Introduction

Glomus tumors are benign lesions originating from glomus bodies, modified neuromyoarterial structures involved in thermoregulation and blood flow control. They are rare neoplasms, accounting for less than 2% of all soft tissue tumors [1, 2]. The typical location is in the subungual region, although they can also occur less frequently in the gastrointestinal tract, lungs, intraosseous, and intraneural regions. The classic clinical presentation of these tumors is described by the triad of precise localization of the painful point, paroxysmal pain, and cold hypersensitivity (Fig. 1). However, some patients may not exhibit all of these symptoms, making diagnosis solely based on clinical data challenging [3, 4].

Fig. 1
figure 1

41-year-old woman with a glomus tumor is depicted in the clinical image. The glomus tumor is indicated by a white arrow and was observed after a partial nail removal

Full size image

Imaging examinations play a significant role in evaluating and differentiating subungual lesions due to their nonspecific clinical features. Among the available diagnostic modalities, high-frequency ultrasound (>18 MHz) stands out as a non-invasive method with good spatial resolution and widespread availability. In addition to characterizing glomus tumors and distinguishing them from other diagnoses, high-frequency ultrasound provides valuable information regarding lesion location and its relationship with adjacent structures, such as the distal phalanx, lateral folds, and nail plates, aiding in therapeutic planning. The primary treatment for these lesions is complete surgical excision to prevent tumor recurrence [5,6,7].

The objective of this study was to describe the most frequently observed imaging features of subungual glomus tumors using high-frequency ultrasound.

Materials and methods

Patient cohort

The population for this study was derived from a retrospective study titled “Ultrasonographic Evaluation of Subungual Glomus Tumors,” which commenced in 2019 and received approval from the Research Ethics Committee of the institution (No. 54001821.7.0000.5505). This is a retrospective study and it was part of a larger project studying nail alterations using high-frequency ultrasound, in which we collected data from all ultrasonographic examinations of the nail apparatus.

A retrospective evaluation of all nail apparatus examinations carried out at the Alta Excellence Diagnostic Clinic in Botafogo, Rio de Janeiro, Brazil, was conducted by C.C. between January 2019 and December 2022. The device used for this evaluation was the Aplio 800 - Canon Medical, equipped with high-frequency linear transducers operating at 24 MHz.

Clinical data were gathered from chart review and analyzing the indications provided in the examination requests. The clinical characteristics of the patients under evaluation included gender, age, cold hypersensitivity, paroxysmal pain, and the time interval between the onset of pain and the examination. Pain intensity was documented, whenever available in medical records, using a numerical pain scale ranging from 0 to 10, where 0 indicated no pain and 10 represented the most severe pain experienced.

Image analysis

The inclusion criteria for this study were confirmed clinical and histopathological evidence of glomus tumors. The ultrasonographic characteristics were evaluated by two radiologists: one with 18 years of experience in musculoskeletal radiology and another with 4 years of experience in general radiology. Inter- and intraobserver reliability was also assessed. The two radiologists were provided with sufficient time for assessment. After the initial categorizations, a second round of readings was conducted 20 days later without access to the initial assessments, and discussions on subungual lesion images were avoided among participants. An analysis of the reports was also conducted to determine the percentage of cases in which the possibility of a glomus tumor was suggested through the ultrasonographic report.

Ultrasound documentation was conducted following established protocols (visualizing the distal interphalangeal joint, nail matrix, nail bed, nail plates, observing the distal phalanx in two planes, and assessing vascularization of the nail apparatus) to ensure data uniformity for retrospective analysis. In this study, the examinations were indeed conducted by a radiologist at the request of a dermatology practice.

The following ultrasonographic aspects were assessed:

  1. 1-

    Size of the lesions in the largest axis, described in millimeters.

  2. 2-

    Boundary of the lesions, described as well-defined, partially defined, or poorly defined.

  3. 3-

    Echogenicity, with lesions categorized as hypoechoic or hyperechoic (Fig. 2).

  4. 4-

    Involvement of the nail plates, presenting irregularities or not.

  5. 5-

    Vascularity on power Doppler, specifying if it was considered discrete, moderate, or intense based on semiquantitative evaluation. Flow quantity was classified as discrete if there were one- or two-pixels containing flow (usually less than 1 mm in diameter). Flow was considered moderate if it involved less than 50% of the lesion and intense if more than 50% of the lesion had intense vascularity (Fig. 3).

  6. 6-

    Presence of the stalk sign, characterized by a hypoechoic area connecting the lesion to the adjacent tissue, showing intense flow on power Doppler and indicating the vascular origin (Fig. 3).

  7. 7-

    Bone remodeling, with varying degrees of depression in the adjacent distal phalanx without cortical erosion (Fig. 4).

  8. 8-

    Measurement of the loss of the circumference of the distal phalanx in the axial plane, measured through a semicircular projection of the altered phalanx, not previously described in the literature.

Fig. 2
figure 2

39-year-old woman with a glomus tumor is shown in the sagittal image of the nail apparatus. The image was captured using a 24-MHz probe and displays a nodular, hypoechoic lesion (delineated by the dashed line) causing bone remodeling (pointed out by the arrowhead). Importantly, the lesion does not affect the nail plates (as indicated by the arrow). This anomaly is situated within the nail bed

Full size image
Fig. 3
figure 3

58-year-old man with a glomus tumor is featured in the sagittal plane image obtained using a 24-MHz ultrasound with power Doppler. This image reveals the presence of the stalk sign (arrow), along with the hypervascularized glomus tumor

Full size image
Fig. 4
figure 4

42-year-old woman with a glomus tumor is depicted. A gray-scale ultrasound image in the axial plane of the right thumb nail bed illustrates the remodeling of the distal phalanx. A B-mode image in the axial plane. B The same image with annotations, where the dashed line represents the bony contour of the distal phalanx, the white dotted circle represents the lesion, and the area in red indicates the loss of circumference of the distal phalanx in the axial plane. The arrowhead highlights the alteration in the nail plates, which are bulging

Full size image

Statistical analysis

For the statistical analysis, the SPSS (Statistical Package for the Social Sciences) program, version 19.0, was used. In this study, inter- and intraobserver reliability was assessed using the Fleiss Kappa test. The interpretation of Kappa values was as follows: above 0.75 indicated excellent agreement, 0.5–0.75 signified good agreement, and values below 0.5 were indicative of poor agreement.

Results

Epidemiological aspects

For this study, 23 patients were selected, out of which 2 had more than one glomus tumor, and 1 patient experienced tumor recurrence, resulting in a total of 31 lesions. The study population comprised 20 females (86.9%) and 3 males (13.1%). The average age of the cases was 55.5 years (standard deviation = 3.56), with a median age of 56 years.

Intra- and inter-observer analysis

In terms of intraobserver analysis of ultrasonographic features, a high level of agreement was observed, with a Kappa value of 1.0 for the more experienced observer and 0.8 for the less experienced one. For the inter-observer analysis, classifications from the radiologists’ second reading were employed, demonstrating nearly complete inter-observer agreement with a Kappa value of 0.9. Subsequent statistical analyses exclusively utilized data from the more experienced observer.

Clinical characteristics evaluation

Details of the clinical characteristics of all cases in the study are provided in Table 1.

Table 1 Patients’ characteristics
Full size table

Clinical information of patients was acquired through examination requests and data provided by referring physicians. Pain intensity was graded using a numerical pain scale. Among the 23 patients, only 6 had information on pain ratings, totaling 11 lesions with pain scores available. The average pain score was 7.0 (standard deviation: 2.7), with the highest reported pain score being 8 and the lowest being zero.

Information about cold hypersensitivity was available for 30 out of 31 lesions, with 11 (36.7%) showing cold hypersensitivity and 19 (63.3%) not presenting this characteristic. As for paroxysmal pain, only 13 cases (46.4%) reported experiencing this type of pain, while 15 cases (53.6%) did not describe having this characteristic.

Data on the time interval between the onset of painful symptoms and the ultrasound examination were available for 21 out of the 31 glomus tumors evaluated. The mean time elapsed between symptom onset and ultrasound examination was 22.6 years (ranging from 1 to 38 years).

Evaluation of ultrasonographic characteristics

Details of the ultrasonographic characteristics of all cases in the study are provided in Table 2.

Table 2 Ultrasonographic characteristics of glomus tumors visualized by high-frequency ultrasound with 24-MHz transducer
Full size table

Among the 31 lesions studied, the average size was 4.6 mm (ranging from 1.6 to 11.8 mm). When assessing lesion boundaries, 12 (38.7%) were characterized as partially defined, 11 (35.4%) as well-defined, and 8 (25.8%) as poorly defined. Regarding echogenicity, 28 lesions (90.3%) were classified as hypoechoic, and 3 lesions (9.6%) were classified as hyperechoic (Fig. 2).

Ultrasonographic evaluation revealed nail plate involvement in only 9 cases (29%) among the adjacent nail plates to the lesions. Most cases (71%) did not show any type of onychodystrophy, such as irregularity, thickening, accentuation of curvature, or other damages to the nail plates identified through physical or ultrasonographic examination.

Lesions were categorized based on their vascularity on power Doppler as showing discrete, moderate, or intense vascularity. Seventeen lesions (54.8%) were considered to have intense vascularity, 5 lesions (16.1%) had moderate vascularity, and 9 lesions (29.1%) had discrete vascularity. The stalk sign was observed in 20 of the 31 glomus tumors (64.5%) (Fig. 3).

Among all examinations, the distal phalanges adjacent to glomus tumors were analyzed for the presence or absence of bone remodeling on the longitudinal image. Bone remodeling was observed in 27 cases (87.1%), and only 4 cases (12.9%) did not show this characteristic (Fig. 4).

Of the 31 glomus tumor cases, 27 (87.0%) presented some degree of loss of the circumference of the distal phalange. The measurement of depression of the distal phalange in the axial plane averaged 0.8 mm (ranging from 0.2 to 1.4 mm).

Among the reports analyzed, in 22 cases (70.9%), the radiologist performing the examination suggested the diagnostic hypothesis of a glomus tumor.

Discussion

In this investigation, 31 subungual glomus tumors were scrutinized, characterized by their small dimensions, with an average lesion size of 4.6 mm. A predominant proportion of these tumors exhibited partially defined boundaries, constituting 38.7% of the cases. Concerning echogenicity, the majority, at 90.3%, were categorized as hypoechoic. Vascularity, evaluated using power Doppler, revealed that 54.8% of the lesions displayed intense vascularity, consistent with the classic description in the existing literature, which characterizes glomus tumors as hypervascular, ovoid to round hypoechoic lesions on ultrasound [8]. Among the ultrasonographic characteristics observed in subungual glomus tumors in this study, one of the most relevant features contributing to the diagnosis is the presence of the stalk sign. The stalk sign is defined as a well-defined efferent artery connecting the tumor to the adjacent tissue [9]. In our sample, this sign was present in 64.5% of all evaluated tumors on power or color Doppler, consistent with previous studies [9,10,11]. The presence of this sign serves as an indicator supporting the diagnostic hypothesis, as glomus tumors originate from myoarteriovenous anastomoses [11].

Another significant aspect observed was the presence of loss of the circumference of the distal phalanx adjacent to the tumor. Among the 31 cases, 27 (87.1%) showed some degree of bone remodeling, resulting in the loss of the distal phalanx’s circular shape when observed in the axial plane (Fig. 5). A recent study by Sechi et al. [8] described that 51.7% of glomus tumors evaluated by ultrasonography exhibited a cup-shaped bone remodeling, characterized by a subtle concave erosion resembling a cup in the periosteum, without periosteal reaction, interruption of the cortical line, or loss of reverberation artifact. Similar findings were observed in our study, where bone remodeling was evident without any cortical discontinuity or erosions [12]. Additionally, Chiang et al. [1] demonstrated in another study that three-quarters (75%) of glomus tumors showed bone erosion on ultrasound while having normal radiographs, suggesting that ultrasonography has higher sensitivity in detecting bone alterations in the distal phalanx of these tumors. Moreover, the authors suggested that the sensitivity for detecting glomus tumors through ultrasonography increases from 86 to 93% when assessing the cortical margins of the distal phalanx. Notably, alterations as small as 1 mm in the cortical bone could be detected using ultrasonography. Although the evaluation in the axial plane had not been previously described in the literature, this finding holds great importance. Given that glomus tumors are very small, with millimeter-sized dimensions and non-infiltrative growth, the bone indentation observed in the axial plane becomes instrumental in locating the lesion during ultrasonographic examination. It has proven to be essential in identifying very small tumors, as reported by the authors of this dissertation based on their experience (Fig. 6). Therefore, incorporating axial plane evaluation into the ultrasonographic assessment of glomus tumors can significantly enhance diagnostic accuracy and improve the localization of minute lesions.

Fig. 5
figure 5

60-year-old man is diagnosed with a glomus tumor, which is responsible for inducing bone remodeling. A Clinical image of a male patient with painful onychodystrophy. B Ultrasound image in the sagittal plane showing mild bone remodeling (dashed line). C Ultrasound in the axial plane shows a small subungual glomus tumor causing remodeling of the adjacent distal phalanx. The measurement of the loss of circumference of the phalanx was 0.6 mm (male, 60 years - case 31)

Full size image
Fig. 6
figure 6

39-year-old woman with a glomus tumor. A Clinical image of onycholysis in a patient previously operated for glomus tumor. B Sagittal plane ultrasound image with high-frequency transducer shows mild bone remodeling (dashed line). C B-mode ultrasound image in the axial plane demonstrates the loss of circumference of the distal phalanx (dashed line), aiding in the localization of the lesion (double arrow) on the ulnar side of the first finger (female, 39 years - case 22)

Full size image

In our case series, among the 11 tumors with information on pain rating using the numerical pain scale, the average pain score was 7.0, indicating intense pain. Glomus tumors are well-known for being highly painful lesions, and pain is considered the most important symptom associated with these tumors. However, other clinical characteristics, such as cold hypersensitivity and paroxysmal pain, were only present in a subset of cases, highlighting the atypical presentation of glomus tumors and the challenges in diagnosing them based solely on clinical symptoms [13,14,15,16].

Ham et al. [17] demonstrated in their study that up to 23% of glomus tumors do not present the classical triad of symptoms. Furthermore, clinical criteria have a sensitivity ranging from 50 to 90% and take years to establish [15]. Relying solely on a physical examination to detect these lesions can lead to delays in tumor recognition and misdiagnosis [17]. Currently, the literature suggests that a preoperative imaging examination should be performed in suspected cases of glomus tumors. This not only assists in a more accurate diagnosis but also allows for precise tumor localization and provides additional data, such as lesion size [16, 17].

The majority of glomus tumors are sporadic, but there is a predisposition to multiple glomus tumors occurring simultaneously in patients with neurofibromatosis type 1. Among the 23 patients evaluated in this study, 2 (8.8%) had multiple tumors, with 1 (4.4%) of them having neurofibromatosis type 1 discovered later in adulthood.

Among the 23 patients evaluated, only one (4.4%) presented with tumor recurrence after excision. Tumor recurrence for glomus tumors is defined as the return of symptoms similar to the previous ones, along with clinical findings consistent with this diagnostic hypothesis. Its incidence ranges from 4 to 15%, given that surgical removal is curative, and recurrence arises from incomplete excision. A study conducted by Gandhi et al. [18] assessed 12 patients with glomus tumors and mapped these tumors during surgery. In this study, five patients experienced recurrence, and in all cases, the recurrent lesion developed in a different anatomical location from the original tumor, as observed through surgical mapping. Interestingly, the recurrent lesions were always found in the same digit, suggesting that recurrences may arise from pre-existing synchronous satellite lesions that were not visualized during the initial surgery. Thus, the term “recurrence” might be inadequate in this context [18]. In light of this, preoperative ultrasonography proves to be significant, as it can precisely map the tumor’s location and size, allowing for complete excision. Additionally, it enables a comprehensive examination of the matrix and nail bed non-invasively to identify possible pre-existing, yet asymptomatic lesions.

Regarding the epidemiological data, recent studies have described a twofold predilection for glomus tumors in women compared to men [8, 19]. In this case series, women accounted for 86% of the study population, aligning with the reported prevalence.

In our investigation, we placed particular emphasis on the use of high-frequency probes operating at 24 MHz, distinguishing our study from previous research that primarily employed low-frequency probes. The rationale behind this focus is the minute size of many subungual glomus tumors, which can measure only a few millimeters. High-frequency probes prove instrumental in augmenting the detection of these diminutive lesions. By doing so, we aim to augment the existing body of literature with compelling evidence supporting the efficacy of high-frequency probes. This may, in turn, facilitate the attainment of more precise and early diagnostic results. Moreover, our study bolsters the expanding realm of research advocating for the application of ultrasound as a diagnostic tool for subungual tumors. This advocacy is particularly pertinent in cases where alternative imaging modalities may fall short in providing informative assessments and could potentially overlook millimetric lesions [20, 21].

This study was subject to several limitations, primarily stemming from the inherent challenges associated with the rarity of the tumor under investigation. The limitations include a relatively small sample size due to the infrequency of glomus tumors, restricted availability of high-frequency ultrasound probes, a retrospective study design, the absence of a control group for comparative analysis, and incomplete clinical data for certain patients, which hindered the ability to establish a comprehensive clinical-radiological correlation. To address these limitations, the study was designed with a focus on utilizing available patient data and conducting a thorough analysis within the constraints of the sample size. Efforts were made to minimize biases through rigorous data collection and analysis techniques. While the absence of a control group remained a limitation, the study aimed to provide valuable insights into the ultrasonographic characteristics of subungual glomus tumors and their clinical implications within the given constraints.

Conclusion

In summary, the ultrasonographic evaluation of subungual glomus tumors offers a highly valuable diagnostic tool with significant clinical implications. This study has shown that the characteristic features, including the identification of the stalk sign and bone remodeling, play a pivotal role in enhancing the accuracy of diagnosis and treatment planning for these tumors. The recognition of these features aids in not only differentiating glomus tumors from other nail pathologies but also contributes to improving patient outcomes and minimizing recurrence rates.

Moreover, the non-invasive nature of ultrasound imaging makes it an appealing option for clinicians, as it can help reduce the need for more invasive procedures. The findings presented here underscore the importance of considering ultrasound as a valuable adjunctive diagnostic tool for subungual glomus tumors, thereby facilitating better-informed clinical decisions and ultimately enhancing the quality of patient care. Future research and larger-scale studies can further refine the utility of ultrasonography in the management of these rare but clinically significant lesions.