Diagnosis, management, and complications of glomus tumours of the digits in neurofibromatosis type 1
- Douglas R Stewart1,
- Jennifer L Sloan2,
- Lawrence Yao3,
- Andrew J Mannes4,
- Armin Moshyedi5,
- Chyi-Chia Richard Lee6,
- Raf Sciot7,
- Luc De Smet8,
- Victor-Felix Mautner9,
- Eric Legius10
- 1Genetic Disease Research Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland, USA
- 2Genetic and Molecular Biology Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland, USA
- 3Department of Radiology, Clinical Center, National Institutes of Health, Bethesda, Maryland, USA
- 4Department of Anaesthesia and Surgical Services, Clinical Center, National Institutes of Health, Bethesda, Maryland, USA
- 5Clinical Center, National Institutes of Health, Bethesda, Maryland, USA
- 6Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA
- 7Department of Pathology, Catholic University Leuven, Leuven, Belgium
- 8Department of Orthopaedic Surgery, University Hospital Pellenberg, Lubbeek, Belgium
- 9Laboratory for Tumour Biology and Developmental Disorders, Department of Maxillofacial Surgery, University Hospital Eppendorf, Hamburg, Germany
- 10Department of Human Genetics, Catholic University Leuven, Leuven, Belgium
- Correspondence to Dr Douglas Stewart, Genetic Disease Research Branch, National Human Genome Research Institute, National Institutes of Health, 49 Convent Drive, Building 49, Room 4A62, Bethesda, MD 20892, USA;
- Received 9 October 2009
- Revised 30 November 2009
- Accepted 5 December 2009
- Published Online First 7 June 2010
Background Glomus tumours are benign painful tumours of the glomus body, a thermoregulatory shunt in the digits. Glomus tumours of the fingers and toes are associated with the monogenic disorder neurofibromatosis type 1 (NF1) and are recently recognised as part of the NF1 phenotype.
Methods and Results A multi-institutional experience with 15 individuals with NF1 and glomus tumours of the fingers or toes is reported. The majority of individuals presented with at least two of the symptoms in the classic triad of localised tenderness, severe paroxysmal pain, and sensitivity to cold. Appearance of the nail and finger or toe is often normal. Women are affected more often than men. Multifocal tumours are common. There is often a delay in diagnosis of many years and clinical suspicion is key to diagnosis, although magnetic resonance imaging may be useful in some scenarios. Surgical extirpation can be curative; however, local tumour recurrence and metachronous tumours are common. Three of our patients developed signs and symptoms of the complex regional pain syndrome.
Conclusions Glomus tumours in NF1 are more common than previously recognised and NF1 patients should be specifically queried about fingertip or toe pain.
- Neurofibromatosis type 1
- glomus tumour
- glomus body
- complex regional pain syndrome
- clinical genetics
Neurofibromatosis type 1 (NF1) is a common (∼1/3000 birth incidence) disorder of increased tumour predisposition that arises secondary to mutations in the RAS regulatory gene NF1.1 Glomus tumours are rare, benign tumours of the glomus body that cause severe paroxysmal pain secondary to changes in temperature or pressure. Glomus tumours of the fingers and toes associated with NF12 arise due to bi-allelic inactivation of the NF1 gene3 and are only recently recognised as part of the NF1 phenotype.4 5 They frequently appear as bluish subcutaneous nodules on the trunk and limbs and can be multifocal.6 7 The first association of NF1 and glomus tumours was published in 1938. The report described a single 13-year-old girl with features consistent with NF1 and multiple soft, dark blue nodules on her right neck, bilateral lower extremities and left heel. A lesion from the right lower leg had histologic features consistent with a glomus tumour.8 9
Glomus bodies are thermoregulatory shunts concentrated in the dermis of the fingertips and other peripheral sites subject to excessive cold, and should be distinguished from unrelated adrenal and extra-adrenal paragangliomas, also commonly called ‘glomus tumours’.10 Glomus tumours of the fingers consist of a convoluted arteriovenous anastomosis surrounded by a thick layer of modified smooth muscle cells and nerve elements (figure 1). Glomus tumours are thought to arise from the modified smooth muscle cells of glomus bodies, although they can occur in regions where glomus bodies do not normally occur.7 Typically, a glomus tumour of the finger presents with a triad of localised tenderness, severe paroxysmal pain (out of proportion to size), and sensitivity to cold. They have a benign clinical course.7 However, glomus tumours of the fingers are under-recognised. One large series of sporadic glomus tumours of the fingers found that an average of 2.5 physicians (range 0–7), including psychiatrists, were consulted before the correct diagnosis was made. The duration of symptoms averaged 10 years (range 1–40 years).11
Until recently, there were only eight cases of glomus tumours of the fingers and toes in individuals with NF1 in the English language literature (table 1).2 12–15 There were no examples of multifocal tumours in the largest retrospective review of 51 sporadic cases of glomus tumours of the fingers.11
In this report, we review the published literature and describe our multi-institutional experience (15 patients) with the presentation, diagnosis, imaging, management and complications of glomus tumours in the digits of individuals with NF1.
NIH NF1 and glomus tumour clinical experience
Table 2 and supplementary table S1 summarise clinical findings from our three groups. Patients NIH-1, NIH-2, and Leu-1 through Leu-7 were previously reported in brief tabular form.3 Leu-2 and Leu-3 were briefly summarised in De Smet et al 2002.2
As part of a natural history study of NF1, four individuals were ascertained with severe fingertip pain, a lesion on magnetic resonance imaging (MRI) that correlated with symptoms and at least one histologically proven glomus tumour (table 2 and supplementary table S1; figures 1 and 2 and supplementary figures S1 to S3). One woman (NIH-5) underwent resection of a pathologically proven glomus tumour in a single finger at another institution but was evaluated at follow-up at the National Institutes of Health (NIH). Repeat surgeries were required in two patients (NIH-1 and NIH-2; three surgeries each), both of whom harboured multifocal glomus tumours in multiple digits (supplementary table S1). Two individuals (NIH-1 and NIH-2) had local recurrence of tumour after resection or the development of new glomus tumours in other fingers and symptoms consistent with the complex regional pain syndrome (CRPS). At 2-year follow-up, one woman (NIH-3) had improvement in her glomus tumour pain after resection but had persistent neuropathic pain in her affected hand consistent with CRPS. Two women, both with single glomus tumours, were pain-free at 18 month (NIH-4) and 4 year (NIH-5) follow-up. There was no correlation between café-au-lait macule burden and number of neurofibromas with the development of glomus tumours.
Belgian NF1 and glomus tumour clinical experience
In the last 10 years, eight patients with NF1 and a glomus tumour of the fingertip were treated. All patients were seen by EL as part of an outpatient clinic for neurofibromatosis, and were subsequently seen by EL and LDS in a multidisciplinary clinic for congenital and genetic hand abnormalities at the human genetics outpatient clinic in Leuven, Belgium. All surgeries were performed by LDS and indication for surgery was based on the typical history and clinical examination compatible with glomus tumour of the fingertip. We did not perform systematic ultrasonography or MRI scanning before surgery. Fourteen pathologically proven glomus tumours were removed from 12 different fingers in eight patients (supplementary table S1). Only one finger needed a surgical re-intervention 13 months after the first surgery because of recurrence of a glomus tumour at exactly the same position as the first tumour (Leu-5). In this patient, the relapsed tumour was diagnosed clinically and was also visible by ultrasonography, but not by MRI scan of the finger. None of our eight patients with NF1 were diagnosed with the complex regional pain syndrome after surgical treatment. With the exception of Leu-5, all patients were cured after surgery. In the case of Leu-5, all symptoms were cured after resection of the relapsed tumour. Five of the 12 affected fingers did not show any visible abnormalities at diagnosis. Two fingers showed only a slight hyperaemic reddish discoloration of the skin, one finger showed a small localised swelling, one a small localised swelling with a hyperaemic reddish discoloration, and one a localised swelling of the base of the nailbed with nail dystrophy distal from the position of the localised tenderness (figure 3). One finger showed a split nail. The only child in the series (Leu-7; 11 years) had a notably swollen distal phalanx of the right fifth finger with a hyperaemic appearance. Radiographs of the fingers showed an impression of the tumour on the distal phalanx with erosion of the distal tuft of the phalanx (figure 4A). The phalanges of the right fifth finger and metacarpal appeared osteoporotic (figure 4B and 4C) and the affected fifth finger showed a mild shortening (4.5 mm) suggesting a longstanding severe problem. She was symptomatic for more than 2 years and protected the finger to prevent touch induced severe paroxysmal pain.
Presentation and diagnosis
Table 3 summarises and compares the demographic and physical characteristics of pathologically proven glomus tumours in NF1 in the published literature and in our combined experience. Recurrent tumours were excluded from the summary. We also summarise data from two large recent surveys of sporadic glomus tumours.11 16 In both sporadic and NF1 associated glomus tumours, the average age of diagnosis and years of symptoms before diagnosis are comparable. The tumour affects more women than men (table 3, ‘Combined NIH/Belgian/Hamburg experience’ and ‘Previously reported NF1-associated glomus tumors’: 4 males versus 16 females; p=0.006, binomial distribution). There is roughly equal distribution between the right and left hand or foot. Interestingly, the fourth (ring) and third (middle) fingers are the two most commonly affected digits in both NF1-associated and sporadic glomus tumours. In the combined series (table 3, ‘Combined NIH/Belgian/Hamburg experience’ and ‘Previously reported NF1-associated glomus tumors’) the frequencies of affected fingers are not equal (χ2, p<0.014); however only the fourth finger is significantly more frequently affected than expected (binomial distribution, p=0.04 after Bonferroni correction for multiple testing). No examples of multifocal tumours were reported in individuals without NF1.
In the eight published descriptions (table 1) of glomus tumours of the fingers associated with NF1 (hereafter, unless noted, ‘glomus tumours’), all reported at least one element of the classic glomus tumour triad: localised tenderness, severe paroxysmal pain, and sensitivity to cold. In our experience, patients, if asked, will have at least two or more features of the triad. Paradoxically, patients may have lived with the pain for so long (in our series, up to 40 years) that they need to be specifically asked about these symptoms. The tenderness is typically continuous and is exacerbated by exposure to cold (or less commonly, warm) temperatures, which can be as minor as holding a cold drink or reaching into a refrigerator. Patients may wear multiple pairs of gloves or mittens in the cooler months to insulate their fingers. The severe paroxysms of pain can be short lived (<1 min), but are debilitating and frightening, and patients can require considerable time (hours) to recover from them. Paroxysms can be triggered by minor everyday vibration (riding a bicycle) and use (typing on a keyboard). Patients are protective of their fingers and may be reluctant to shake hands. The pain is typically non-radiating and localised to a particular (ulnar/radial/central) aspect of a finger. If the pain radiates, it is often limited in extent to the elbow. The distinguishing feature of glomus tumour pain is its severity and location; the quality of the pain is akin to slamming a heavy drawer on a finger. Out of frustration for a lack of diagnosis (and thus effective treatment), we have had patients request amputation of the affected fingers.
On physical examination, the nail and pulp of the affected fingertip is often entirely normal. Patients may guard the affected finger since even a gentle examination is likely to elicit paroxysms of pain. The location of glomus tumours may either be subungual or in the pulp of the fingertip. In a large series of sporadic glomus tumours, a blue discoloration of the nail was observed in 43% of subungual tumours and 10% of tumours of the pulp.11 In the same series, a nail deformation was noted in 47% of subungual tumours. We observed no examples of bluish discoloration of the nails and three examples (NIH: 1; Belgium: 2) of nail abnormalities in our patients. Tumour nodules are infrequently palpable on examination. However, the location of the tumour can be mapped with some precision by applying gentle pressure from the head of a sterile straight pin (Love's test).17 Love's test has a sensitivity of 100% and 78% accuracy.18 Other useful physical examination techniques include Hildreth's test (application of a tourniquet to the base of an affected finger and repeating Love's test; pain should be abolished) and the cold sensitivity test (placing affected hand in cold water). Hildreth's test has a sensitivity of 71.4%, a specificity of 100%, and an accuracy of 78%, and the cold sensitivity test has a sensitivity, specificity and accuracy each of 100%.18
There is an extensive literature on the imaging of glomus tumours.19–21 In one large series, plain radiography revealed a bony defect in the distal phalanx in 36% of patients,11 and may be a useful adjunct given sufficient clinical suspicion. Four of our patients (NIH-1, NIH-2, NIH-3, Leu-7) had plain radiographs of the hands. Of the eight fingers in these four patients harbouring glomus tumours at the time of radiography, only two (NIH-2, right F1; Leu-7, left F5) had lytic lesions consistent with glomus tumours. We have had little experience with fingertip ultrasound; it has some theoretical advantages and is reported to be useful in locating glomus tumours as small as 3 mm in size, especially if located in the pulp of the fingertip, but is operator dependent.20 Other modalities (thermography, scintigraphy, arteriography) are no longer indicated.
High resolution MRI is capable of detecting normal glomus bodies with a T2 weighted sequence after gadolinium injection.19 Glomus tumours are proliferations of glomus cells with a relative paucity of vascular lumen. Their signal is thus similar to the surrounding tissue (especially those in the highly vascular nailbed) and can be difficult to detect on T1 sequences.20 Post-contrast imaging with gadolinium may be helpful. Normal glomus bodies and glomus tumours exhibit hyperintensity on T2 sequences relative to surrounding reticular dermis. Glomus tumours are delimited by a pseudocapsule that forms as a reaction to the surrounding connective tissue. The capsule is best seen on T2 weighted images or on three dimensional gradient echo images.20 The NIH imaging protocol includes coronal STIR (short TI inversion recovery), axial T1 spin echo, axial fast spin echo, T2 weighted and three dimensional gradient echo post-contrast imaging in axial and coronal planes with a dedicated receive-only wrist coil.
The usefulness of MRI in the diagnosis of glomus tumours is controversial.21 In one study of 42 individuals with glomus tumours symptoms undergoing hand surgery, 40 had histologically proven glomus tumours. All had MRI imaging preoperatively. MRI had 90% sensitivity, 50% specificity, a positive predictive value of 97% and a negative predictive value of 20%. The low negative predictive value arose from patients with smaller (2–3 mm) tumours with a lack of clear delineation.21 The authors note that a glomus tumour of the hand is a clinical diagnosis and that compelling evidence to use preoperative MRI imaging is lacking.
The NIH experience with the use of MRI to diagnose glomus tumours in NF1 is summarised in supplementary table S1. In classic, uncomplicated presentations of glomus tumour symptoms (eg, NIH-4 and the initial surgery for NIH-1 and NIH-2), we observed a perfect correlation among clinical impression, MRI imaging, and pathology. In these cases, the clinical examination was likely sufficient to diagnose and even plan surgical resection. However, in more complicated scenarios like patients with CRPS (eg, NIH-3 and second and third surgeries of NIH-1 and NIH-2), the predictive power of the clinical examination declines, and the usefulness of MRI increases. In these cases, we found that MRI correctly predicted the pathological diagnosis of glomus tumour with greater sensitivity than the clinical examination. In patients with CRPS many fingers (indeed the entire hand or limb) hurt, making it difficult for the patient and examiner to distinguish ‘glomus tumour’ pain from ‘CRPS’ pain. In CRPS, where avoidance of additional ‘pain generators’ (like surgical exploration) is paramount, MRI can be useful to avoid unnecessary procedures.
In our experience, glomus tumours on MRI tend to be nodular and exhibit hyperintensity on T2 weighted imaging. A capsule or pseudocapsule was not typically appreciated on T2 imaging. Lesions tend to be isointense and poorly visualised on non-enhanced T1 imaging. Glomus tumours exhibited a variable degree of enhancement on T1 post-contrast imaging, with most lesions showing mild to moderate enhancement, but some lesions remaining isointense to the surrounding, enhancing nailbed. Potential lesions that exhibit focal hyperintensity only on a STIR or T2 sequence (eg, NIH-1, left F3, surgery number 2) but without correlative findings on other sequences are more likely to be false positives. In addition, it is important to remember that multiple glomus tumours may be present in one finger (supplementary table S1: NIH-1, left F4, surgery 1, procedure 1; Leu-3, left F4, surgery 2, procedure 1). Evaluation of the postoperative nailbed may be hampered by artefact, and requires careful comparison to prior imaging exams.
For symptomatic glomus tumours, surgical extirpation is the only effective treatment and can be curative. Two surgical approaches are used, depending on the location of the tumour: direct transungual excision (for sub-ungual tumours) or a lateral sub-periosteal approach (for both subungual tumours and tumours of the pulp).16 The lateral approach may reduce the risk of nail deformity but may afford a more narrow view of the tumour bed, and thus a higher risk of incomplete excision and thus recurrence.22 Either can be performed under local anaesthesia with or without sedation. Although typically small (<5 mm), glomus tumours of the fingers have a pseudocapsule, making excision relatively easy. The tumours are frequently apposed to the bone, necessitating bony curettage to reduce the chance of recurrence. Nail deformities are seen in a small minority of patients especially after use of the transungual approach.16 Pathologically, glomus tumours are almost always benign but can be misdiagnosed as haemangiomas or venous malformations.7
In the Belgian and Hamburg experience, all patients brought to surgery had pathologically proven glomus tumours of the fingers or toes. In the NIH experience, three patients (NIH-1, NIH-2, and NIH-3; supplementary table S1) had, concurrent with successful tumour resection in other digits, fingertip explorations in which no glomus tumour was found. However, all three patients had multiple painful fingertips due to CRPS, lack of imaging of the painful but non-glomus tumour finger, and/or atypical MRI abnormalities (see ‘Recurrence of pain’, below).
Recurrence of tumour
In the majority of published cases of glomus tumours in NF1 (table 1), no comment is made on the recurrence of tumour following surgery. In six large published series (n=12 to 51 patients) of sporadic glomus tumours of the fingers, the recurrence rate, when noted, ranged from 0–33.3%.16 18 23 Early recurrence may occur within weeks to months of surgery, and presumably reflects inadequate excision. Later recurrence (years) is probably the result of the development of a new tumour.16 Three of our patients had repeat surgery on five fingers with histologic evidence of tumour recurrence within 13 months of the initial procedure (supplementary table S1: NIH-1: left F4; NIH-2: left F2 and left F4 and right F1; Leu-5: left F3). In four of the five fingers, bony involvement or erosion was noted on initial preoperative MRI imaging or at time of the first surgery. Recurrence occurred despite bony curettage at the time of initial operation.
Recurrence of pain
In our experience, recrudescence of glomus tumour symptoms may be due to recurrence of tumour (above), growth of a new tumour in a previously unaffected finger (eg, NIH-2: right F4) or the development of CRPS, which we diagnosed in three patients (NIH-1, NIH-2, NIH-3). Of these three causes, CRPS is the most difficult to treat.
CRPS is a poorly understood chronic pain syndrome affecting one or more limbs. It has a variable course and a heterogeneous pathogenesis.24 CRPS is classified into type 1 (no known inciting injury; formerly known as reflex sympathetic dystrophy, RSD) and type 2 (known or suspected inciting injury, formerly known as causalgia). We diagnosed three patients (NIH-1, NIH-2, and NIH-3) with CRPS type 2 before excision, suggesting that the syndrome arose secondary to the glomus tumours, and not surgery. Patients presented with allodynia, hyperesthesia, dysesthesia, oedema, sudomotor and vasomotor changes primarily affecting one limb, a clinical picture not consistent with neurofibromatous neuropathy, an uncommon but well recognised neurologic complication in NF1.25–27 To our knowledge, these three patients are the first reported with CRPS in NF1 and the first with CRPS secondary to glomus tumours of the fingers. One case report described reflex sympathetic dystrophy in a female with a glomus tumour of the leg.28 Individuals with NF1 may be at an increased risk of developing chronic pain syndromes such as CRPS since neurofibromin (the protein product of NF1) plays a key role in the excitability regulation of nociceptive sensory neurons.29
Avoidance or elimination of chronic inciting ‘pain generators’ in CRPS is critical. Specifically in the case of glomus tumours, this is difficult to achieve in practice since the tumours are under-recognised and patients may have symptoms for years or decades before proper diagnosis. Thus, we advocate prompt removal of symptomatic glomus tumours. Similarly, in patients with a previous resection of glomus tumours, MRI of recurrent or newly painful fingers may help to distinguish between true tumour recurrence and the development of CRPS.
Lastly, prevention is the best strategy in managing CRPS. Although the prevalence of glomus tumours in NF1 is not known, it is likely to be higher than previously supposed. To screen for glomus tumours of the fingers, we suggest simply asking all individuals with NF1 about chronic, cold sensitive fingertip pain and mechanical allodynia.
Summary and recommendations
From our experience, we advocate the following:
Glomus tumours of the fingers (and, more rarely, toes) are associated with NF1 and are part of the tumour spectrum of the disorder. Recently published genetic, functional and clinical evidence shows that glomus tumours in NF1 arise secondary to bi-allelic inactivation of NF1.3
Glomus tumours in NF1 are more common than previously suspected. Although some of our patients were evaluated following specific referral, many were not, and were diagnosed because we specifically asked about fingertip pain. From our experience, we estimate that glomus tumours of the fingers may affect up to 5% of the adult NF1 population. Their prevalence in toes is unknown but much lower than that of fingers.
Glomus tumours can cause considerable morbidity and screening efforts should be established. In our experience, glomus tumours profoundly affect quality of life (disability, CRPS) and, in the published literature, can result in unnecessary surgery (see patient 1 from Sawada et al 199513). A simple question (‘Do the tips of your fingers ever hurt, especially when cold or bumped?’) should be asked as part of the routine care of adults with NF1.
The natural history of sporadic and NF1-associated glomus tumours is similar (presentation, recurrence of tumour), with some important differences. Specifically, individuals with NF1 can present with multiple tumours in single or multiple fingers. There may be theoretical reasons why individuals with NF1 are at an increased risk to develop CRPS.
Glomus tumours are a clinical diagnosis. In our experience, in uncomplicated cases, MRI correlated highly with physical exam findings. In certain scenarios (recurrence of tumour, CRPS, surgical planning), MRI may be useful.
The establishment of glomus tumours of the fingers and toes as part of the tumour spectrum of NF1 over 100 years after von Recklinghausen's description exemplifies the need for open mindedness and creativity, even when evaluating familiar phenotypes. It is our hope that greater awareness of glomus tumours in NF1 leads to improved care.
Funding The work was supported in part by the Division of Intramural Research of the National Human Genome Research Institute (DRS), research grants from the Fonds voor Wetenschappelijk Onderzoek Vlaanderen (G.0096.02, EL), the Interuniversity Attraction Poles granted by the Federal Office for Scientific, Technical and Cultural Affairs, Belgium (2007-2011; P5/25, EL), and by a Concerted Action Grant from the K.U. Leuven (EL, RS).
The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organisations imply endorsement by the US Government.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Institutional Review Board of the National Human Genome Research Institute.
Provenance and peer review Not commissioned; externally peer reviewed.