Vertebra Plana in Children May Result from Etiologies Other Than Eosinophilic Granuloma

Abstract

Background

Vertebra plana in children is a diagnostic dilemma for orthopaedic surgeons. This radiographic finding sometimes has been said to be pathognomonic for eosinophilic granuloma (Langerhans cell histiocytosis); however, vertebra plana may also be caused by a range of other conditions. We sought to determine whether vertebra plana can be associated with malignancies other than eosinophilic granuloma.

Questions/purposes

(1) To report the underlying diagnoses for children with vertebra plana and determine how frequently these patients were found to have eosinophilic granuloma as opposed to an underlying malignant process, (2) to evaluate the occurrence of nondiagnostic results on biopsy, and (3) to determine whether the presenting characteristics of spinal lesions were associated with the ultimate clinical diagnosis.

Methods

As part of a retrospective review, our institutional electronic medical record was searched for all patients younger than 18 years between 1976 and 2017 whose clinical record included the term vertebra plana. Patients with trauma were excluded. Twenty-seven patients met the inclusion criteria (mean [range] age 9 years [0 to 18]; 12 girls). To address our first research purpose about the underlying diagnoses of patients with vertebra plana, we reviewed the final clinical diagnosis. To address our second research purpose about the utility of biopsy, we reviewed which patients underwent a biopsy and whether it had been diagnostic. To address our third research purpose about the radiographic criteria, we classified the radiographs and compared this to the clinical diagnosis. Vertebral collapse was described as less than 50% collapse, 50% to 100% collapse, symmetrical, and asymmetrical. The location of each lesion was noted.

Results

Twelve of 27 patients had a diagnosis of eosinophilic granuloma. Six of 27 had other neoplastic etiologies, including acute lymphoblastic leukemia, primary germ cell tumor, giant cell tumor, rhabdomyosarcoma and teratoma. Seventeen of 27 patients underwent biopsy to confirm the diagnosis; six biopsies were consistent with eosinophilic granuloma, six for other etiologies, and five were nondiagnostic. With the limited patient numbers available, there was no difference in the frequency of less than 50% loss of height or 50% to 100%, or symmetric and asymmetric loss of height, and location of the lesion among patients with eosinophilic granuloma and those with other diagnoses.

Conclusions

Eosinophilic granuloma or Langerhans cell histiocytosis is a common cause of vertebra plana, but other causes must be considered in children presenting with this radiographic finding. Six of 27 of patients presenting to our center with vertebra plana had an underlying neoplasm other than eosinophilic granuloma. With the limited patient numbers available, pain, spinal location, and the degree and symmetry of collapse were not associated with a diagnosis of eosinophilic granuloma in this series. Thus, patients presenting with vertebral plana and back pain need a comprehensive work-up and potentially tissue biopsy to determine diagnosis and appropriate treatment.

Level of Evidence

Level IV, diagnostic study.

Introduction

Atraumatic vertebral collapse in children is rare, but when it occurs, it is a diagnostic challenge for surgeons. Substantial, uniform loss of height is described as vertebra plana, a radiographic finding frequently associated with eosinophilic granuloma or Langerhans cell histiocytosis [3]. A new diagnosis of vertebra plana without a history of high-energy trauma represents a diagnostic dilemma because it is unclear how much testing should be performed to exclude the possibility of underlying metabolic bone disease or malignancy [4, 7]. Patients may present with back pain or even with an incidental finding seen on chest radiographs or other images. Further work-up may include a full skeletal survey, MRI, CT, open or CT-guided spine biopsy, bone marrow biopsy, bone scan, or other invasive testing.

Previous reports have described cohorts of patients with vertebra plana and confirmed tissue diagnoses of eosinophilic granuloma [2, 5, 6, 13, 18]. Eosinophilic granuloma is defined as the pathologic accumulation in bone of large histiocytes, a phagocytic cell of the connective tissue. Eosinophilic granulomas occur most commonly in patients younger than 20 years, with peak predominance between the ages of 5 and 10 years [2, 11, 18, 29]. Case series have reported that patients present with vertebra plana for reasons other than eosinophilic granuloma [15, 17, 21, 23]. However, there is little information to guide physicians at the time of diagnosis regarding the chances that a pediatric patient with vertebra plana will have underlying eosinophilic granuloma and whether biopsy is helpful to promptly arrive at a diagnosis.

We therefore sought (1) to report the underlying diagnoses for children presenting with vertebra plana and determine how frequently these patients were found to have eosinophilic granuloma as opposed to an underlying malignant process, (2) to evaluate the occurrence of nondiagnostic results on biopsy, and (3) to determine whether the presenting characteristics of spinal lesions were associated with the ultimate clinical diagnosis.

Patients and Methods

Approval was obtained from our institutional review board to retrospectively study medical records and images (IRB #16-001098). Institutional software was used to search for all patients 18 years or younger whose clinic or radiography reports included the text “vertebra plana” or “vertebra collapse” between 1976 and 2017 at a tertiary referral center. This provided a minimum of 2-year follow-up for patients. Radiographs at our institution before 1976 have been discarded. Thus, we limited the search to patients presenting after 1975. Patients with trauma were excluded from the study. STROBE criteria were completed. From the search, 156 patients were identified, of whom 129 had a history of trauma and were excluded, leaving 27 patients to include in the current study. One author (FB) reviewed the records with senior authors available for consultation and further radiographic review (ANL, MTH, TAM). The review included age at presentation, symptoms at presentation, and gender/sex. Final patient diagnosis was determined through retrospective review of existing medical results and conclusions of the patients’ treating physicians based on consideration of testing and pathology results. To address our first research purpose about the underlying diagnoses of patients with vertebra plana, we evaluated the final clinical diagnosis based on medical records, work-up, and pathology results. To address our second research purpose about the utility of biopsy, we determined which patients underwent a biopsy and whether it had been diagnostic. To explore our third research purpose about the radiographic criteria, we classified the radiographs and evaluated the final clinical diagnosis. For the third research purpose, radiographs for all patients were reviewed, and the degree of vertebral collapse was classified according to the method of Garg et al. [13]. Type 2 lesions were defined as greater than 50% loss of height and Type 1 lesions were defined as 50% or less loss of height. A Type A lesion reflected symmetrical loss of height, and Type B represented asymmetrical loss of height.

Due to the retrospective nature of this study and variable patient presentation, a variety of diagnostic studies were performed to determine the final diagnosis and patient follow-up was determined by the treating physician. Patients underwent treatment according to the treating physician’s diagnosis. During the follow-up period, no patients presented with a novel diagnosis who would have presented a different explanation for vertebra plana.

We used a chi-square analysis for categorical variables (diagnosis and radiographic classification), and a two-tailed t-test for continuous outcome variables, and p values less than 0.05 were considered significant. Given the small sample size, we did not perform a regression analysis.

Results

Our first study aim was to determine the underlying diagnoses for pediatric patients presenting with vertebra plana. Of the 27 patients who met the inclusion criteria, six had a diagnosis of other neoplasms and 12 had a diagnosis of eosinophilic granuloma (Table 1). Six of 12 patients had biopsy-proven eosinophilic granuloma, and the remainder had presumptive diagnoses based on clinical findings alone (three patients) or on biopsy with indeterminate findings (three patients).

Table 1.

Demographics and lesion characteristics

Patient number	Sex	Age (years)	Level	Classa	Presentation	Diagnosis
1	M	6	L4	2A	Back pain	Eosinophilic granuloma
2	F	11	T11	2B	Back pain	Eosinophilic granuloma
3	F	13	T6, T8	2A	Incidental finding	Eosinophilic granuloma
4	M	2	L1	1A	Back pain	Primary germ cell tumor
5	M	12	T7	2A	Back pain	Idiopathic
6	M	5	T4	1A	Orbital swelling	Eosinophilic granuloma
7	M	2	L4	1A	Back pain	Eosinophilic granuloma
8	M	12	T6	1A	Lower-extremity deformity	Spondylometaphyseal dysplasia
9	F	8	T7	1B	Back pain	CRMO
10	F	15	T2	1A	Back pain, fevers	Eosinophilic granuloma
11	M	10	L2	2A	Back pain	Eosinophilic granuloma
12	M	9	T12	2A	Back pain	Eosinophilic granuloma
13	M	12	T10	2A	Back pain	Eosinophilic granuloma
14	M	6	C4	1A	Neck pain	Eosinophilic granuloma
15	F	11	T7	1B	Back pain	CRMO
16	F	3	L2	2A	Sacral mass	Teratoma
17	F	10	T5	1A	Back pain	Eosinophilic granuloma
18	M	0	L2	2A	Ultrasound	Osteogenesis imperfecta
19	F	9	T4, T5, T6, T7	2A	Back pain	Idiopathic
20	F	10	T2	1A	Back pain	Idiopathic
21	F	18	L5	2B	Back pain	Giant cell tumor
22	F	18	T3	2B	Back pain	CRMO
23	M	8	T10, T11, T12	1B	Back pain	ALL
24	F	10	T8-L4	1B	Back pain	ALL
25	M	11	T6	2B	Back pain	Eosinophilic granuloma
26	M	0	T9-L2	2A	Incidental finding	Osteogenesis imperfecta
27	M	16	L3	1A	Back pain	Primary rhabdomyosarcoma

^aType 1 is < 50% collapse, Type 2 50-100%, A is symmetrical collapse, B is asymmetrical; CRMO = chronic recurrent multifocal osteomyelitis; ALL = acute lymphoblastic leukemia.

Three patients had a known underlying diagnosis at the time of presentation, including an existing diagnosis of Langerhans cell histiocytosis with recurrent orbital swelling, one patient had chronic, recurrent osteomyelitis at other skeletal sites, and one patient had osteogenesis imperfecta demonstrated by multiple other characteristic skeletal fractures. Of the remaining 24 patients with no known underlying diagnosis at presentation, six had a new diagnosis of other neoplasms including acute lymphoblastic leukemia (two patients), primary germ cell tumor (one patient), giant cell tumor (one patient), rhabdomyosarcoma (one patient), and teratoma (one patient).

We further sought to evaluate the utility of biopsy to determine a diagnosis. A variety of diagnostic testing performed among the 27 patients over the 41-year study period (Table 2). Seventeen of 27 patients underwent biopsy to confirm the final diagnosis; 12 biopsies confirmed the diagnosis of chronic recurrent multifocal osteomyelitis (CRMO) (2), eosinophilic granuloma (6), primary germ cell tumor (1), teratoma (1), giant cell tumor (1), rhabdomyosarcoma (1). Of the five nondiagnostic biopsies, two demonstrated the presence of inflammation but were negative for Langerhans histologic staining (CD1A, Langerin) and these patients received a diagnosis of chronic recurrent multifocal osteomyelitis. Two biopsies provided inadequate samples, but these biopsies were not repeated. Both patients received a presumptive diagnosis of eosinophilic granuloma. The final nondiagnostic biopsy had no findings of abnormality, and this patient received a diagnosis of idiopathic vertebral collapse. The two patients with an ultimate diagnosis of acute lymphoblastic leukemia initially presented with insidious back pain. Radiographs revealed vertebra plana. Laboratory findings showed pancytopenia and elevated lactate dehydrogenase levels and inflammatory markers. Both patients underwent bone marrow biopsy and were given a diagnosis of acute lymphoblastic leukemia. Thus, including the bone marrow biopsies, 19 of 27 patients underwent biopsy, of which 14 produced a tissue diagnosis.

Table 2.

Work-up of the patients included in this study

Patient	Imaging	Biopsy performed	Biopsy diagnosis	Diagnosis
1	Plain film	None		Eosinophilic granuloma (presumptive)
2	Plain film	None		Eosinophilic granuloma (presumptive)
3	Plain film	Open biopsy	Eosinophilic granuloma	Eosinophilic granuloma
4	Plain film, CT, abdomen/pelvis MRI	Bone marrow biopsy, CT-guided spine	Undifferentiated malignant neoplasm	Primary germ cell tumor
5	Plain film, CT, MRI, skeletal survey	CT-guided biopsy	Nondiagnostic	Idiopathic
6	CT, skeletal survey	Orbital mass biopsy	Eosinophilic granuloma	Eosinophilic granuloma
7	MRI spine, skeletal scintigraphy	CT-guided biopsy	Positive for S-100 and negative for CD68 and LCA	Eosinophilic granuloma
8	Plain film	None	None	Spondylometaphyseal dysplasia
9	Plain film, MRI, skeletal survey	CT-guided and open biopsy	CD1a and Langerhans-negative	CRMO
10	Plain film, MRI, skeletal scintigraphy	CT-guided biopsy	Nondiagnostic	Eosinophilic granuloma (presumptive)
11	Plain film, MRI, skeletal survey, skeletal scintigraphy	CT-guided biopsy	Nondiagnostic	Eosinophilic granuloma (presumptive)
12	MRI	(A) CT-guided biopsy (B) Open biopsy	(A) Nondiagnostic (B) Positive for CD1a and S100; LCA and KP-1-positive in macrophages	Eosinophilic granuloma
13	CT, MRI	(A) CT-guided biopsy (B) Open biopsy	(A) Nondiagnostic (B) Negative LCA, Langerhans staining	Eosinophilic granuloma (presumptive)
14	Plain film, CT, MRI, skeletal survey	CT-guided biopsy	Positive for Langerhans, CD1A, S100, and CD68	Eosinophilic granuloma
15	Plain film, CT, MRI, skeletal survey, skeletal scintigraphy	CT-guided biopsy	Acute and chronic inflammation with reactive or remodeling changes; negative Langerhans staining result	CRMO
16	Ultrasound, MRI, skeletal scintigraphy	Open biopsy	Teratoma	Sacral teratoma
17	Plain film, MRI, skeletal survey, skeletal scintigraphy	None		Eosinophilic granuloma (presumptive)
18	Ultrasound, plain film, skeletal survey	None		Osteogenesis imperfecta
19	Plain film, MRI, DEXA, skeletal scintigraphy	None		Idiopathic
20	Plain film, MRI, CT	None		Idiopathic
21	Plain film, MRI, CT	(A) CT-guided biopsy (B) Excisional biopsy	(A) Nondiagnostic (B) Bland giant cell-rich lesion	Giant cell tumor of the bone
22	Plain film	CT-guided biopsy	Nondiagnostic	CRMO
23	Plain film, CT, MRI	Bone marrow biopsy		ALL
24	Plain film, CT, MRI	Bone marrow biopsy		ALL
25	Plain film	Open biopsy	Langerhans, CD1A, S100, and CD68-positive	Eosinophilic granuloma
26	Plain film, skeletal survey	None		Osteogenesis imperfecta
27	Plain film, CT, MRI	Open biopsy	Cytologic atypia, spindle cells	Rhabdomyosarcoma

LCA = leukocyte common antigen; CRMO = chronic recurrent multifocal osteomyelitis; DEXA = dual-energy x-ray absorptiometry; ALL = acute lymphoblastic leukemia.

We also sought to evaluate whether the presenting characteristics of spinal lesions were associated with a diagnosis of eosinophilic granuloma (Fig. 1). Radiographic appearance did not correlate with the diagnosis (Table 3). Fourteen patients (52%) had greater than 50% loss of height. In our limited sample size, there was no difference in the frequency of greater than 50% loss of height and 50% or less loss of height between patients with eosinophilic granuloma (n = 7) and those with other diagnoses (n = 7; p = 0.67, Table 4). Additionally, in our limited sample size, there was no difference in the frequency of symmetrical loss of height and asymmetrical loss of height between patients with eosinophilic granuloma and those with other diagnoses (p = 0.10). Back pain was the most common initial complaint in patients with a diagnosis of eosinophilic granuloma (n = 10) and also in those with other diagnoses (n = 11; p = 0.53). Thoracic vertebrae were commonly involved in patients with eosinophilic granuloma (n = 8) and also in those with other diagnoses (n = 9; p = 0.49). Twenty-two of 27 patients had a single involved vertebra and five had multiple vertebrae.

Fig. 1.

Representative examples of vertebra plana are shown here. (A) Posteroanterior and (B) lateral thoracic spine radiographs of an 8-year-old girl presenting with atraumatic back pain showing T7 vertebra plana. Bone scan revealed additional lesions at L4, S1, bilateral femurs, and right tibia, and left humerus. Biopsy of the tibia was consistent with chronic regional multifocal osteomyelitis. She was initially treated with scheduled naproxen, and by 6-year follow-up, she was asymptomatic. (C) Posteroanterior and (D) lateral thoracic spine radiographs of an 11-year-old boy who presented with back pain and vertebra plana. Open biopsy of T6 revealed eosinophilic granuloma.

Table 3.

Radiographic findings, percentage of collapse

Type	Type 1 (< 50% collapse)	Type 2 (> 50% collapse)
A	Primary germ cell tumor Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Spondylometaphyseal dysplasia idiopathic Eosinophilic granuloma	Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Eosinophilic granuloma Osteogenesis imperfecta Osteogenesis imperfecta Idiopathic Idiopathic Sacrococcygeal teratoma
B	CRMO CRMO ALL ALL	Eosinophilic granuloma Eosinophilic granuloma CRMO Giant cell tumor

CRMO = chronic recurrent multifocal osteomyelitis; ALL = acute lymphoblastic leukemia.

Table 4.

Radiographic findings, asymmetric collapse

Diagnosis	50% or less loss of height (Type 1)		> 50% loss of height (Type 2)
	Symmetric (Type A)	Asymmetric (Type B)	Symmetric (Type A)	Asymmetric (Type B)
Eosinophilic granuloma	5	0	5	2
Solid tumor	2	0	1	0
CRMO	0	2	0	1
ALL	0	2	0	0
Other	2	0	4	1

CRMO = chronic recurrent multifocal osteomyelitis; ALL = acute lymphoblastic leukemia.

Discussion

The radiographic definition of vertebra plana is an “extreme collapse of a vertebral body with sparing of the posterior elements and normal to slightly widened adjacent disk spaces.” Calvé [7] first described this finding in 1925 while treating patients whom he believed had extrapulmonary tuberculosis. Over time, however, the finding of vertebra plana has been linked to the osseous manifestation of Langerhans cell histiocytosis—eosinophilic granuloma—and grew to be considered pathognomonic for the condition in the pediatric population. Although vertebra plana is frequently due to an underlying diagnosis of eosinophilic granuloma, we noted several patients in our practice who presented with vertebra plana associated with other causes. Consequently, we wished to study more formally the frequency of this event, the role of a diagnostic biopsy, and whether the radiographic and clinical findings that might be associated with the patient’s diagnosis. We found that vertebra plana can be due to neoplastic causes other than eosinophilic granuloma. This is important because the treatment recommendations are different for each type of neoplasm. Delay in diagnosis of ALL or a malignant tumor can result in treatment delay and potentially increased mortality.

The limitations of this study include that it is a retrospective chart review, which can result in selection bias, transfer bias, and assessment bias. This represents a cohort of pediatric patients presenting with vertebra plana to a tertiary referral center over a 41-year period. Further, due to small patient numbers, there may be associations between diagnosis and other patient characteristics that were not detected. Again, due to the small patient numbers, our study may be underpowered to detect whether specific radiographic criteria are associated with the underlying diagnosis because the study is likely underpowered for these parameters. Technology including MRI, positron emission tomography scans, and histologic diagnosis strategies have changed during the study period. Further, six of 12 patients with vertebra plana were diagnosed based on clinical data in the absence of a tissue diagnosis. Although it is possible that these patients may have had alternative diseases, none went on to develop systemic malignancy or multifocal disease during the follow-up period.

Additionally, it is important to note that our search relied on the text used in the radiology and clinical records. As a result, our patient selection was subject to the language biases of providers caring for these patients. For instance, as Scheuermann kyphosis is described by multiple adjacent wedged vertebra and not vertebra plana, no Scheuermann disease was noted in our list of possible patients to include in the study. We chose to include all patients who had a radiographic finding of vertebra plana without preceding trauma. This included patients who may have had other physical exam or clinical findings indicative of their diagnosis and was not limited to patients with an unknown diagnosis at the time of radiograph. We do not believe that this detracts from the conclusion that the finding of vertebra plana is not limited to the diagnosis of eosinophilic granuloma. Finally, our study was performed at a large tertiary referral center and that may have been affected our presenting patient population. A multicenter study including institutions with different referral patterns may provide a better cross-section of patients presenting with atraumatic vertebra plana. As our study was retrospective, there were no patient-reported outcomes to assess the degree of back pain each patient experienced. Further studies could pursue more in-depth assessment of patient symptoms or potentially MRI findings associated with other diagnoses. In our experience, MRI is helpful to rule out an associated soft tissue mass, which may indicate a malignant process and guide plans for biopsy.

We found that a wide range of diagnoses other than eosinophilic granuloma were responsible for vertebra plana. Our paper is different from previous reports in that it examines a cross-section of patients presenting with vertebra plana, rather than patients with a known histologic diagnosis [2, 5, 6, 11]. Our approach is clinically useful because it parallels the physician’s experience of having a patient present for care with an unknown diagnosis rather than a collection of patients with an already established diagnosis. Twelve of 27 patients in our series had a diagnosis of eosinophilic granuloma, including one patient with a known prior diagnosis because of orbital swelling. Our series showed a higher-than-expected incidence of non-eosinophilic granuloma diagnoses, and, most concerning, a high incidence of malignancy. Although eosinophilic granuloma represents most reported patients in previous studies, several case reports have shown that various disease processes present as vertebra plana, including aneurysmal bone cyst, Ewing sarcoma, tuberculosis, lymphoma, leukemia, acute osteomyelitis, and chronic recurrent multifocal osteomyelitis [8-10, 15, 17, 19, 21-25, 28]. Interestingly, our institution treated two patients with chronic recurrent multifocal osteomyelitis initially presenting with vertebral collapse. Chronic recurrent multifocal osteomyelitis is an autoinflammatory disorder with biopsy recommended for diagnosis, although primary presentation in the spine is atypical [1, 12, 14, 28]. Our study provides information that vertebra plana may be an initial presentation for conditions other than eosinophilic granuloma.

Given the high rate of malignancy, we argue for performing CT-guided biopsy if CT, MRI, labwork, and physical exam have not provided a diagnosis and when an accessible target has been identified [27]. Despite biopsy, however, only 12 of 17 biopsies of bone lesions performed in our series were diagnostic, pointing to the need for a careful multisystem evaluation. Further, open biopsy in the spine may lead to substantial morbidity and cannot be recommended as first-line work-up procedure.

The most common presenting symptom for vertebra plana associated with eosinophilic granuloma in our study was pain and limited or restricted ROM, a finding that echoes the work of others [2, 5, 6, 11, 16, 20, 26]. But it is important to note that with the numbers available in our study, the initial complaint of back pain was no more common in children with eosinophilic granuloma than it was among children with other diagnoses. More generally, presenting symptoms, demographics, lesion location, and radiographic findings were not different in our patients with eosinophilic granuloma than they were in patients with a malignancy, suggesting that all patients with vertebra plana warrant a thorough work-up examination. Unfortunately, in our limited series, radiographic findings also did not distinguish between patients with eosinophilic granuloma and those with a malignancy, suggesting that all patients with vertebra plana warrant a thorough work-up.

In summary, we found that vertebra plana in a child is not pathognomonic for eosinophilic granuloma, and that in fact fewer than half of the patients in our series who presented with that finding had that diagnosis. Thus, pediatric patients presenting with vertebra plana require a comprehensive evaluation. If a cause cannot readily be determined by laboratory findings, history-taking, and physical examination, then biopsy should strongly be considered to exclude the presence of a malignancy. It is important to develop a thoughtful diagnostic approach, recognizing that all diagnostic tests have strengths and weaknesses, and that vertebra plana can represent neoplasm other than eosinophilic granuloma.