—  SYMPOSIUM #09  —

What is New in Bone Tumors
Moderators: K. Krishnan Unni and Franco Bertoni

Section 4 - Notochordal and Notochord-Like Lesions: Old and New Concepts

Michael Kyriakos
Washington University School of Medicine
St. Louis, Missouri


"Nothing Quite New is Perfect"
-- Cicero (104-43 BC)

An appreciation of anything new requires a knowledge of what has gone before, and with notochordal lesions this "old stuff" includes information contributed by anatomists, embryologists and pathologists over the past 150 years on the development of the notochord (NC) and its relationship to tumor or tumor-like lesions.

Old Stuff :
The notochord (NC) develops during the third week of gestation [1, 2, 3, 4], as a dorsal rod-like structure composed of compact, small, non-vacuolated epithelial cells. At its cephalic and caudal ends it pursues a convoluted or coiled course, but a rather straight course within the central portion of the future vertebral bodies of the mobile spine. Over several weeks, the cells enlarge, and assume a secretory function, develop multiple clear cytoplasmic vacuoles (physaliphorous cells), become less cohesive, and finally reside loosely within an extracellular myxoid matrix. As the future vertebrae develop through chondrification, the NC degenerates, breaking up into segments, and the cells are "squeezed" into the site of the future intervertebral disc to form the nucleus pulposis. NC tissue is not usually found in the vertebral body beyond 10 weeks of gestation [3, 4, 5, 6], and while NC cells may be present within the intervetebral discs of term infants, they usually are not found after the age of 10 years, although rare isolated examples are reported in middle aged adults [1, 2, 7]. Studies of human embryos and fetuses have demonstrated that intra- and extraosseous ectopic offshoots from the main NC may occur and persist anywhere along the axial path of its development, especially at its caudal and cephalic ends [1, 2, 4].

The first existence of a notochordal remnant in an adult was reported by German workers in the last half of the 19th century, who identified an intradural, pre-pontine, jelly-like mass occasionally found incidentally at autopsy. Although first considered cartilaginous, morphologic and experimental studies eventually showed it to be notochordal in origin and was termed "chordoma." At the turn of the century cases began to be reported of clinically malignant lesions of the axial skeleton composed of multivacuolated cells similar to those of the NC [8]. In 1923, Stewart and Burrow [9] divided chordal lesions into two groups, reserving "chordoma" for the malignant forms, and "ecchordosis physaliphora," (EP) for the usually asymptomatic base of skull lesion [(see historical review by Stewart [8]]. Found in up to 2% of autopsies, EP is usually a few millimeters to 1-2 cm, but those 3.5 cm are reported [10, 11, 12]. It is virtually universally accepted that EP represents a notochordal vestige (remnant; rest) that has undergone hyperplasia from its microscopic structure, accepting the proposition that notochordal tissue may be both benign and proliferative.

Although not normally found in post-natal vertebrae, rare examples of NC remnants were mentioned as occurring in adult vertebrae [1, 2, 13], but not histologically demonstrated until 1952 when Congdon [14] illustrated notochordal tissue involving the coccyx that apparently arose from an intervertebral disc. The first clearly described and illustrated example of benign appearing chordal tissue within an adult vertebral body was reported by Ulich and Mirra [15] in 1982 as an incidental microscopic autopsy finding, consisting of two nodules of physaliphorous and signet-ring cells admixed with islands of cartilage. They found a similar incidental "notochordal rest" within a surgical specimen of the coccyx. They termed these remnants "ecchordosis physaliphora vertebralis."

In view of the studies that demonstrated ectopic NC remnants in fetuses and embryos, it was quickly postulated that chordoma arose from malignant transformation of these embryonic rests retained in the adult [8]. Virtually all articles on chordoma still pay homage to this idea, only rarely is it suggested that this tumor might arise from other tissue elements [16, 17]. Data indicating that chordoma most frequently arose in the sacrococcygeal and sphenooccipital regions, sites where ectopic NC rests are most often found in fetuses, supported this view, as did the reported similarities between chordoma, NC and EP cells, in their light, electron microscopic and immunohistochemical features [18]. The idea that adult tumors arose through the transformation of embryonic rests was advocated by Cohnheim in 1877 [19], and although this view has long since been rejected, chordoma remains the only major human malignancy still believed to be derived from embryonic rests. It is of some interest that the intravertebral disc, the site where NC cells are most often found post-natally, has never been shown to be the site of origin of chordoma. Recent epidemiologic data from the SEER program [20] on chordoma in the United States, found the tumor location to be uniformly distributed throughout the axial skeleton, in sharp contrast to the reported cephalic and caudal preponderance sited to support the NC remnant origin view. The reason for this discrepancy is unclear.

Dabska [21] reported a peripheral soft tissue tumor with light microscopic features that superficially resembled chordoma and which was designated as a parachordoma. Although subsequently shown to differ from true chordoma (86,95), rare tumors of the peripheral soft tissues and long bones have been described having all of the light microscopic, immunohistochemical, and electron microscopic features of true chordoma, and designated as chordoma periphericum [23]. Although suggested to arise from ectopic NC cells drawn from their axial position during the migration of developing embryonic structures, NC cells have never been described in an abaxial location. Koh et al. [24] suggested their possible origin from totipotential mesenchymal cells. If one accepts this latter proposition for peripheral chordoma, it raises the possibility that notochordal remnants (rests) need not be responsible for the origin of all axial chordomas.

Conventional chordoma, which accounts for <5% of malignant bone tumors, is histologically characterized by its lobulation, and an abundant mucoid matrix in which are distributed nests, sheets, and syncytial cords of epithelioid cells, many containing single or multiple clear cytoplasmic vacuoles that form the "bubbly" cytoplasm of physaliphorous cells. The cells show some degree of nuclear atypia, at times marked; mitotic figures are usually not numerous; and areas of necrosis may be present. Marked variability in cellularity and cell composition occurs from field to field throughout the tumor. The cells are immunoreactive for a variety of epithelial markers, vimentin, S-100 protein, and carcinoembryonic antigen. At diagnosis, vertebral chordoma invariably shows radiologic evidence of bone destruction, and almost always an associated soft tissue mass [18]. Roentgenograms and computed tomographic (CT) studies usually show lytic bone destruction that may be associated with amorphous calcification. Sclerotic change may be extensive, but only rarely is a totally "ivory" vertebral body produced. Magnetic resonance imaging (MRI) studies show low to intermediate signal intensity T1-weighted images, and high signal intensity T2-weighted images, and most often show enhancement with contrast agents.

New Stuff :
The concept that retained benign notochordal tissue might grow to produce a symptomatic benign vertebral lesion, was first advanced by Joseph Mirra at the 1996 meeting of the International Skeletal Society (ISS) during discussion of a case, presented by others, of an adult man with a 3.0 cm, painful, lumbar vertebral body lesion; confined to bone without bone destruction or a soft tissue mass.. Histologically, the lesion was composed of a uniform sheet-like proliferation of bland unvacuolated and multivacuolated chordal-type cells which, on low power examination, simulated the appearance of fatty marrow. Immunostains were reactive for epithelial markers and S-100 protein. On the basis of having seen other non-destructive vertebral body lesions with a similar histology, and which did not clinically progress, Mirra suggested a diagnosis of giant notochordal rest (GNR). When this case was subsequently published [25], the authors rejected this view, and considered the lesion an "early stage" chordoma, despite the patient's 10-year history of back pain. At the 1998 and 1999 ISS meetings four additional cases [18, 26, 27] of symptomatic non-destructive vertebral body lesions having identical histologic features were presented – three adults, two of whom were Mirra's original patients [27], and one in a child [18]. In their 1998 textbook, Dorfman and Czerniak illustrated the MRI images of a similar lumbar vertebral body lesion, in an adult woman, which they termed a notochord rest [28]. Mirra, based on criticism that the lesions were too large to be considered a "rest," changed the diagnosis to "giant notochordal hamartoma" in his published report [27], although the formal definition of an embryonic rest does not include a size limit.

An additional 8 similar cases have been published [29, 30, 31, 32, 33]. The 14 patients, 8 male and 6 female, ranged in age from 11 ˝ to 57 years, with a mean of 39.5 years (median 41.5 years). All but one were symptomatic with back pain, at times severe, present for from 2 months to 20 years. In some, the pain appeared not to be related to the specific vertebral lesion. In the asymptomatic patient the lesion was an incidental radiologic finding. Twelve patients had a single involved vertebra; one had two; and another had three, with lumbar vertebrae involved in 7, thoracic in 1, cervical in 3, sacral in 4, and the coccyx in 2. Roentgenograms, described in 10 patients, were normal in 9, and in one the affected vertebral bodies were totally sclerotic, but with a normal contour and size. Pre-biopsy bone scans were normal in 9, while 2 showed a "cold spot" in the vertebral body. In 12 patients with pre-biopsy CT studies, a variable degree of vertebral body sclerosis was present in 8, while 4 showed no abnormality; no case showed evidence of bone destruction. MRI studies all showed a lesion that partially or totally replaced the vertebral body, without extraosseous extension. In all but two cases, the lesion had low T1- and high T2-weighted signal intensity images. The lesions ranged from 0.7 to 3.0cm, to those that completely filled the vertebral body.

Treatment consisted only of biopsy in 4 patients; total vertebrectomy/resection in 7, and needle biopsy with cement injection and curettage with bone graft in one patient each, and one patient died interoperatively. On follow-up, all of the remaining patients were alive, without recurrence, from 10 months to 9 years, including those in whom only biopsy was done; in none followed with radiologic studies has progression of the lesion occurred.

Histologically [18, 27, 34], GNR is similar to EP, lacking lobularity, a mucinous matrix, syncytial cell cords, significant nuclear atypia, mitotic activity, or necrosis. Rather, a uniform, sheet-like proliferation of compact cells, with granular to clear univacuolated and multivacuolated cytoplasm, permeates between existing bone trabeculae. Cell nuclei are uniform, round to oval, and dense. Cell borders may appear thickened, and at low power magnification, simulate mature fat cells. The interface with the marrow is sharp, GNR having a pushing border. Foci of marrow elements may be entrapped within the lesion, and the bone trabeculae thickened by appositional new bone or cores of residual cartilage. The cells are immunoreactive for cytokeratin, EMA and S-100 protein [18, 25, 27, 29, 31, 32, 33] and non-reactive for p53 and Ki-67 [18, 27, 32]. Needle biopsy tissue may show only physaliphorous cells, and in such cases the diagnosis of GNR, rather than chordoma, is greatly dependent upon the radiologic findings mentioned above, with no evidence of bone destruction or an extraosseous mass. The radiologic features are also static, unlike chordoma where they always show progressive change with bone destruction. The lack of enhancement of the lesion following the use of contrast agents would also argue against a diagnosis of chordoma. The relationship, if any, of GNR to what Rosenberg et al. [35] term "lipoid chordoma," is unclear.

Yamaguchi and colleagues [29, 34, 36, 37], have recently done extensive and careful studies in the Japanese population, on benign notochordal-type lesions of the axial skeleton, and their relationship to chordoma. Using autopsy spinal columns and clival biopsy cores, they found that at least 20% of the cases (14% of the vertebral specimens) contained microscopic foci of what they termed "benign notochordal cell tumor" (BNCT), whose histomorphology is identical to the symptomatic macroscopic lesions described above as GNR [34, 36]. These "tumors," however, only ranged from 1mm to 13mm, with all the vertebral foci less than 1.0 cm (median 2 mm) and over 50% only 1-2mm in maximum size. The rationale given for designating such small lesions as tumors was that they differed morphologically from the notochordal vestiges within the intervertebral discs of fetuses where the NC cells are in syncytial cords within a myxoid stroma [36]. However, the gestational ages of the four fetal cases examined ranged from 21 to 35 weeks, a time when NC tissue is well along not only in its secretory phase, but in its involution, and its morphology is not the same as in its earlier intravertebral phase when it begins as a compact group of small cells without a myxoid stroma. Since BNCT(s) lacked a myxoid stroma, it was concluded that they were not NC remnants, but true neoplasms since sclerotic change in the adjacent bone indicated lesional growth [34, 36]. Since the location of BNCT(s} overlapped that of chordoma, the authors also concluded that they were the probable precursors of chordoma. They further claimed that ecchordoses physalifera, which are histologically similar to BNCT(s), are therefore also not notochord remnants [29]. The authors also based some of their conclusions on the work of Naka et al. [38], in Japan , who found that chordoma cells were reactive for CK18, but fetal NC cells were not. Yamguchi et al. [34] demonstrated that the cells of BNCT were also reactive for CK18, but the NC cells in the fetal intervertebral discs were not, and thus again considered BNCT not to be notochordal. However, the specimens studied by Naka et al. had gestational ages of 17 to 39 weeks, and those of Yamaguchi et al., 21 to 35 weeks, stages that are relatively late in notochord degeneration. In addition, other investigators have demonstrated CK18 positivity in the NC cells of both embryos and fetuses [6, 39], bringing into question the conclusions drawn about the nature of BNCT.

Yamaguchi and co-workers also reported what they claimed to be the first histologically confirmed case of a chordoma arising from a BNCT [29], the chordomatous tissue being present adjacent to small nests of benign appearing physaliphorous cells with a sharp interface between them. However, in a subsequent report [37] they conceded that they could not exclude the possibility that a de novo chordoma had developed "coincidentally" in the vicinity of a pre-existing BNCT. Similar findings were reported by Deshpande et al. [40] who, in a histologic examination of sacrectomy specimens removed for chordoma, found similar foci of benign notochordal tissue in 8.5%. These aggregates (1 to 20 mm) were "adjacent" to the chordoma, but separate from it. The authors believed these findings supported BNCT as the precursor lesion of chordoma.

Finally, Yamaguchi et al. [37] described two autopsy cases in which incidental "miniature," and "incipient" extraosseous coccygeal chordomas, 17 mm and 20 mm , were found that seemed to invade the soft tissue and had microscopic features "reminiscent" of classic chordoma, but lacked lobularity or nuclear aytpia. Several separate intraosseous foci of solid benign notochordal tissue were also present, two of which had an adjacent myxoid component containing strands and cords of cells similar to the extraosseous lesions. The authors indicated that these "incipient" chordomas, and their relationship to the areas of BNCT, supported the view that chordoma arises from BNCT.

The work of Yamaguchi and co-workers has shed new light on the prevalence of axial chordal lesions. Their high prevalence rate for "BNCT" must be taken as minimal, as complete sectioning of the entire axial skeleton was not done so that additional microscopic foci could well have been missed. However, the specimens were from a uniform Japanese population, and whether examination of such material from other ethnic populations would yield similar results is open to question, as there are marked differences in the prevalence of chordoma in white and non-white populations [20]. Some of the data and conclusions drawn by these authors also require some reservation. Nowhere is the concept entertained that what they call neoplastic BNCT, and which is histologically identical to GNR, might actually represent simple hyperplasia of retained notochordal tissue. The cell of origin of "BNCT" is not discussed other than to state that it is not notochordal, and no discussion is given to the fate of the ectopic axial NC remnants noted by embryologists – do they never occur in the adult? The use of involuting nucleus pulposis tissue as the morphologinc and immunohistochemical benchmark for all NC tissue is challengable, as is the use of cytokeratin CK18 as a means of distinguishing NC tissue from both "BNCT" and chordoma, and by which they reject the existence of remnant intraosseous or extraosseous NC tissue EP. Finally, the average age of their patients with BNCT was similar to that of patients with chordoma [36]. If these lesions are the precursors of chordoma, one might expect them to be found in a younger population. We do not hold the view that the notochordal lesions described here as GNR ("BNCT") never undergo malignant transformation. If one accepts the conclusions of Yamaguchi et al., drawn on the basis of their "incipient" chordoma case, it is possible that malignant transformation of these lesions might occur, although if their microscopic prevalence in the axial skeleton is proven to be universally true, such an occurrence must be quite rare.

Patients whose lesions are symptomatic may require surgical resection for relief. If discovered as an incidental radiologic finding, a biopsy to confirm the diagnosis is required, and these patients could be followed with periodic radiologic examinations to insure that no progression occurs. Whether we are correct that what we designate as a GNR is not an early stage chordoma as advocated by Darby et al, [25] remains an open question, as is the origin of chordoma. Only long term follow-up of additional patients, who have not been treated, will provide answers to these questions, and establish whether what we now believe is a new idea, is really true.

References
  1. Horwitz T. Chordal ectopia and its possible relation to chordoma. Arch Pathol 1941;31:354.

  2. Horwitz T. The human notochord. A study of its development and regression, variations and pathologic derivative, chordoma. Private printing, Indianapolis , IN , 1977.

  3. Steegmann AT. Notochordal tissue: its development and reactions. In: Minckler J, ed. Pathology of the Nervous System, Vol. 1, New York , NY : McGraw-Hill Book Co., 1968.

  4. Willis RA. The Borderland of Embryology and Pathology, 2nd ed. London : Butterworths & Co., 1962

  5. Goto S, Uhthoff HK. Notochord action on spinal development. Acta Orthop Scand 1985;57:85.

  6. Götz W, Kasper M, Fischer G, et al. Intermediate filament typing of the human embryonic and fetal notochord. Cell Tissue Res 1995;280:455.

  7. Trout JJ, Buckwalter JA, Moore KC, et al. Ultrastructure of the human intervertebral disc. I. Changes in notochordal cells with age. Tissue Cell 1982;14:359.

  8. Stewart MJ. Malignant sacrococcygeal chordoma. J Pathol Bacteriol 1922;25:40.

  9. Stewart MJ, Burrow J le F. Ecchordosis physaliphora spheno-occipitalis. J Neurol Psychopathol 1923;4:218.

  10. Ho K-L. Ecchordosis physaliphora and chordoma: a comparative ultrastructural study. Clin Neuropathol 1985;4:77.

  11. g S-H, Ko S-F, Wan Y-L, et al. Case report. Cervical ecchordosis physaliphora: CT and MR features. Br J Radiol 1998;71:329.

  12. Salisbury JR. The pathology of the human notochord. J Pathol 1993;171:253.

  13. Beadle, OA. The intervertebral discs. Observations on their normal and morbid anatomy in relation to certain spinal deformities. His Majesty's Stationery Office, Medical Research Council, Special Report Series No. 161, 1931.

  14. Congdon CC. Benign and malignant chordomas. A clinico-anatomical study of twenty-two cases. Am J Pathol 1952;28:793.

  15. Ulich TR, Mirra JM. Ecchordosis physaliphora vertebralis. Clin Orthop Rel Res 1982;163:282.

  16. Birrell, JHW. Chordomata. A review of nineteen cases of chordomata including five vertebral cases. Aust NZ J Surg 1953;22:258.

  17. Bottles K, Beckstead JH. Enzyme histochemical characterization of chordomas. Am J Surg Pathol 1984;8:443.

  18. Kyriakos M, Totty WG, Lenke LG. Giant vertebral notochordal rest. A lesion distinct from chordoma: Discussion of an evolving concept. Am J Surg Pathol 2003;27:396.

  19. Hajdu SI. Thoughts about the cause of cancer. Cancer 2006;106:1643.

  20. McMaster ML, Goldstein AM, Bromley CM, et al. Chordoma: incidence and survival patterns in the United States, 1973-1995. Cancer Causes Control 2001;12:1.

  21. Dabska M. Parachordoma. A new clinicopathologic entity. Cancer 1977;40:1586.

  22. Folpe AL, Agoff SN, Willis J, et al.. Parachordoma is immunohistochemically and cytogenetically distinct from axial chordoma and extraskeletal myxoid chondrosaroma. Am J Surg Pathol 1999;23:1059.

  23. Nielsen GP, Mangham DC, Grimer RJ, et al. Chordoma periphericum. A case report. Am J Surg Pathol 2001;25:263.

  24. Koh JS, Chung JH, Lee SY, et al. Parachordoma of the tibia: Report of a rare case. Pathol Res Pract 2000;196:269.

  25. Darby AJ, Cassar-Pullicino VN, McCall IW, et al. Vertebral intra-osseous chordoma or giant notochordal rest? Skeletal Radiol 1999;28:342.

  26. McCarthy E. Case 35, International Skeletal Society Meeting, Dublin , Ireland , 1998.

  27. Mirra JM, Brien EW. Giant notochordal hamartoma of intraosseous origin: a newly reported benign entity to be distinguished from chordoma. Report of two cases. Skeletal Radiol 2001;30:698.

  28. Dorfman HD, Czerniak B. Bone Tumors, St. Louis, MO: Mosby, Inc., 1998: 974.

  29. Yamaguchi T, Yamato M, Saotome K. First histologically confirmed case of a classic chordoma arising in a precursor benign notochordal lesion: differential diagnosis of benign and malignant notochordal lesions. Skeletal Radiol 2002;31:413.

  30. Murakami H, Kawahara N, Gabata T, et al. Vertebral body osteonecrosis without vertebral collapse. Spine 2003;28:E323.

  31. Murakami H, Kawahara N, Tomita K. Letter to the Editor. Spine 2005;30:593.

  32. Knapik JA, Vlasak R, Reith JD. Notochordal hamartoma and its necessary distinction from chordoma (abstract). Mod Pathol 2003;16:15A

  33. Chauvel A, Taillat F, Gille O, et al. Giant vertebral notochordal rest: A new entity distinct from chordoma. Histopathology 2005;47:646.

  34. Yamaguchi T, Suzuki S, Ishiiwa H, et al. Benign notochordal cell tumors. A comparative histological study of benign notochordal tumors, classic chordomas, and notochordal vestiges of fetal intervertebral discs. Am J Surg Pathol 2004;28:756.

  35. Rosenberg AE, Nielsen GP, Keel SB Lipoid chordoma. Mod Pathol 2000;13:15A.

  36. Yamaguchi T, Suzuki S, Ishiiwa H, et al. Intraosseous benign notochordal cell tumours: overlooked precursors of classic chordomas? Histopathology 2004;44:597.

  37. Yamaguchi T, Watanabe-Ishiiwa H, Suzuki S, et al. Incipient chordoma: a report of two cases of early-stage chordoma arising from benign notochordal cell tumors. Mod Pathol 2005;18:1005.

  38. Naka T, Iwamoto Y, Shinohara N, et al. Cytokeratin subtyping in chordomas and the fetal notochord : An immunohistochemical analysis of aberrant expression. Mod Pathol 1997;10:545.

  39. Stosiek P, Kasper M, Karsten U. Expression of cytokeratin and vimentin in nucleus pulposus cells. Differentiation 1988;39:78.

  40. Deshpande V, Nielsen GP, Mankin HJ, et al. Intraosseous benign notochord cell tumors (BNCT): Further evidence supporting a relationship to chordoma (abstract). Mod Pathol 2006;19 (suppl):10A