Erlenmeyer Flask Deformity: Causes, Types and How to Measure It

Erlenmeyer flask deformity is a failure of metaphyseal modelling. The di-metaphysis of a long bone, classically the distal femur, is not narrowed down to a normal diaphyseal waist during growth, a process called undertubulation. The result is a broad flared metaphysis, straight or outwardly bowed di-metaphyseal borders instead of the normal concave curve, and a thin cortex, so the bone end resembles a conical laboratory flask.

The single most useful thing to know about it is that it is a shape, not a diagnosis. It has been reported in at least 20 distinct disorders, and the shape alone does not tell you which. What narrows the differential is the trabecular bone around the deformity, and that is exactly what the Faden classification below is built on.

Why is it called so?

The deformity is named for the Erlenmeyer flask, the conical laboratory flask with a wide flat base, straight uncurved sides and a narrow cylindrical neck. It was introduced in 1861 by the German chemist Richard August Carl Emil Erlenmeyer, usually cited simply as Emil Erlenmeyer. The radiographic analogy is exact: the flared metaphysis is the base of the flask, the narrow diaphysis is the neck, and the loss of the normal concave waist between them is what makes the two look alike.

The finding long predates the name. The earliest recognised example is a radiograph of the left lower limb of a Nubian mummy published by Smith and Jones in 1910, with no diagnosis attached. In 1967 Urteaga and Moseley described the deformity in an ancient skeleton from the Mochica culture of Peru and suggested it represented Pyle disease.

Pathophysiology

Growing tubular bone is continuously reshaped at the metaphysis. As the bone lengthens, the wide metaphyseal segment laid down at the growth plate must be resorbed from the outside by osteoclasts while bone is added endosteally, so that yesterday’s metaphysis becomes today’s narrower diaphysis. This is tubulation, and it is what produces the normal concave di-metaphyseal waist.

Erlenmeyer flask deformity is what that process failing looks like. Three different upstream problems converge on the same shape:

• Defective osteoclast-mediated modelling. The osteoclasts cannot resorb the metaphyseal collar, either because they do not work (osteopetrosis) or because the signalling that governs cortical modelling is disrupted (Pyle disease, the craniotubular dysplasias).
• Osteoclast failure with abnormal bone quality. In the sclerosing disorders the unresorbed primary spongiosa persists, so the bone is both misshapen and abnormally dense.
• Marrow expansion. The marrow cavity is physically expanded from within by storage macrophages (Gaucher cells, Niemann-Pick histiocytes) or by erythroid hyperplasia (thalassaemia). The cortex thins and the metaphysis widens under pressure.

Those three mechanisms are precisely the three radiographic types below, which is why the classification is worth learning rather than just the list of causes.

The three types of Erlenmeyer flask deformity

Faden and colleagues interrogated the International Skeletal Dysplasia Registry radiographic database (1988 to 2007), reviewed 50 registry cases across 12 disorders alongside a literature review, and proposed the classification that is still the standard framework. The discriminator is not the flask shape, which is shared, but what the trabecular bone looks like.

How to use this at the workstation. Ask one question of the film before you reach for a differential list: is the bone around the flask normal, too dense, or too lucent? Normal trabeculae send you to the craniotubular dysplasias, and the skull then separates them. Dense, structureless bone sends you to osteopetrosis and dysosteosclerosis. Lucent, expanded marrow with a thinned cortex sends you to the storage diseases and the haemoglobinopathies, where the spleen and the blood count usually close the case.

How reliable is the sign in each disorder?

Faden’s registry review answered a question the textbooks usually skip: in a patient who definitely has the disease, how often is the deformity actually there? The answer is that it varies enormously, and the associations most people quote are not the most consistent ones.

Two things fall out of this table. First, frontometaphyseal dysplasia was the most consistent association in the registry, at 11 of 11, and it does not appear in the usual mnemonics at all. Second, the two causes everyone names, Gaucher disease and craniometaphyseal dysplasia, showed the deformity in only half and about a third of cases respectively. Faden’s group also recorded that involvement was variable in both appearance and location within the skeleton, so a negative distal femur does not exclude any of these diagnoses.

How to measure it instead of eyeballing it

For most of the sign’s history there was no definition at all. Carter and colleagues at Addenbrooke’s made that point directly: the reported prevalence of the deformity in Gaucher disease varied widely between series precisely because every series was using a different subjective impression of the same shape.

They took knee radiographs on which three experienced radiologists agreed about the presence or absence of the morphology, measured the femur serially along the metadiaphysis, and derived a single ratio.

The practical message is that you can settle an equivocal distal femur with a ruler. The Carter ratio needs one frontal knee radiograph and two diameters. The Adusumilli method, produced by the GAUCHERITE Consortium, matters more for research and for follow-up, because it replaces a present-or-absent call with a graded severity that can be tracked on treatment and is amenable to automation.

Which bones are involved

The distal femur is where the term was coined, where it is most often seen, and the only site for which validated measurements exist. It is not the only site.

• Distal femur, by far the most common and the reference site
• Proximal tibia
• Proximal humerus
• Distal radius and ulna

The deformity may be unilateral or bilateral. In Pyle disease the undertubulation is characteristically widespread, affecting distal femora, proximal humeri and tibiae, and distal ulnae and radii together, with an S-shaped tibia, which is a useful clue that you are dealing with a generalised modelling disorder rather than a focal one.

Gaucher disease: the classic cause, with caveats

Gaucher disease is the most common lysosomal storage disorder, caused by biallelic GBA variants and deficient acid beta-glucosidase, so glucocerebroside accumulates in macrophages. These distended Gaucher cells infiltrate the marrow and expand it, producing the EFD-ME pattern.

Skeletal disease is the most disabling part of the illness. Registry data show that nearly all patients have radiological evidence of skeletal involvement and that a majority have a history of serious skeletal complications, with up to 20% having impaired mobility. Wenstrup and colleagues separated it into three processes:

The caveats matter. Only 3 of 6 registry cases of type 1 Gaucher disease had the deformity, so it is neither sensitive nor required for the diagnosis. It is also not specific: the same marrow-expansion mechanism operates in Niemann-Pick disease type B and in thalassaemia. Bone pain, a bone crisis, hepatosplenomegaly and cytopenias are what should prompt the enzyme assay, not the shape of the femur.

Modern assessment is by MRI, not radiography. The radiograph shows established structural change; MRI shows the marrow burden that is actually driving it and that actually changes on treatment. The bone marrow burden (BMB) score of Maas and colleagues is the widely used semi-quantitative measure, scoring signal in the lumbar spine and femora. Response is real: in 18 Brazilian children with type 1 disease followed on plain films after starting imiglucerase, 13 (72%) improved clinically and radiologically, while the 5 who worsened had received significantly lower doses (29 plus or minus 2 U/kg versus 55 plus or minus 10 U/kg per 15 days, p less than 0.03). Established modelling deformity in a skeletally mature patient, however, will not remodel away.

Osteopetrosis and the EFD-A group

In osteopetrosis the osteoclasts fail, so neither resorption-dependent modelling nor normal remodelling occurs. The metaphysis stays broad and the bone becomes abnormally dense, which is the EFD-A pattern: flask shape plus radiographically abnormal trabecular bone.

Crucially, the deformity is not equally distributed across osteopetrosis subtypes. Spinnato and colleagues set out which imaging features track with which inheritance pattern:

So an Erlenmeyer flask femur plus a rugger-jersey spine in an adult with a low-energy fracture is a coherent picture of autosomal dominant osteopetrosis type II. For the density differential, see pycnodysostosis versus osteopetrosis; pycnodysostosis has acro-osteolysis and an obtuse mandibular angle and does not produce this deformity.

Dysosteosclerosis is the other EFD-A entity, and it showed the deformity in 3 of 3 registry cases. It combines flask-shaped, club-like bones that have thin cortices with platyspondyly, skull base sclerosis and a characteristic red-violet macular skin atrophy. Biallelic SLC29A3 variants are the commonest cause, with TNFRSF11A and TCIRG1 also reported.

Pyle disease versus craniometaphyseal dysplasia

These two are the classic pairing that candidates confuse, because both are named for metaphyseal disease and both flare the distal femur. The separation is the skull, and it is clean.

The genetics explain the paradox in Pyle disease. Kiper and colleagues showed in the New England Journal of Medicine that loss of sFRP4, a secreted Wnt inhibitor, produces widened, poorly modelled metaphyses with thin, fragile cortical bone at the same time. The bones look bulky and break easily. That dissociation between trabecular expansion and cortical fragility is now studied as a model for cortical bone homeostasis in ordinary osteoporosis.

Haemoglobinopathies and the marrow expansion group

In thalassaemia, ineffective erythropoiesis drives massive erythroid hyperplasia that expands the marrow cavity, thins the cortex and widens the metaphysis, the same EFD-ME mechanism as Gaucher disease. It usually travels with the other marrow-expansion findings, including a hair-on-end skull, coarse trabeculae, and facial bone overgrowth from maxillary marrow expansion.

Sickle cell disease is routinely listed with thalassaemia in mnemonics, including the CHONG aid above. Be more careful here. Marrow hyperplasia certainly occurs in sickle cell disease, but the dominant skeletal findings are infarction related: bone infarcts, H-shaped or step-off vertebrae, avascular necrosis of the femoral head, dactylitis and osteomyelitis. Erlenmeyer flask deformity is a far weaker association than in thalassaemia, and it did not feature in the Faden registry cohort at all.

Causes that are widely listed but poorly supported

Textbook lists for this sign have accumulated entries that do not survive contact with the evidence. Worth knowing which ones to demote.

Mimics and how not to overcall it

• Normal metaphyseal flare. The commonest reason for a false positive. A normal distal femur flares, but keeps a concave di-metaphyseal waist. Loss of that concavity, not the width of the metaphysis, is the finding. When in doubt, apply the Carter ratio.
• Rickets. Produces metaphyseal widening, but with cupping, fraying and splaying of an irregular, indistinct physeal margin and a widened growth plate. The Erlenmeyer flask metaphysis has a sharp physeal margin and is abnormal in its modelling, not in its mineralisation.
• Projection and rotation. An obliqued or rotated knee can flatten the apparent di-metaphyseal curve. Measure on a true frontal film.
• Post-treatment and growth arrest lines. Metaphyseal banding after chemotherapy or bisphosphonates alters density, not shape.

Alternative names: Erlenmeyer flask bone deformity (EFD), metaphyseal flaring, undertubulation of the distal femur, flask-shaped femur

Other associated named signs: Bone-in-bone appearance, rugger-jersey spine and sandwich vertebra in osteopetrosis, cod-fish vertebrae in Pyle disease, and the hair-on-end skull in thalassaemia

For the quick exam aid on its own, see the CHONG visual mnemonic for causes of Erlenmeyer flask deformity.

Frequently asked questions

What is the Erlenmeyer flask deformity?

It is a modelling (undertubulation) abnormality of the distal femur in which the metaphysis stays broad and the normal concave waist of the di-metaphysis is lost, so the bone end looks like a conical laboratory flask. The cortex is characteristically thin. It is a descriptive shape, not a diagnosis, and it points to about 20 different disorders.

What causes Erlenmeyer flask deformity?

Faden and colleagues found 20 distinct disorders in a literature review plus a cohort study of the International Skeletal Dysplasia Registry. They fall into three mechanisms: craniotubular dysplasias with otherwise normal trabecular bone (frontometaphyseal dysplasia, craniometaphyseal dysplasia, craniodiaphyseal dysplasia, Engelmann disease, Pyle disease, Melnick-Needles osteodysplasty, otopalatodigital syndrome type I), sclerosing disorders with abnormal trabecular bone (osteopetrosis, dysosteosclerosis), and marrow expansion (Gaucher disease, Niemann-Pick disease type B, thalassaemia).

What are the types of Erlenmeyer flask deformity?

Three, from the 2009 Faden classification. EFD-T (typical) is absent di-metaphyseal modelling with normal looking trabecular bone. EFD-A (atypical) is the same modelling defect but with radiographically abnormal, usually sclerotic, trabecular bone. EFD-ME is deformity caused by marrow expansion from storage material or marrow hyperplasia. Deciding which of the three you are looking at narrows the differential far faster than reciting a list of causes.

How is Erlenmeyer flask deformity measured?

Carter and colleagues derived the first objective definition. On a frontal knee radiograph, divide the femoral shaft diameter 4 cm proximal to the physis by the diameter at the physeal baseline. A ratio above 0.57 predicted the deformity with 95.6% sensitivity and 100% specificity in 43 adults with Gaucher disease. Adusumilli and colleagues later added a 3-point severity scale using volume ratios, with areas under the curve of 0.95 and 0.93.

Which bone is most often affected?

The distal femur, which is where the term was coined and where nearly every published series measures it. The proximal humerus, proximal tibia, and distal radius and ulna can show the same undertubulation. Involvement can be unilateral or bilateral, and Faden’s review specifically noted that the finding is not uniform even within a single disorder.

Is Erlenmeyer flask deformity always present in Gaucher disease?

No, and this is the most over-taught association in the whole topic. Only 3 of 6 registry cases of type 1 Gaucher disease showed it. Reported prevalence has varied widely across series, which is largely why Carter and colleagues set out to define the deformity numerically in the first place. Absence of the deformity does not exclude Gaucher disease.

What is the mnemonic for Erlenmeyer flask deformity?

CHONG: Craniometaphyseal dysplasias, Hemoglobinopathies, Osteopetrosis and osteochondromatosis, Niemann-Pick, Gaucher. It is a useful exam aid for the common causes, but note that it captures none of the EFD-T craniotubular group beyond craniometaphyseal dysplasia, so frontometaphyseal dysplasia, Pyle disease, Melnick-Needles and otopalatodigital syndrome are all missing from it.

How do you tell Pyle disease from craniometaphyseal dysplasia?

By the skull. Both produce marked metaphyseal flaring, but Pyle disease (biallelic SFRP4 loss) has no craniofacial dysmorphism and no cranial nerve compression, with only mild skull base and vault sclerosis. Craniometaphyseal dysplasia (usually ANKH) is dominated by sclerosis of the cranial vault, base and facial bones with foraminal narrowing, causing progressive deafness, optic atrophy and facial palsy. Pyle disease also causes cortical thinning and fragility.

Does the deformity reverse with treatment?

Bone disease in Gaucher disease responds to adequate enzyme replacement. In 18 Brazilian children with type 1 disease followed on plain films, 13 (72%) improved clinically and radiologically, and the 5 who worsened had received significantly lower imiglucerase doses (29 plus or minus 2 U/kg versus 55 plus or minus 10 U/kg). An established modelling deformity in a skeletally mature patient, however, is fixed.

Is Erlenmeyer flask deformity seen in lead poisoning?

It is widely listed, on the strength of older textbook references describing it as a late change, but it is not what lead poisoning actually looks like. The reproducible radiographic finding is dense transverse metaphyseal bands, the lead lines. Treat lead poisoning as a weak, late association rather than a differential you should be offering from the shape alone.

What is the difference between Erlenmeyer flask deformity and normal metaphyseal flaring?

A normal distal femur has a concave waist between the flared metaphysis and the narrower diaphysis. In the deformity that concavity is lost, so the di-metaphyseal borders run straight or even bow outwards, and the cortex is thin. If you are unsure, measure: the Carter ratio above 0.57 separates the two reproducibly, and it is quicker than arguing about the shape.

Is the deformity visible on CT and MRI?

Yes. It is a shape, so any modality that shows the di-metaphysis shows it, and it has been reported specifically on CT. In practice the radiograph remains the reference standard for calling it, while MRI is the modality that grades the marrow disease behind an EFD-ME deformity, using scores such as the bone marrow burden score in Gaucher disease.

References

1. Faden MA, Krakow D, Ezgu F, Rimoin DL, Lachman RS. The Erlenmeyer flask bone deformity in the skeletal dysplasias. Am J Med Genet A. 2009;149A(6):1334-45. PMID 19444897.
2. Carter A, Rajan PS, Deegan P, Cox TM, Bearcroft P. Quantifying the Erlenmeyer flask deformity. Br J Radiol. 2012;85(1015):905-9. PMID 22010032.
3. Adusumilli G, Kaggie JD, D’Amore S, et al. Improving the quantitative classification of Erlenmeyer flask deformities. Skeletal Radiol. 2021;50(2):361-9. PMID 32734372.
4. Spinnato P, Pedrini E, Petrera MR, et al. Spectrum of skeletal imaging features in osteopetrosis: inheritance pattern and radiological associations. Genes (Basel). 2022;13(11):1965. PMID 36360203.
5. Kiper POS, Saito H, Gori F, et al. Cortical-bone fragility: insights from sFRP4 deficiency in Pyle’s disease. N Engl J Med. 2016;374(26):2553-62. PMID 27355534.
6. Wenstrup RJ, Roca-Espiau M, Weinreb NJ, Bembi B. Skeletal aspects of Gaucher disease: a review. Br J Radiol. 2002;75 Suppl 1:A2-12. PMID 12036828.
7. Maas M, van Kuijk C, Stoker J, et al. Quantification of bone involvement in Gaucher disease: MR imaging bone marrow burden score as an alternative to Dixon quantitative chemical shift MR imaging. Radiology. 2003;229(2):554-61. PMID 14526090.
8. Mota RM, Mankin H. Use of plain radiography to optimize skeletal outcomes in children with type 1 Gaucher disease in Brazil. J Pediatr Orthop. 2007;27(3):347-50. PMID 17414023.
9. Simpson WL, Hermann G, Balwani M. Imaging of Gaucher disease. World J Radiol. 2014;6(9):657-68. PMID 25276309.
10. Campeau PM, Lu JT, Sule G, et al. Whole-exome sequencing identifies mutations in the nucleoside transporter gene SLC29A3 in dysosteosclerosis, a form of osteopetrosis. Hum Mol Genet. 2012;21(22):4904-9. PMID 22875837.
11. Alvarenga Fernandes D, Assuncao Jorge M, de Melo Alexandre Fraga A, Reis F. Radiological changes in lead poisoning. Clin Toxicol (Phila). 2024;62(6):404-6. PMID 38899783.
12. Rueda-de-Eusebio A, Gomez-Pena S, Moreno-Casado MJ, et al. Hereditary multiple exostoses: an educational review. Insights Imaging. 2025;16(1):46. PMID 39982564.
13. Ejindu VC, Hine AL, Mashayekhi M, Shorvon PJ, Misra RR. Musculoskeletal manifestations of sickle cell disease. Radiographics. 2007;27(4):1005-21. PMID 17620464.
14. Simpson WL Jr, Mendelson D, Wasserstein MP, McGovern MM. Imaging manifestations of Niemann-Pick disease type B. AJR Am J Roentgenol. 2010;194(1):W12-9. PMID 20028884.
15. Stylianou C, Deftereos S. The Erlenmeyer flask deformity on computed tomography. Acta Med Acad. 2023;52(2):146-7. PMID 37933512.