PRIMITIVE NEUROECTODERMAL TUMORS OF T H E BRAIN IN CHILDREN

~ I I C H A E L NOEL. HART, MD, MAJ, MC, USA, AND KENNETH M. EARLE, M D

Predominantly undifferentiated tumors occurring in the cerebrum of young individuals are referred to as “primitive neuroectodermal tumors.” Twenty- three of these cases were collected which had certain common features. Grossly they tended to be cystic and compressed adjacent brain. Microscopically they were malignant-appearing and demonstrated a prominent mesenchymal com- ponent. Sixteen cases showed focal evidence of glial and/or neuronal-appear- ing differentiation. It is suggested that the mesenchyme is also an aspect of tumor differentiation. The biologic behavior of these tumors was malignant, with an average duration of 18 months from onset of symptoms to death.

RIMITIVE NEUROECTODERMAL” OR “NEURO- P epithelial” are names used to describe tumors composed of undifferentiated cells resembling germinal or matrix cells6 of the embryonic neural tube. Implied in both of these names is the capacity of these tumor cells to differentiate along either neuronal or glial lines. Neuroectodermal is the preferable name for our purposes since neuroepithelial implies a covering or lining tissue, and also because there is a specific tumor named neu- roepithelioma. The term neuroectoderm has long been in use to describe the total sub- stance of central nervous system tissue minus the mesenchymal elements of blood vessels and microglia.

In a recent address concerning specific types of primitive neuroectodermal tumors, Rubinstein12 cited the medulloepitheliomas, cerebral neuroblastomas, polar spongioblas- tomas, ependymoblastomas, pineal parenchy- mal tumors, and cerebellar medulloblastomas as tumors that have earned special designa- tions because of their more or less predictable location and histologic appearance.

In contrast to these tumors, there remains a group of largely undifferentiated cerebral neo- plasms occurring in children. These tumors remain a classifier’s enigma because they are uncommon, are sometimes mistaken for meta- static tumors, and often show focal areas ot glial and/or neuronal differentiation in adtli-

tion to a prominent mesenchymal component, thereby adding to the confusion. For these reasons, primitive neuroectodermal tumors are generally called by a variety of names: “neuro- blastoma,” “cerebral medulloblastoma,” “un- differentiated small cell neoplasm,” and “un- classified glioma.” We prefer the designation primitive neuroectodermal tumor because i t allows for a spectrum of histologic appear- ances ranging from completely undifferen- tiated tumors to those showing focal or diffuse areas of neuronal and/or glial differentiation. At the same time, these tumors display certain common clinical, gross, and microscopic fea- tures justifying this grouping. The following 23 cases from the files of the Armed Forces In- stitute of Pathology illustrate these features.

MATERIALS AND METHODS

Twenty-three cases were selected from the files of the Armed Forces Institute of Pathol- ogy with the diagnoses of “primitive neuroec- todermal tumor,” “unclassified glioma,” “cere- bral neuroblastoma,” and “cerebral medullob- lastoma.”

All sections were stained with hematoxylin and eosin. In 11 cases special stains of Nissl, phosphotungstic acid hematoxylin, reticulin, and Bodian or Bielschowsky were performed in order to enhance detection of differentia- t ion.

From the Neuropathology Branch, Armed Forces 1 1 1 - stitrite of Pathology, Washington, D.C. 20306.

T h e opinions or assertions contained herein are the private views of the authors and arc not to he con- strued as official or as reflecting the views of thc De- partment of the iZrmy or the Department of Defense.

Received for publication April 10, 1973.

CLINICAL FEATURES

Fifteen patients were male and eight were female. T h e age range was stillborn to 24 years with the average age being 8.1 years. Present-

890

No. 4 NEUROECTODERMAI. TUMOR - Harl and Earle 89 1

ing signs or symptoms in 18 cases where this information was available included: headache, 9 cases: vomiting, 6 cases; seizure, 5 cases: eye signs, 4 cases: hemiparesis, 4 cases; and hydro- cephalus, 2 cases. Duration of symptoms from onset to diagnosis ranged froni zero (sudden unexpected death in two cases) to 3 years, with the average Iieing 4.8 months. If one case in which symptoms lasted for 3 years is elimi- nated, then the average duration of symptoms was 3.0 months. Fol low-~p until death was obtained in 6 cases with the average duration of symptoms from onset until death being 18 months. Again, i f the same case in which the patient lived 5 years after onset of symptoms is eliminated, then the average duration of symptoms from onset until death was 10 months.

PATHOLOGY

Nine tumors were located in a frontal lobe, 4 in a temporal lobe, 4 in a parietal lobe, 2 in an occipital lobe, 2 in the corpus callosum, 1 at a foramen of Monro, and 1 bilateral in the hemispheres. Relationship of the tumors to ependymal lining, other than the specimen lo- cated at the foramen of Monro, was not stated. Most of the hemispheric tumors ap- peared to arise in the white matter but exact location in most cases was not known. Grossly, 9 tumors were described as being cystic: the others were variously described as hemor- rhagic, soft, lobulated, and well demarcated from surrounding parenchyma. Three tumors were stated to be encapsulated.

Microscopically these tumors were com- posed predominantly of small, undifferen- tiated cells with dark, oval to irregular nuclei and usually without observable cytoplasm (Fig. 1). All tumors were histologically malig- nant using the parameters of cellularity, pleo- morphism, endothelial hyperplasia (Fig. Z), numbers of mitotic figures, and areas of necro- sis (Fig. 3). A mesenchymal component was prominent in 13 cases; 9 cases showed colla- genous trabeculae, and 4 cases showed an al- veolar pattern with the alveolar walls com- posed of reticulin. Focal perivascular arrange- ment of cells was present in 6 cases (Fig. 4) and rosettes (with either central canal or cen- tral tangle) were present in 7 cases (Fig. 5). Mitotic figures were seen readily in 20 cases. Confirming the gross impression, the margins of tumor were noted to be compressing adja-

cent brain in 17 cases where margins of tumor were present in the specimen (Fig. 6). Charac- teristically, a few cells were usually noted in- filtrating adjacent brain beyond the compress- ing margins. Areas of necrosis were present in 15 cases (Fig. 3). Each tumor was carefully screened for areas demonstrating differentia- tion along either glial or neuronal lines (Figs. 7, 8). Spongioblastic appearance, and fine cel- lular and intercellular processes were accepted as evidence of probable glial differentiation. Flexner-Wintersteiner rosettes (so-called “true” rosettes with a central canal) were considered to be evidence of either ependymal or retinal differentiation. Electron microscopy will be necessary in the future to determine whether or not the Flexner-Wintersteiner rosettes show features of photoreceptor cells as demonstrated in some retinoblastomas by Ts’o, Fine, and Zimmermanl3 or whether they show features of ependymal cells as suggested by Rubinstein*Z and observed in some of our cases of ependymoma. Neuronal differentia- tion was acknowledged when two or more of the following criteria were present: increase in cell size, increase in amount of cytoplasm, presence of prominent nucleoli, and presence of Homer-Wright (central fibrillary tangles) rosettes. Whether or not these cells with histo- logic features of neurons are truly neurons or not will require electron microscopy for deter- mination. Zimmerman, Font, and Andersenl5 have shown that cells that appeared to be neu- rons in some of their cases of intraocular med- ulloepithelioma, were in fact rhabdomyoblasts by electron microscopy. Many of their medul- loepitheliomas of the eye showed mixed fea- tures of neuroepithelial, mesenchymal, and undifferentiated elements similar to some of our neoplasms in the brain. It was further noted that in areas showing what appeared to be obvious glial differentiation, the supposed glial fibrils stained positively with PTAH in less than half the cases, with staining in sev- eral cases being equivocal. Three cases showed focal areas of glial differentiation only. Seven cases showed focal areas of neuronal differen- tiation only, and 6 cases showed focal glial dif- ferentiation plus focal neuronal differentia- tion. Another 7 cases were judged as showing no evidence of either glial or neuronal differ- entiation. Four of the latter group were among those cases showing prominent colla- genous trabeculae. There was no correlation between amount or degree of differentiation and the age of the patient.

892 CANCER October 1978 Vol. 32

FIG. 1 (top). Typical histology of primitive neuroectodermal tumor illustrating undifferen- tiated cells and fine stromal network (AFIP Neg. 52-11232: H and E, ~ 3 0 5 ) . FIG. 2 (bot tom). Tu- mor with undifferentiated cellularity and prominent endothelial hyperplasia (AFIP Neg. 72-11231; Hand E, x110).

No. 4 NEUROECTODERMAL TUMOR * Hart and Earle

FIG. 3 ( top) . Undifferentiated pattern with prominent vessels and “pseudopalisading” of tumor cells around areas of necrosis (AFIP Neg. 72-11228; H and E, x80). FIG. 4 (bottom). Focal peri- vascular arrangement of cells in an otherwise undifferentiated tumor (AFIP Neg. 72-1 1229; H and E, x110).

893

894 CANCER October 1973 Vol. 32

FIG. 5 ( top ) . In the center of the picture are islands of darkly staining tumor cells containing rosettes with centtal tangles. T h e surrounding tissue is mesenchymal and stained positively with reticulin stains (AFIP Neg. 72-9843; H and E. ~ 1 0 0 ) . FIG. 6 (bottom). Abrupt margin of tumor. This section also shows trahcculae coursing through tumor (.-ZFIP Neg. 72-9896: H and E, ~ 3 5 ) .

so. 4 NEUROECTODERMAL TUMOR * Hart und Eade

FIG. 7 (top). Focus of enlarged cells with increased cytoplasm and large nuclei with prominent nucleoli suggesting neuronal differentiation. Large cell in center closely resembles a ganglion cell. (AFIP Seg. 52-11771; H and E, x440). FIG. 8 (bottom). Focal spongioblastic-like arrangement of cells with fine cellular processes suggesting glial differentiation (AFIP Neg. 52-115iS; H and E, X3.50).

895

Vol. 32 896 CANCER Orlober 1973

DISCUSSION

We have presented 23 tumors with the com- mon features of (a) predominance in early life, (b) clinical malignancy, (c) Occurrence in the cerebrum (usually hemispheric and deep), (d) grossly cystic and hemorrhagic with sharp bor- ders, (e) microscopically malignant and pre- dominantly undifferentiated with evidence of focal attempts to differentiate along glial and neuronal lines, and (f) prominent mesenchy- ma1 component. It is to be emphasized that not all, but most of these characteristics are present in each tumor, justifying their com- mon grouping. It is also to be recalled that the specific neuroectodermal tumors men- tioned in our introduction as well as malig- nant gliomas display many of the anaplastic features similar to this group of tumors. At what point a tumor should cease to be consid- ered a “primitive neuroectodermal tumor” and named according to its avenue of predom- inant differentiation is arbitrary; for our pur- poses, we have restricted the present series to those tumors appearing at least 90-95% undif- ferentiated.

The appearance of mesenchyme as a promi- nent histologic component of many of these tumors is interesting. We feel that this proba- bly represents a third avenue of differentia- tion open to tumors of neuroectodermal ori- gin. Mesenchyme is often distributed heavily throughout cerebellar medulloblastomas,4 and in our experience is not uncommonly found forming a prominent part of oligodendrog- liomas and glioblastomas.

Most reviews of brain tumors in chil- drenlp3~7,s,9J1 do not list a category such as ours but in general concede that certain spe- cific neuroectodermal tumors such as cerebral neuroblastomas or ependymoblastomas may be predominantly undifferentiated. Willis,ls however, mentions six tumors in his collection entitled “Gliomas of Uncertain Nature,” which he felt were glial but preferred to leave unclassified. Although several of his tumors are quite bizarre, others certainly were very close in their pathology to our primitive neuroectodermal tumors. Bodian2 reported

15 cases of supratentorial medulloblastomas in a series of 129 brain tumors which he termed “differentiating medulloblastomas.” He reported these tumors to be slow growing, however. In a series of 116 gliomas, Cushing6 had 6 cases in his “unclassified” category in children which were atypical or transitional. Some of these may have been what we are describing as primitive neuroectodermal tu- mors.

In a recent analysis of 167 brain tumors oc- curring in children under two years of age, Leestma and Earlel0 listed 19 cases which were difficult to classify because of lack of differen- tiating features. They referred to four of these as “blue tumors”; so named because of their dense, undifferentiated cellular patterns with a strong affinity for hematoxylin. These tu- mors are included in the present series.

It cannot be explained why this tumor oc- curs only in younger individuals. In general, however, many primary cerebral neoplasms of children tend to appear more histologically primitive than their adult counterparts. Cellu- lar rest theories cannot explain why these tu- mors would lie dormant for so long and then become suddenly so aggressive. More plausible is the hypothesis that whatever the carcino- genic agent is, the substrate cells of the childs’ central nervous system respond to this agent differently than do adults.

In our series, contributors’ diagnoses of these tumors were most often neuroblastoma, glioblastoma, undifferentiated tumor (or some variation), cerebral medulloblastoma, malig- nant cellular ependymoma, or astroblastoma, reflecting the desire to place each tumor in a known category for which there is ample justification. We are fully aware that each of these tumors, with slightly more differen- tiation or de-differentiation, might fit into one of these categories. However, once i t is recognized that these largely undifferentiated brain tumors share common biological and pathologic properties, then focal areas of di- vergent differentiation can be tolerated with- out losing sight of the ultimate reason for any classification-prediction of clinical out- come.

REFERENCES 1 . Bailey, P., Buchanan, D. N., and Bucy, P. C.: In-

tracranial Tumors of Infancy and Childhood. Chicago, Univ. of Chicago Press, 1939.

2. Bodian hl., and Lawson, D.: The intracranial neo- plastic diseases of childhood. Br. .J. Surg. 40:368-392,

3. Buchanan, D.: Intracranial tumors in infancy and childhood. Am. J . Surg. 93:935-940, 1957. 4. Chatty, E. M., and Earle, K. M.: Medulloblas-

toma-A report of 201 cases with emphasis on the re- lationship of histologic variants to survival. Cancer

1953. 28:977-983. 1971.

No. 4 NEUROECTODERMAL TUMOR - Hurl and Earle 897

5. Cushiiig, H.: l h e intracranial tumors of preado- lescence. Ani. J. Dis. Child 33:551-584,1927.

6. Fujita, S.: The matrix cell and cytogenesis in the developing central nervous system. J . Conap. Neiirol. 120:3’1-42, 1963.

7. Gross, S. W.: Tumors of the brain in infancy- Clinical and pathologic study. Am. J. Dis. Chifd 48:

8. Koos, W. T., and Miller, M. H.: Intracranial T u - mors of Infants and Children. Stuttgart, Georg Thienie Verlag, 1971.

9. Leavitt, F. H.: Brain tumors in childhood-A Clinico-pathologic study. Ant. J . Med. Sci. 178:229-236, 1929.

10. Leestma, J. E., and Earle, K.M.: Neoplasms of the central nervous system in infants 2 years of age and under-Analysis of 163 cases. Delivered at the Ameri-

739-763, 1934.

can Association of Neuropathologists 48th Annual Meeting. Chicago, Ill., June 9, 1972.

11. Michael, P.: Tumors of Infancy and Childhood. Philadelphia, J. B. Lippincott Co., 1964.

12. Rubinstein, I. J.: Cytogenesis and differentiation of primitive central neuroepithelial tumors. J. Neuro- p t h o l . Exp . Neurol. 31:7-26, 1972.

13. Ts’o, M. 0. M., Fine, B. S., and Zimmerman, L. E.: The Flexner-Wintersteiner rosettes in retinoblas- toma. Arch. Pathol. 88:664-671, 1969.

14. Willis, R. A.: The Pathology of Tumors of Chil- dren. Springfield, Ill., Charles C Thomas, 1962; pp. 31-35.

15.Zimmerman, L. E., Font, R. L., and Andersen, S. R.: Rhabdomyosarcomatous differentiation in malig- nant intraocular medulloepitheliomas. Cancer 30817- 835, 1972.