent for publication

Author contributions

Funding

Declaration of conflicting interests

Data availability statement

References