J. Neurol. Neurosurg. Psychiat., 1969, 32, 99-110

Intrathoracic meningoceleIts development and association with neurofibromatosis

JOHN MILES,1 JOE PENNYBACKER, AND PHILIP SHELDON

From the Department of Neurological Surgery, Radcliffe Infirmary, Oxford

An antero-lateral meningocele is a protrusion of thespinal meninges through an intervertebral foramen.It contains an extension of the subarachnoid space,filled with cerebrospinal fluid, and this gives rise toa paravertebral cystic swelling, usually spherical inshape. Theoretically such a swelling could occur inthe neck, or in the thoracic, abdominal, or pelviccavities, but practically all the recorded cases (andwe have discovered only 66 in literature) have oc-curred in the thoracic cavity and are there referred toas intrathoracic meningoceles.A feature of the abnormality is that nearly two-

thirds of the recorded cases have occurred inassociation with neurofibromatosis. It has thus beenof especial interest to neurological and thoracicsurgeons because of the differential diagnosis fromdumb-bell neurofibroma-that is, a neurofibromabeginning on a spinal nerve root, enlarging in thevertebral canal, extending out through the inter-vertebral foramen and enlarging again in theparavertebral space. Indeed the original observationby Pohl (1933) concerned a 47-year-old woman withvon Recklinghausen's disease who was mistakenlythought to have an intrathoracic neurofibroma. Themajority of the recorded cases have come into thehands of thoracic surgeons, and the increasingpractice of mass radiography of the chest may beexpected to provide more cases in the future.

Because of their rarity, little is known about thenatural history of intrathoracic meningoceles and,that being the case, the indications for treatment andthe most effective methods of treatment are by nomeans clear. We have not encountered a comprehen-sive review of the literature.We present below a report of four cases seen in

the Department of Neurological Surgery, RadcliffeInfirmary, Oxford, during the past 18 years, togetherwith a review of the literature, a discussion of the

'Present address: Department of Neurological Surgery, QueenElizabeth Hospital, Birmingham.

A selected set of references is appended, but a full list of 69 referencesmay be obtained by application to: The Department of NeurologicalSurgery, Radcliffe Infirmary, Oxford.

various theories of aetiology, and suggestions as totreatment.

CASE REPORTS

CASE 1 A.G. (R.I. No. 96373), a 66-year-old retiredbarman, with lifelong evidence of neurofibromatosis,presented in January 1949 with a year's history ofprogressive paraparesis. On examination he had anupper thoracic scoliosis and a moderate spastic para-paresis. There was a relative sensory deficit, most markedto pain and temperature stimulation, with an upperlevel at the 4th thoracic dermatome. Radiographs(Fig. 1) showed the gross scoliosis, convex to the left, atthe apex of which there was a rounded opacity filling theupper pole of the thoracic cavity on that side. Lateralradiograph confirmed this position and also revealedgross scalloping of the posterior surfaces of the bodiesof the 2nd, 3rd, and 4th thoracic vertebrae.Lumbar puncture produced slightly yellow fluid, which

proved to contain 100 mg/100 ml. protein, at a pressureof 50 mm water. Queckenstedt's manoeuvre produced arise of only 55 mm in 10 seconds and the return to normaltook 45 seconds-that is, there was thought to be a partialmanometric block. Lipiodol myelography showed anincomplete arrest at the 6th thoracic vertebra and noneof the contrast medium passed out of the vertebral canal.The presumptive diagnosis was compression of the spinalcord by dumb-bell neurofibroma.At operation on 14 January 1949, the laminae of the

2nd, 3rd, and 4th thoracic vertebrae were found to beabnormally thin. When they were excised, no tumourwas visible in the vertebral canal. On opening the duraa very capacious subarachnoid space was revealed, butagain there was no tumour. The three intervertebralforamina on the left, at the apex of the scoliosis, appeareddilated, especially that between the 2nd and 3rd thoracicvertebrae, which had a cross-sectional diameter of about5 cm, and through this the meninges passed out into thechest as a cystic diverticulum. The wall of the cyst was sothin that the carbon-mottled lung surface and its move-ment on respiration could be seen through it. The spinalcord was tightly stretched across the apex of the scoliosis.There seemed no indication to interfere with the meningo-cele. Lipiodol was placed in the cavity, and its position,in the upper chest opacity, was confirmed radio-graphically.

Thirty-six hours post-operatively, the patient developed99

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

S

FIG. 1. Case 1. Upper thoracic scoliosis with meningocelefilling the left thoracic apex.

a left-sided pneumothorax which proved fatal. Necropsyprovided no explanation for the pneumothorax andmerely confirmed the surgical findings (Fig. 2). Duringthe operation it had been observed that there wasabnormal mobility of the vertebral column at the site ofthe meningocele, and this also was confirmed at necropsy.It was suggested that this abnormal mobility at the sitewhere the cord was tethered over the kyphus may havecontributed to the paraparesis.

CASE 2 E.S. (R.I. No. 147752), a 48-year-old housewife,was admitted on 25 September 1951 after a roundedopacity in the right lower chest had been discoveredduring a barium meal examination for dyspepsia (Fig. 3).She also complained of low backache and slight weaknessand nocturnal paraesthesiae of the left leg.On examination she was found to have 'cafe-au-lait'

patches of neurofibromatosis. There was slight weaknessof the left leg without reflex or sensory changes. Chestradiographs showed the opacity to be centred over thethe right sides of the 11th and 12th thoracic vertebrae,with considerable scalloping involving the posteriorsurfaces and the right sides of the bodies of the 10ththoracic to the 1st lumbar vertebrae. The vertebral canalwas widened with thinning of the pedicles over the sameregion, and the head of the right 12th rib was eroded.There was slight kyphotic deformity of the spine at thislevel, but no significant scoliosis. Lumbar puncturerevealed normal cerebrospinal fluid with normal man-ometry. The diagnosis of antero-lateral meningocele was

FIG. 2. Case 1. Necropsy specimen showing the meningo-cele seenfrom the thoracic cavity, with the vertebral bodiesremoved to show the continuity of the spinal meninges withthe wall of the meningocele.

confirmed by air myelography: air injected into thelumbar subarachnoid space passed freely into themeningocele (Fig. 4).

Because of the symptoms and the slight left leg weak-ness, thoraco-lumbar laminectomy was performed on22 October 1951. As in the first case, there were thinnedlaminae, and expansion of the vertebral canal and sub-arachnoid space extending over at least three segmentswith no evidence of either tumour or compression ordistortion of cord or roots. The meningocele passed outthrough an intervertebral foramen dilated to 2 5 by 3 cm.The intercostal nerve and accompanying artery wereseen passing through the cavity. The neck of the meningo-cele was dissected free from the bony margins of theintervertebral foramen and was ligated but not divided.There were no post-operative complications and the painand paraesthesiae disappeared immediately.

Fifteen years later (1966), there has been no recurrenceof symptoms. On examination there is now definite

100

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

Intrathoracic meningocele

11

FIG. 4. Case 2. Air myelography has resulted in thereplacement of some of the cerebrospinal fluid in themeningocele by air. With the patient in the vertical positiona fluid level results as seen. The arrows indicate the thinmeningocele wall outlined by air.

FIG. 3. Case 2. Lateral radiograph of the chest with themeningocele outlined by arrows, and showing the posteriorvertebral body scalloping.

kyphosis with little scoliosis, and, while the patientregards the power in her legs as normal, the left thigh isa little wasted and the left knee and ankle reflexes areexaggerated. Sensation is normal. Radiographs show theopacity to be essentially unchanged in size with com-parable posterior scalloping but increased lateral scallop-ing. The rib erosion is unchanged.

CASE 3 A.G., a 24-year-old student with neurofibro-matosis, was found to have a rounded opacity in the leftlower thorax at routine chest radiography in 1955. Theposterior surfaces of the bodies of the 8th to the 12ththoracic vertebrae were considerably scalloped and thebodies themselves were irregular in outline. There wasa slight scoliosis concave towards the opacity. The leftpedicles of the 10th to the 12th thoracic vertebrae werethinned and that of the 9th was absent. The interpediculardistance was increased at these levels. The heads andnecks of the left 9th and 10th ribs were eroded (Fig. 5,a + b). There was a prominent posterior 'cafe-au-lait'patch in the left paravertebral position over the lower

six ribs. The character of the opacity and more particu-larly the associated bony abnormalities suggested thediagnosis of antero-lateral meningocele, but, as therewere neither symptoms nor signs referable to the chestor the spinal cord, no treatment was advised. Over thesucceeding 12 years he has continued to be symptom-free,but, radiologically, there has been a definite increase inthe size of the meningocele: its volume was estimated as55 ml. in 1955, and as 241 ml. in 1967.

CASE 4 T.M. (R.I. No. 2563), a 24-year-old chef, abinovular twin, had neurofibromatosis though there wasno family history of such. At routine chest radiographyon 13 April 1966 an opacity was seen over the heads of theleft 10th and 11th ribs, which were grossly eroded (Fig. 6).There was slight kyphoscoliosis and gross posteriorscalloping of the bodies of the 8th to the 11th thoracicvertebrae, with posterior collapse (reversed wedging) ofthe 10th thoracic vertebral body. There was expansionof this segment of the vertebral canal. He had complainedfor three years of slight mid-thoracic and lumbar back-ache with occasional cramp in the left leg, and had ob-tained some relief from a plaster of Paris support.On examination the power and the tone in the legs

were normal, although the left knee reflex was diminishedand the ankle reflex was absent. There was no sensory

101

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

FIG. 5a and b. CASE 3. Penetrated radiographs (1955) showing (a) the meningocele lying over the eroded heads of theleft 9th and 10th ribs; (b) the meningocele and the degree ofposterior scalloping.

deficit. Again it was felt that at present no treatment wasindicated and that no further investigations werenecessary. However we had previous records of thispatient. He was first seen in 1943 as a 10-month-old baby,when he presented with right upper lobe pneumonia, andreview of the chest radiographs taken at that time nowreveal thinning of the left 10th and 11th ribs and thesuggestion of a lower thoracic scoliosis. Subsequentradiographs taken because of recurrent chest infectionconfirmed these abnormalities. At the age of 5 hepresented again with a left occipital skull defect, and atthis examination 'cafe-au-lait' patches were found on histrunk and these had appeared only over the precedingfew months. Because of the known association of bonydysplasia with neurofibromatosis, further radiographswere taken and those of the vertebral column (Fig. 7)showed posterior scalloping of the 8th to 11th vertebraeand, to a lesser extent, the 2nd to 4th lumbar vertebraeinclusively, with widening of the canal and thinning ofthe pedicles. The rib changes were confirmed, and thin-ning of the 10th and 11th thoracic transverse processeswith, in fact, costo-transverse disarticulation was now

also seen. As there was no neurological deficit andlumbar puncture revealed normal cerebrospinal fluidwith normal manometry, no surgery was undertaken.Frequent radiographs were taken up to the age of 13(1953) and these showed only slight increase in thescalloping. The child remained well and active.

It was not until 1966, when he was 24, that the meningo-cele was discovered on a routine chest radiograph takenin connection with his work as a chef. Retrospectiveexaminations of previous chest radiographs show thatit was not present in 1957, although by 1961 it was justvisible; at that time it was smaller and almost obscuredby the cardiac shadow. The volumetric increase, estimatedradiographically, was from a few ml. in 1961 to 65 ml. in1966. There has been a further increase in the last year to85 ml.

REVIEW OF THE LITERATURE

Of the 70 cases now reported in the literature, 38were women and 31 were men. We have chosen to

102

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

Intrathoracic meningocele

1967

exclude from this series three cases that have beenincluded in some previous series. In each we feelthat the evidence for the protrusion being an antero-lateral meningocele is inconclusive.

AGE OF PRESENTATION The meningocele is usuallydetected in middle life, the average age in therecorded cases being 43-3 years; 74% presentedbetween the age of 30 and 60 years.

SITE In 37 cases (52 %) the meningocele was on theright; in 27 cases (40%) it was on the left. In eightcases there was more than one meningocele and infive of these there was a meningocele on each side.In one case the side was not specified. The apparentright-sided preference may be artificial, as smallleft-sided meningoceles may be obscured in theradiograph by aortic and heart shadows. As an

example of this, LaVielle and Campbell (1958)thought that they were dealing with a single men-ingocele until the post-operative film, as well as

FIG. 6. Case 4 in 1967.Penetrated radiographs show-ing (a) the meningocele in thelateral projection and alsoshowing the severe posteriorvertebral scalloping with

~10 posterior collapse of the 10ththoracic vertebral body;(b) underlying left 10th and11th ribs are shown to haveeroded heads and thinnednecks, and there is upwardidslocation of the 10th rib.

showing satisfactory removal of the right-sided10th thoracic meningocele, also showed a previouslyunrecognized meningocele at the same level on theleft side. In our fourth case, similarly, the men-ingocele was recognized only when it had expandedbeyond the cardiac shadow. Wilhelm (1954)suggested that there might be particular paucity oflower left meningoceles, and that this could be dueto the buttressing effect of the descending aorta. Wefound a random distribution along the thoracicspine, with the exception that, for no obviousreason, as many as 13 (20%) occurred between the10th and 11th thoracic vertebrae.

MODE OF PRESENTATION (Table I) Chance findingsThirty-two meningoceles were discovered on routineradiological examination of the chest, nine onradiography for completely unrelated symptoms,and there was one chance necropsy finding. A total,therefore, of 42 (60%) could be considered chancefindings. Many of the others presented with symp-

103

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

FIG. 7. Case 4 in 1948. On the left show-ing the thinning of the proximal parts ofthe 10th and 11th ribs. In the lateralprojection posterior vertebral body scallop-ing is seen in the lowest thoracic vertebraand again in the mid-lumbar vertebrae.

toms only doubtfully related either to the meningo-cele or to the part of the spine bearing it.

Pain Though not a characteristic feature, painwas present in 16 cases (23 %). All but one of thosewho had pain also had kyphosis, and in the exceptionthe presence or otherwise of kyphoscoliosis was notspecified. In nine the pain was referred to some-though not necessarily the appropriate-part of thevertebral column, and in six it was of a radiating,intercostal type, conceivably referable to themeningocele. In no case was there mention ofpleuritic pain.

Neurological abnormalities These were on thewhole slight and uncommon. Only six patients (9 %)had significant paraparesis, but another seven hadslight abnormalities such as hyper-reflexia, slightweakness, vague sensory impairment, and the like.Kyphoscoliosis was specified as being present in allbut one of those who had any kind of neurologicaldeficit in the legs.

TABLE IMODE OF PRESENTATION

Presentation (no.) (%)

Chance finding 42 60Pain 16 23Dyspnoea 8 11Cough 7 10Paraparesis 6 9Direct presentation: 2 3

1 rupture1 bleed

Dyspepia 1

Pulmonary symptoms Cough was present inseven (10%) and dyspnoea in eight (11 %). Only inthose cases in which the meningocele assumedmassive proportions was the respiratory embarrass-ment severe, as in the case reported by Waterfall(1966).

Presentation directly related to the meningoceleThis occurred twice. Portigliatti-Barbos (1953) hada case in which the meningocele ruptured during about of coughing, when it was penetrated by apreviously fractured rib. The patient died frommedullary coning due to leakage of cerebrospinalfluid into the pleural cavity. Another patient pre-sented with spontaneous haemothorax originatingfrom the very vascular wall of the meningocele, andsurvived its excision. This vascularity proved fatal atone operation and was a serious problem at another.Dyspepsia and dysphagia were not previouslyrecorded but may have been a feature of our secondcase.

RADIOLOGICAL FINDINGS (Table II) KyphoscoliosisSome degree of kyphoscoliosis varying from slight tovery gross was present in 47 cases (67 %). In 15 cases(21 %) there was said to be none. The kyphoscoliosisusually had the meningocele at its apex on the convexside, but in our third case there was a slight scoliosisconcave toward the meningocele. When there werebilateral meningoceles, there was no obvious reasonwhy the scoliosis should have been to one side ratherthan to the other.The incidence of kyphoscoliosis in people with

104

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

Intrathoracic meningocele

neurofibromatosis has been variously estimated as

being between 9% according to Hunt and Pugh(1961) and 38% by Laws and Pallis (1963). Thisassociation has been further emphasized in thisseries in that, if a patient had a meningocele,kyphoscoliosis appeared to be present when therewere other recognizable stigmata of neurofibro-matosis. Thus of 46 patients with meningocele andobvious neurofibromatosis, 39 (85%) had kypho-scoliosis, while of the 15 without any evidence ofneurofibromatosis only five (33%) had kypho-scoliosis.

Scalloping This was present in 48 cases (69%)and again was usually associated with kyphoscoliosis.The number of vertebral bodies affected varied fromtwo to six, and in our fourth patient the spineseemed to be involved at two separate levels, lowerthoracic and lower lumbar. Reversed wedging-that is, collapse of the posterior part of the vertebralbody-has been noted by Laws and Pallis (1963)and in our fourth patient was shown tomographic-ally to be related to the degree of posterior scalloping.

Enlargement of the intervertebral foramen Thiswas a common finding but was specifically said to beabsent in five cases. It was also common for severaladjacent foramina to be enlarged, even though themeningocele emerged through a single foramen.

Rib changes These were recorded in 36 cases(51 %) and were probably not present in nine cases.They varied from splaying with increase in theintercostal space, to gross thinning of the adjacentsurfaces of the heads and necks at the site of themeningocele. Costo-transverse dislocation occurredwith the most marked thinning, and Waterfall (1966)described progression of rib changes accompanyingthe expansion of the meningocele. The possiblesignificance of rib changes before the developmentof the meningocele will be discussed later. Theseproximal rib changes are different from thosecommonly associated with neurofibromatosis-that is, the 'twisted ribbon' type of deformity whichusually occurs in the more distal parts of the shaftof the rib.

Other bony abnormalities These are rare,although synostosis of adjacent vertebral bodies was

seen in the vicinity of the first meningocele that wasdescribed by Pohl (1933).

Radiological evidence of enlargement In 12 casesincrease in the size of the meningocele was known tohave occurred over periods of from two to 10 years.In two cases, the meningocele ultimately attaineda massive size, occupying two-thirds of the hemi-thorax. No enlargement of the meningocele occurredin six instances followed for periods of from one tonine years.

Abnormalities in adjacent organs Pulmonarycysts have been seen at the same level as themeningocele and at laminectomy an arachnoid cystwas found just above another meningocele.

THE INCIDENCE OF NEUROFIBROMATOSIS In 46 cases(64 %), there was definite evidence of neurofibro-matosis; six were said to have none, but one of thesehad a positive family history. In an attempt to explainthe absence of signs of neurofibromatosis in somecases, Wilhelm (1954) suggested that their appear-ance might be delayed until after the meningocelewas discovered. Voisin, Macquet, Wattel, Mahieu,Leduc, and Jacob (1964) likewise emphasize theevolutionary progression of overt evidence ofneurofibromatosis. Our fourth patient had bonyabnormalities before cutaneous manifestationsappeared, and Teng and Eastman's (1958) patienthad but one small subcutaneous occipital neuro-fibroma to suggest the presence of this disease.Cross, Reavis, and Saunders (1949) considered thatthe meningocele might be a forme fruste of neuro-fibromatosis, and the only apparent manifestation.The siting of the skin stigmata did occasionallyseem to have some relevance, as with the largeoverlying caf&-au-lait patch in our fourth patient,and adjacent intercostal neurofibromata have alsobeen described.

Other conditions known to be associated withneurofibromatosis also occurred in this series andincluded two phaeochromocytomata, two pontinegliomata, and a presumed optic nerve glioma. In onepatient sarcomatous change occurred in a peripheralneurofibroma.

The Differential Diagnosis This always included

TABLE IIRADIOLOGICAL DATA

Kyphoscoliosis Vertebral Foraminal Rib changes Evidence ofscalloping enlargement enlargement

(no.) ( %) (no.) (5/,) (no.) (%) (no.) (5%) (no.) (%)

Present 47 67 48 69 50 71 36 51 12 17Definitely absent 15 21 7 10 5 7 9 13 6 9

105

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

neurogenic tumour-for example, neurofibroma,neuroblastoma, ganglioneuroma, etc-and in one

case ruptured aortic aneurysm. Thirty-six of themeningoceles were operated upon with a presumptivediagnosis of intrathoracic neurogenic tumour-usually because of the risk of malignancy, which isquoted by Kent, Blades, Valle, and Graham (1944) as

being as high as 40% in this type of tumour. How-ever some authors, including Hilton and McCarthy(1959), have suggested that in the presence of neuro-

fibromatosis, a paravertebral shadow is more likelyto be a meningocele than a neurogenic tumour. Sup-porting this relative dissociation between generalizedneurofibromatosis and intrathoracic neurogenictumour, Bikfalvi (1964) found evidence of the generaldisease in only one of the 28 cases of intrathoracicneurogenic tumour upon whom he operated. How-ever Wilhelm (1954) questions the practical value ofsuch deductions when intrathoracic neurofibroma isso much more common than intrathoracic meningo-cele.Laws and Pallis (1963) support the view that the

associated adjacent bony abnormalities-that is, inthe vertebrae and ribs-are themselves sufficientlycharacteristic for the correct diagnosis to be made,while others argue the need for special investigations.With Laws and Pallis we think that by virtue of thesecharacteristic vertebral abnormalities, and particu-larly their extending over several segments, it shouldbe possible to make the correct diagnosis withoutresorting to myelography, as long as there is no

evidence of paraparesis. In all cases with paraparesismyelography is necessary to assess possible cordcompression due to a dumb-bell tumour or

kyphoscoliosis. If the chest opacity is felt to be an

intrathoracic neurofibroma, and in spite of therebeing no evidence of cord compression, there is stillconcern at the possibility of an intravertebralextension, the relatively simple procedure of lumbarpuncture might be of some help: a normalQueckenstedt response and a normal amount ofprotein in the cerebrospinal fluid might provideconsiderable reassurance. However if there were

evidence of a partial or complete block, myelographyis needed to differentiate between the possiblecauses.

Regarding the cerebrospinal fluid itself, we havenot been able to explain the findings in our firstpatient. There was evidence of obstruction by theimpaired response to jugular venous compression,but, whereas the protein content of the fluid was

only 100 mg/100 ml., the fluid was yellow, whichone usually expects only in association with muchhigher concentrations of protein. We found no

similar reports in the literature.Diagnosis The correct diagnosis was made pre-

operatively in 29 cases: in 18 by Myodil or Lipiodolmyelography as first used by Schuller and Uiberall(1938). Myelography failed to reveal the meningoceleon two occasions, merely indicating some degree ofblock at the appropriate level, probably due tokyphoscoliosis. Diagnosis was achieved by directpuncture and aspiration twice, and confirmed bydirect instillation of Lipiodol on two other occasions.Indigo carmine was instilled directly into anotlhermeningocele and then stained cerebrospinal fluidwas withdrawn at lumbar puncture. The diagnosiswas made in one case by comparing the cerebrospinalfluid pressures of simultaneous direct puncture andlumbar puncture. The safest diagnostic procedurewould seem to be air-myelography, which was firstperformed in this type of abnormality by Cross in1949 and has been used successfully eight times since.

TREATMENT Corrective surgery This was per-formed on 42 meningoceles, with excision viathoracotomy on 33, and by laminectomy on two.A simple ligation after laminectomy was made on oursecond case. Exploration without any attempt atsurgical correction was made twice and was com-bined with aspiration of the exposed meningoceleonce. Neurotomy for intractable pain was necessaryin one case.

Observation byfollow-up This procedure, withoutoperation, was decided upon for 20 cases.

RESULTS OF OPERATION Improvement It has beendifficult to evaluate improvement after operationbecause the symptoms have often been vague andsometimes difficult to relate purely to the men-ingocele. When pain was a feature, relief wasobtained in four cases, though it was unaffectedin three others. Dyspnoea and headache were eachimproved in two cases. Our second patient'sdyspepsia was cured and weakness of her left legsubjectively and objectively improved, though whenseen 15 years after operation there were stillabnormal neurological signs to be found in thelimb. There has been no report of improvement ofparaparesis.

Operative mortality There were eight (19%)deaths directly related to the operation, though thisfigure is somewhat artificially weighted by the threeearly failures which occurred before the use ofantibiotics. Three died from leakage of cerebro-spinal fluid and subsequent infection, two fromrespiratory failure, one due to pneumonitis andthe other, our first case, due to unexplainedpneumothorax. Wilhelms' (1954) patient died fromcirculatory failure due to a coincidental phaeo-chromocytoma, and Byron, Alling, and Samson's(1949) because of operative haemorrhage. The reason

106

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

Intrathoracic meningocele

for one death is not stated, and there was anotherdeath arising from the consequences of a paraplegiathat had appeared only after two operations on themeningocele.

Operative morbidity Non-fatal leakage of cere-brospinal fluid occurred after four operations, andin two there was evidence of meningitis. Operationhas been followed by permanent paraparesis,temporary aggravation of existing paraparesis, andHomer's syndrome.

Recurrence Radiological evidence of recurrencewas seen only once by Bogedain, Carpathios, andLawland (1963) and that was four years after opera-tion. In this case the closure of the defect had beendifficult, and an immediate cerebrospinal fluid fistulahad necessitated re-operation and fascial repair. Atnecropsy, five years after the original operation,the graft was found to be thinned but intact.

HISTOLOGY In 16 cases histology of the meningocelewall showed it to be thin (measuring only 1 to 2 mm),smooth lined, and looking very similar to the spinalmeninges with which it became continuous at theintervertebral foramen. Microscopically dura materwas seen in 13 cases, while both dural andarachnoidal meninges were recognizable in three.Hutchin and Mark (1964) found well-defined lininglayers of cells in each of his three cases. In the firstthe lining cells were cuboidal, while in the secondand third cases the cells were flattened andsquamous. Bikfalvi (1964) also found a well-definedlining layer of cells that were either cuboidal orcylindrical. Nerve elements have been found in thewall, being isolated nerve fibres in six cases withadditional ganglion cells in two instances. Portigliatti-Barbos (1953) found some smooth muscle fibresincorporated in the wall of the meningocele that heexamined. The meningocele removed by Nanson(1957) appears to have been a little different from theothers examined in that, although the upper part wasonly 2 mm thick and smooth lined, the lower wallwas 2-6 cm thick and lined by gelatinous tissue.Microscopically this lower part seemed to be thepicture of a degenerating neurofibroma. Thehistology of another case is described only as beingthat of a 'cystic lesion'.

DISCUSSION

One interesting aspect of this unusual condition isits uncertain aetiology. In discussing the aetiologicalpossibilities it would seem necessary first to considerthe frequent association with neurofibromatosis.

Neurofibromatosis (von Recklinghausen's dis-ease) is a hereditary condition transmitted by thedominant gene, and the well-known visible

stigmata, if not already present at birth, develop orevolve in a progressive manner throughout life.With the obvious ceeects of skin and nerves it wasoriginally considered to be an example of neuro-ectodermal dysplasia, but since it has become clearthat all organs can be affected, it is now thought tobe an abnormality sited at the meso-ectodermaljunction. For our purpose we can confine ourinterest to the skeletal abnormalities that Hunt andPugh (1961) found to occur in 50% of those withneurofibromatosis. More specifically, consideringvertebral abnormalities alone, the most commonlyrecognized is scoliosis, which is variously quoted asoccurring in from 9 to 38 %. Laws and Pallis(1963), in an intensive radiological survey of thevertebral abnormalities in 18 patients with neuro-fibromatosis, found that 38% had scoliosis, 28%had posterior scalloping, 22% had enlargement ofthe intervertebral foramen, and 18 % erosion ofpedicles. In the region of the spine bearing suchabnormalities of the vertebrae, the subarachnoidspace was found to be expanded in each of the threecases that had myelography. The expansion occurrednot only into the posterior excavations of thevertebral bodies, but also as saccular pouchespassing out into the intervertebral foramina. Theyalso described one intrathoracic meningocele with

FIG. 8. Myodil outlining the saccular antero-lateralexpansions of the cervical subarachnoid space in a womanwith neurofibromatosis.

107

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

proximal rib erosion. Such abnormalities, extendingover several segments, have been found at all levelsalong the vertebral column and have been seen inassociation with antero-lateral meningocele in thecervical (Fig. 8) and in the lumbar region.Even such a brief account of the vertebral abnor-

malities already recognized as occurring in neuro-fibromatosis must emphasize the close relationshipbetween these and the abnormalities which review ofthe literature shows so often to be associated withintrathoracic meningocele. The appreciation ofthese similarities should make consideration of thevarious theories of aetiology a little clearer.

AETIOLOGY Nerve root sleeve prolongationSengpiel, Ruzicka, and Lodmell (1948) suggestedthat the sleeve of meninges that normally accompan-ies the nerve root and fuses with the intercostal nerveperineurium in the intervertebral foramen might beabnormally prolonged, ending outside the foramenand thereby giving rise to a form of congenitalherniation of the subarachnoid space. The pressuredifference between the cerebrospinal fluid inside the'hernia' and the intrapleural pressure outside,especially during manoeuvres such as coughing andsneezing, would then tend to cause progressiveexpansion of this meningocele. There is somesupport for this hypothesis, especially in the factthat nerve elements have frequently been found inthe wall of the meningocele. The finding, at operationon our second case, of the intercostal nerve andartery passing through the meningocele must alsogive some support to this argument.

Cystic degeneration in a neurofibroma As pre-viously mentioned, Nanson (1957) found histologicalevidence of a degenerating neurofibroma in thelower pole of one excised meningocele and suggestedthat such cystic degeneration occurring in anintrathoracic neurofibroma might be the primaryaetiological process and that communication with thesubarachnoid space occurred only as a secondaryfeature. Cystic change can occur in intrathoracicneurogenic tumours. Neither of these theories,however, explains the associated bony changes inadjacent vertebrae.Trauma Cross et al. (1949) described an injury

to the back immediately before the development ofthe presenting symptom of pain in their patient,while Stuber (1949) in his report of a meningocelefound at necropsy noted a history of back traumathat had occurred seven years previously. Someauthors consider that trauma could be important,especially when there was no evidence of generalizedneurofibromatosis. In general, however, there islittle to implicate trauma in the development ofantero-lateral meningoceles.

Dural dysplasia Several authors have suggestedthat the meningocele develops at a site of duraldysplasia. If the dura were to be poorly developed,the pulsatile pressure of the cerebrospinal fluid andits erosive effect would be directly transmitted to theadjacent skeleton. Beginning with the vertebralbodies, the pedicles, and the laminae, and eventuallyalso affecting the ribs, the picture of multiple bonyerosions would ultimately be arrived at. This is alogical explanation, but at present without factualsupport. It is perhaps worth noting that pulsationof the meningocele at operation has never beendescribed.Bone dysplasia If the vertebrae themselves were

dysplastic, either congenitally thin or in some otherway more vulnerable, then herniation of theunsupported, although normal, meninges wouldlikewise occur, again as a result of the differentialpressure across the meninges. Against this hypoth-esis is the evidence that there has been only onerecurrence after excision, in spite of the fact that atno time has any attempt been made to repair thebony defect. Again, the suggestion that the ab-normality of bone alone is responsible for thedevelopment of the meningocele must be purelyhypothetical.

Regional dysplasia or dystrophy Being lessspecific as to which are the primary and which thesecondary effects, some authors have suggestedregional dysplasia or dystrophy as being responsiblefor the development of these meningoceles. Affectingmany different tissues and in this way relating thewidespread changes in meninges, vertebrae, and ribsthat are found in neurofibromatosis and also in thevicinity of an antero-lateral meningocele, it must bean attractive hypothesis.Moore (1941) considered the regional changes in

neurofibromatosis to be dystrophic, and due tofaulty autonomic innervation. Hagelstam (1946)found histological evidence of an endarteriticphenomenon in the affected tissues that he studied,and considered that it supported the theory of'neurotic atrophy'.

Congenital derangement Intrathoracic menin-gocele is probably rarely truly congenital-that is,present at birth. In support of this statement manyof the cases, including our fourth, have had theevidence of previously normal chest radiographs.However the extent of the adjacent congenitalderangement of vertebrae and ribs is such that thesecondary development of a meningeal herniationshould not be surprising. The manner in which suchabnormalities of vertebral body, arch, and ribs, arelinked can be explained embryologically. Themesenchymal sclerotome migrates dorsally from itscondensation position around the notochord to

108

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

Intrathoracic meningocele

form the neural arch and laterally to form thetransverse processes and ribs. Initially in con-tinuity-and presumably therefore similarly vul-nerable to dysplastic influences-they only laterseparate and by means of specific ossification centresdevelop independently. Our fourth case provides arecord of congenital derangement preceding, andpossibly contributing to, the development of themeningocele. It does not, however, exclude thepossibility of there being both a congenital derange-ment and a regional dysplasia or dystrophy. Thedysplasia or dystrophy could be considered second-ary in that its evolution might be due to someautonomic or neurotrophic derangement. The factoror factors responsible for the greater frequency andsize of antero-lateral meningoceles in the thoracicregion is also unexplained. It is probably necessaryto invoke as an explanation more than just thefacility for chest radiography in modern medicine.The simplest explanation lies in the difference inpressure between the spinal subarachnoid space andthe pleural cavity, which at rest would amount toaround 10 cm water and which at times of inspirationor cough might be much greater. Perhaps supportingthis, there was no further expansion in the size ofthe meningocele in our second case after ligation ofits neck. Against this hypothesis is the fact that afterexcision there has been only one recurrent men-ingocele, whereas the pressure gradient should haveremained unchanged by the operation.

CONCLUSION

Intrathoracic meningoceles, per se, rarely cause anysymptoms. The one symptom that seems unequivo-cally related to the meningocele, but then only whenit is a very large one, is dyspnoea. Pain is almostcertainly related to the associated kyphoscoliosis, inparticular to the degree of angulation, and LaVielleand Campbell (1958) were able to relieve it by fusionafter it had been uninfluenced by the prior removal ofthe meningocele. It is difficult to imagine themeningocele influencing the spinal cord and causingparaparesis, and more probably again it is thekyphoscoliosis that is to blame. However Buonoand Osacar (1961) have described displacement ofthe cord towards the meningocele until incision ofthe arachnoid, after which it returned to a morenormal position; but this patient had no paraparesis.Chandler and Herzberger (1963) at laminectomyobserved a previously non-pulsatile cord to pulsateafter excision of a right-sided meningocele at thelevel of the fifth thoracic vertebra. They felt that asa result of this procedure there had been someimprovement in the power of the right arm, whichtogether with both legs had been weak before the

4

operation. The paraparesis remained unchanged.Our first case had a paraparesis that was attributedto his kyphoscoliosis, possibly due to excessivemobility of the spine at this point.

There is little information as to the influence ofthe meningocele or of its removal on the progressionof the adjacent bony changes. Our fourth case hasshown some added rib thinning since the develop-ment of his meningocele, while Waterfall's (1966)case showed progressive rib splaying with increasein the size of the meningocele. Our second case hassome added kyphosis and increased lateral vertebralbody scalloping in spite of ligation of the meningoceleneck.

TREATMENT As intrathoracic meningoceles are oftensymptomless, and as they do not invariably increasein size, and as the risks of operation are by no meansnegligible, the question of treatment is difficult. Ifthe meningocele is a chance radiological finding andis causing no symptoms, the right course would seemto be continued observation to determine whetherit is enlarging, and the temptation to remove itsimply because it is there may have to be resisted.If there is evidence of progressive enlargement-and this has occurred in two-thirds of the 18 casesin which there have been serial observations-aninitial attempt should be made to ligate the neck ofthe sac via laminectomy. This may not always bepossible, but, depending on the anatomical arrange-ments, it may be feasible to occlude the neck of thesac in other ways-for example, by plicating sutures.

If it seems impossible to deal with the lesion vialaminectomy, and more specifically if the meningo-cele is massive and producing respiratory symptoms,it should be approached by thoracotomy. If it canbe excised, this should be done, but it is essentialthat a watertight closure of the neck be obtained. Ifthe thickness of the wall of the sac, or the presenceof adhesions to surrounding structures appears toprohibit a satisfactory closure, the attempt shouldbe abandoned. The same principles apply if themeningocele is discovered accidentally at thora-cotomy-for example, in an exploration for asupposed neoplasm.

SUMMARY

The four cases here presented make a total of 70recorded cases of intrathoracic meningocele, andthese have been reviewed. Discovery has usuallybeen as a chance finding, but dyspnoea, due to alarge meningocele, has been reported. There isevidence both for and against progressive increase insize.The possible aetiology has been discussed, and

109

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from

John Miles, Joe Pennybacker, and Philip Sheldon

we think there is evidence for a congenitally based,regional bony derangement, affecting vertebrae andribs, being present before and possibly contributingto the later development of a meningocele at thissite. Other factors, such as abnormal peripheralprolongation of meningeal sheaths along the inter-costal nerves, and the pressure difference betweenthe spinal subarachnoid space and the pleuralcavity, may also play a part.The frequent (640 ) association with neuro-

fibromatosis (von Recklinghausen's disease) maybe related to the frequency of skeletal, and especiallyvertebral, abnormalities in this condition. Howeverantero-lateral meningocele may occur withoutexternal evidence of neurofibromatosis, and thevertebral changes described may therefore be theonly evidence of the disease.As long as there is no clinical evidence of spinal

cord compression, it should be possible to reach thecorrect diagnosis without resorting to myelography.On diagnosis, follow-up observation should beundertaken, and operation carried out only forrespiratory distress due to a large meningocele, orif there is evidence of rapid progression in size.

We are grateful to Professor Brodie Hughes, Departmentof Neurological Surgery, Queen Elizabeth Hospital,Birmingham, for permission to mention his case.Figure 10 is reproduced by permission of the Editor,The British Journal of Bone and Joint Surgery, andMr. J. C. Scott, Radcliffe Infirmary, Oxford, who usedit in a previous publication.

ADDENDUM

Since compiling the material for this publication,one of us (J.B.M.) has been fortunate in being ableto observe a further case of intrathoracic meningo-cele which illustrates the risks attendant uponoperation for this condition.A 49-year-old man with neurofibromatosis had his

right apical opacity explored by a thoracic surgeon.It proved to be a meningocele and he developed apost-operative cerebrospinal fluid fistula and was

transferred for neurosurgical opinion. Happily itresolved without the need for further surgery.

REFERENCES

Bikfalvi, A. (1964). Zur Frage der intrathorakalen lateralen Meningo-zele. Bruns' Beitr. klin. Chir., 208, 292-304.

Bogedain, W., Carpathios, S. T., and Lawland, F. (1963). Intra-thoracic meningocele. Amer. Rev. resp. Dis., 87, 757-761.

Buono, M. S. del, and Osacar, E. M. (1961). Intrathoracic meningoceleassociation with cutaneous neurofibromatosis. Acta neurochir.(Wien), 9, 561-581.

Byron, F. X., Alling, E. E., and Samson, P. C. (1949). Intrathoracicmeningocele. J. thorac. Surg., 18, 294-303.

Chandler, A., and Herzberger, E. E. (1963). Lateral intrathoracicmeningocele. Case report with preoperative diagnosis. Amer. J.Roentgenol., 90, 1216-1219.

Cross, G. O., Reavis, J. R., and Saunders, W. W. (1949). Lateralintrathoracic meningocele. J. Neurosurg., 6, 423-432.

Hagelstam, L. (1946). On the deformities of the spine in multipleneurofibromatosis (von Recklinghausen). Acta chir. scand., 93,169-193.

Hilton, H. D., and McCarthy, H. H. (1959). Intrathoracic meningo-cele. J. thorac. Surg., 37, 261-268.

Hunt, J. C., and Pugh, D. G. (1961). Skeletal changes in neuro-fibromatosis. Radiology, 76, 1-20.

Hutchin, P., and Mark, J. B. D. (1964). Intrathoracic meningocelenot associated with neurofibromatosis. J. thorac. cardiovasc.Surg., 48, 29-39.

Kent, E. M., Blades, B., Valle, A. R., and Graham, E. A. (1944).Intrathoracic neurogenic tumors. J. thorac. Surg., 13, 116-161.

LaVielle, C. J., and Campbell, D. A. (1958). Neurofibromatosis andintrathoracic meningocele. Radiology, 70, 62-65.

Laws, J. W., and Pallis, C. (1963). Spinal deformities in neuro-fibromatosis. J. Bone Jt Surg., 45B, 674-682.

Moore, B. H. (1941). Some orthopaedic relationships of neuro-fibromatosis. J. Bone Jt Surg., 23, 109-140.

Nanson, E. M. (1957). Thoracic meningocele associated with neuro-fibromatosis. J. thorac. Surg., 33, 650-652.

Pohl, R. (1933). Meningokele im Brustraum unter dem Bilde einesintrathorakalen Rund schattens. Rontgenpraxis, 5, 747-749.

Portigliatti-Barbos, M. (1953). Neurofibromatosi e meningoceletoracico. Friuli mned., 8, 477-484.

Schuller, A., and Uiberall, H. (1938). A case of neurofibromatosisRecklinghausen combined with lateral spinal meningocele.Confin. neurol. (Basel), 1, 312-317.

Sengpiel, G. W., Ruzicka, F. F., and Lodmell, E. A. (1948). Lateralintrathoracic meninigocele. Radiology, 50, 515-520.

Stuber (1949). Doppelseitige seitliche Meningocele thoracalis. Zbl.allg. Path. path. Anat., 85, 275.

Teng, P., and Eastman, P. (1958). Intrathoracic meningocele.Neurology (Mineapp.), 8, 153-156.

Voisin, C., Macquet, V., Wattel, F., Mahieu, P., Leduc, M., andJacob, M. (1964). Les manifestations thoraciques de laneurofibromatose de Recklinghausen. Lille med., 9, Suppl.129-39.

Waterfall, M. C. (1966). Intrathoracic meningocele and neuro-fibromatosis. Proc. roy. Soc. Med., 59, 564-565.

Wilhelm, E. (1954). Meningozele des Brustraumes. Thoraxchirurgie,2, 147-155.

110

Protected by copyright.

on February 5, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.32.2.99 on 1 April 1969. D

ownloaded from