chordoma and chondroid neoplasms of the spheno-occiput: an immunohistochemical study of 41 cases...

2943

Chordoma and Chondroid Neoplasms of the Spheno-Occiput An Immunohistochemical Study of 41 Cases with Prognostic and Nosologic Implications Andrew Mitchell, M.D.,* Bernd W. Scheithauer, M.D.,* K. Krishnan Unni, M.D.,* P. J . Forsyth, M.D.,t Lester E. Wold, M.D.,” and D. J . McGivney, B.S.

Background. Chordomas are rare neoplasms that show a proclivity for the spheno-occiput and sacral regions. A “chondroid” variant involving the spheno-occiput has been associated with improved survival. “Clas- sic” or nonchondroid chordomas are uniformly immunoreactive for keratins. Chondroid chordomas are said to be immunonegative for epithelial markers, a feature used to support the concept that they represent chondrosarcomas.

Methods. The authors performed immunohistochemical studies on 25 patients with chondroid chordoma (mean age, 40.0 years) and on 16 patients with classic chordomas (mean age, 44.2 years) to establish tumor subsets based upon immunophenotype, specifically reactivity for epithelial markers. Kaplan-Meier survival curves were then constructed for each group with age as an added variable.

Results. All classic chordomas reacted for keratins as did 8 (32%) of the 25 chondroid chordomas. Forty-four percent of classic and 85% of chondroid chordomas were positive for S-100 protein. At 5 years, all patients younger than 40 years of age were alive in both the classic and chondroid groups. In contrast, of patients older than 40 years of age, only 22% with classic chordomas and 38% with chondroid chordomas were alive.

Conclusions. Regardless of tumor subtype, age is the single most important variable in determining survival; patients younger than 40 years of age do better than older patients. There are no significant survival differences be- tween patients with cartilage-containing tumors that are keratin immunopositive (“true” chondroid chordoma) or negative (chondrosarcoma). Immunostaining for keratins is of no prognostic value in assessing chondroid lesions of the spheno-occiput. Cancer 1993; 722943-9.

From the Departments of *Laboratory Medicine and Pathology and tNeurology, Mayo Clinic and Mayo Foundation, Rochester, Min- nesota.

Address for reprints: Bernd W. Scheithauer, M.D., Department of Pathology, Mayo Clinic, 200 First Street, S.W., Rochester, MN 55905.

Accepted for publication June 22, 1993.

Key words: chordoma, chondroid chordoma, chondrosarcoma, immunohistochemistry, prognosis.

Chordomas are rare neoplasms of presumed notochor- dal origin that arise along the vertebral axis and show a proclivity for the spheno-occipital and sacral regions. ’,’ Slow growing with a pronounced tendency to recur de- spite surgical resection and/or radiation therapy, they usually result in death of the patient in months to years.3

The gross appearance is of a lobulated, gelatinous tumor that typically infiltrates bone but may grossly appear somewhat demarcated. Histologically, chordomas exhibit a variable mix of sheets and cords of round to polygonal cells with eosinophilic cytoplasm, variably abundant cytoplasmic mucin that, when abundant, forms ”physaliphorous” cells and extracellular mucus pools. Fibrous septae parcel the cells and matrix into lobules. Cytologic atypia is absent or minimal.’-4

A subpopulation of what have been considered chordomas arises largely in the spheno-occiput and ex- hibits cartilaginous differentiation (Fig. 1). Such tumors have been ascribed a more indolent clinical course.’ The extent to which cartilage is present varies from minimal to extensive. Though such tumors have been termed ”chondroid’ chordomas to distinguish them from “classic” chondromas, some authors dispute the existence of chondroid chordomas, preferring to regard these cartilage-containing neoplasms as chondrosar- comas5

A defining feature of classic chordomas is their im- munopositivity for markers of epithelial differentiation, specifically In most series, virtually all tumors are keratin immunopositive; reactivity for epithelial membrane antigen is common, whereas carcinoembryonic antigen staining is less frequently seen .6,7,10

2944 CANCER November 15, 1993, Volume 72, No. 10

drosarcomas, defined herein as tumors lacking epithelial features.

Materials and Methods

A total of 41 cases of spheno-occipital chordoma, both classic and chondroid, arising from the midline exclu- sively, were obtained for study: 29 from Mayo Clinic patients and 12 from our consultation files. Fifteen cases had been included in the previously published clinicopathologic study of Forsyth et al.; the remainder of our cases were obtained separately for the purposes of this report. The other 36 cases used in the paper by Forsyth were not used in this ~ t u d y . ' ~ Each case was reviewed by three pathologists (B.W.S., K.K.U., A.M.) to confirm the diagnosis of chordoma and to assess the presence or absence of chondroid differentiation. Only those neoplasms with myxomatous change were classi-

Figure 1. Representative chondroid chordoma demonstrating typical chordomatous (right) and chondroid (left) regions.

Although there are reports of individual cases of chondroid chordoma expressing keratin reacti~ity,~,~, '" chondroid chordomas and, particularly, chondrosarcomas have traditionally been regarded as immunonegative for epithelial marker^.^,^ Both classic and chondroid chordomas usually exhibit staining for S-100 protein.6

Efforts have been made to define clinical parameters that may portend a more favorable prognosis for spheno-occipital chordomas. In a recent separate study from our institution dealing with clinical parameters and their potential prognostic ~ignificance,'~ 5 1 cases of spheno-occipital chordomas were reviewed with regard to the following features: (1) age at presentation; (2) signs and symptoms, including duration of symptoms, presence or absence of headache, diplopia, papille- dema, and/or extraocular muscle palsy; (3) pathologic features of the neoplasm, including the presence or absence of chondroid differentiation, mitoses, atypia, and/or necrosis; and (4) treatment, i.e., surgical resection and/or radiation therapy. Of these variables, we concluded that only young patient age, defined as younger than 40 years, was an independent, favorable prognostic indicator. The observed better survival of patients with chondroid chordomas was attributed to the younger age of patients in that group; those of similar age with classic and chondroid chordomas had simi-

fied as classic chordomas, and only those tumors with clear-cut cartilaginous differentiation were considered chondroid. The review yielded 16 classic and 25 chondroid chordomas, with follow-ups ranging from 90 days-1 7 years, 2 months (mean, 4.92 years). This series of cases was purposely weighted toward chondroid tumors, accounting for their greater numbers than classic chordomas. Our aim was to study sufficient chondroid neoplasms to allow potential subpopulations to be identified. Additionally, 12 cases of low-grade chondrosarcoma of extracranial bones were included for immunohistochemical comparison.

Immunohistochemical staining was performed in each case using antiserums directed toward the following: keratins (AE1-AE3) (Boehringer Mannheim Corp., Indianapolis, IN; monoclonal, dilution of 1: 75), the low-molecular weight keratin CAM 5.2 (Becton Dickin- son; Mountain View, CA; monoclonal, dilution of 1:25), the low-molecular weight keratin MAK 6 (Triton Diag- nostics, Alameda, CA; monoclonal, dilution of l : l O ) , epithelial membrane antigen, carcinoembryonic antigen (Dako, Carpinteria, CA; polyclonal, dilution of 1:800), 5-100 protein (HSC Research Development Corp., Toronto, Canada; polyclonal, dilution of 1:800), vimentin (Dako; monoclonal, dilution of 1: lo), and glial fibrillary acid protein (Dako; polyclonal, dilution of

lar prognoses. The purpose of this study was to examine the im-

munophenotype and flow cytometric characteristics of both classic chordoma and cartilage-containing tumors to identify subsets associated with a more favorable

'. Age and Sex Spheno-occipital Chordomas

Of 41

No. of patients prognosis-and to examine the relationship of true chon- ~~$"ema Male Female M.F ratio Average (yr) age

droid chordoma, i.e., lesions exhibiting an eDithelia1 " immunoprofile, to histologically similar tumors' lacking Classic 16 9 7 1:3:1 44.2 epithelial markers. The latter may be considered chon- Chondroid 25 16 9 1:8:1 40.0

Skull Base Chordomas and Chondroid Neoplasms/Mitchell et at. 2945

Table 2. Immunoprofiles of 41 Spheno-occipital Chordomas

Chordoma Keratin membrane Carcinoembryonic s-100 Epitheluial

subtvve No. (AEI-AE3) (O/O) antigen (Yo) antigen (YO) protein (%) Vimentin (%)

C 1 ass i c 16 16 (100) 14 (88) Chondroid 25 8 (32) 6 (24)

7 (44) 15 (94) 21 (84) 22 (92)

1 :300). Overall, immunoreactivity was considered as positive or negative in classic chordomas, whereas in chondroid tumors, reactivity was similarly assessed in both chordomatous and/or chondroid areas.

Flow cytometry for ploidy was performed on 14 classic and 16 chondroid cases; in the remaining 11 cases, insufficient tissue precluded flow cytometric study. The block from which histologic sections were prepared was used for flow cytometry. Cell nuclei were extracted from the block by the modified technique of Hedley et al.I4 Nuclei were stained for DNA content by modification of the method of Vindelov et al.15 Modifi- cations included the hydration of tissue for 2 1 / 2 hours, nuclear extraction with 10-minute trypsinization, and 60-minute RNase treatment. Sonication was performed upon the samples as described in a previous publication from our laboratory.'6

Using the resulting immunohistochemical profiles, overall survival and disease-free survival distributions were estimated with Kaplan-Meier curves for the following tumor subsets: (1) classic chordomas, ( 2 ) keratin-immunopositive chondroid chordomas (true chondroid chordomas), and (3) keratin-immunonegative chondroid chordomas (chondrosarcomas). Kaplan- Meier survival curves were also constructed for these tumor subsets with age at diagnosis, i.e., younger than or older than 40 years, as an independent factor. The

log-rank test was used to determine if survival differences in each group were statistically significant.

The therapies of the patients in our series did not vary based on histology or age; similar numbers of patients were debulked or radiated. Therefore, Kaplan- Meier survival curves based on treatment consider- ations were not constructed.

Results

Table 1 lists the profiles (sex and age) of the 16 classic and 25 chondroid chordomas. As demonstrated in pre-

Figure 3. Chondroid area of case illustrated in Figure 1 demonstrating (top) keratin positivity and (bottom) epithelial membrane antigen positivity.

Figure 2. Chordomatous area of case illustrated in Figure 1 demonstrating keratin positivity.


100

80-

8 d, 60 C > 5 4 0 -

v)

.- .-

2o

t

7 (lo)

I Chondroid (n-25) (3)

-

Classical ( n = 16) (11 - P=NS

6, 60 C >

f 40 0

2o

.- .-

0 1 I I I I I 0 1 I 1 I I I 0 2 4 6 8 10 0 2 4 6 0 10

Years Years

% Epithelial marker negative (n= 18) -

(2)

(2)

(9) -

Epithelial marker positive (n = 23)

- P=NS

100

Epithelial marker positive (n = 9)

(1) 80 -

2o

Epithelial marker negative (n= 16)

(21

Classic ( 1 )

I I I I I

- P=NS - P=NS chordoma, n = 16'

I I I I I

Chondmid chordoma

vious studies, men predominate in both groups but es- pecially in the chondroid category. The latter also oc- curred in younger patients (mean age, 4.2 years younger) than did classic chordomas.

The results of immunohistochemical staining are listed in Table 2. As expected, all classic chordomas were keratin-immunopositive; in this group, variable positivity for epithelial membrane antigen (88%) and carcinoembryonic antigen (38%) was present. Of greater interest was the observation that 8 (32%) of 25 chondroid chordomas exhibited a positive keratin reac- tion, both in the chordoid and chondroid areas (Figs. 2 and 3, top). Of these tumors, six (24%) were also reactive for epithelial membrane antigen (Fig. 3, bottom) and carcinoembryonic antigen. In both of these chordoma subgroups, positivity for epithelial membrane antigen and carcinoembryonic antigen did not occur in the absence of keratin staining.

Only 7 (44%) of 16 classic chordomas were S-100 protein positive, a lower rate of positivity than reported in other series. Twenty-one of 25 chondroid chordomas were S-100 protein positive, an expected finding. Of the 12 control cases of chondrosarcoma, all were 5-100 protein immunoreactive but lacked stains for the full spectrum of epithelial markers.

Kaplan-Meier survival curves constructed on the basis of immunophenotype revealed the following: survival rates of patients with chondroid chordoma were not significantly greater than for those with classic chordomas (Fig. 4, top left), and survival rates of patients with epithelial marker-positive chordomas (classic and chondroid) versus epithelial marker-negative chordomas (classic and chondroid) were nearly identical (Fig. 4, top right). Of greater interest, survival rates of patients with epithelial marker-positive chondroid chordomas versus epithelial marker-negative chondroid chordomas, ostensibly chondrosarcomas, were nearly identical (Fig. 4, bottom left). Figure 4 (bottom right) demonstrates the survival differences among the three subgroups of classic chordoma, epithelial marker- positive chondroid neoplasms, and epithelial marker- negative chondroid neoplasms. Again, the survival differences are not statistically significant.

Age as an independent variable had a definite im- pact on survival rates, one overriding the effects of histology or immunotype. Separating all chordoma patients, both those with classic chordomas and all chondroid neoplasms, into a "young" (I 40 years) and an "old" (> 40 years) group at diagnosis revealed the following: the survival of young patients was significantly

Skull Base Chordomas and Chondroid Neoplasms/Mitchell et al.

(3)

2947

(1)

5 " v)

'11, '-77%

7 All chordomas <40 years old (n= 18) (l)

All chordornas 240 years old (n = 23)

:0.001 (3)

I I I I I

0 2 4 6 8 10

Years 100

80

8 d, 60 c > 5 40

0 20

.- .-

1 Classical <40 years old (n = 5)

I 'I

I

0 2 4 6 8 10

Years

better than that of those over 40 years of age, regardless of tumor subtype ( P = 0.001) (Fig. 5, top left). This difference was noted in the chondroid subsets ( P =

0.01) (Fig. 5 , top right) but not in the subset of classic chordomas (P = 0.11) (Fig. 5, bottom left). It is possible that the survival differences in the latter group may lack significance due to the small number of classic chordomas, 5 of 16, occurring in patients younger than 40 years of age.

The results of flow cytometric ploidy studies are listed in Table 3. Eleven of the 30 neoplasms were un- classifiable or had insufficient nuclei for assessment. The remaining 19 were predominantly diploid. Three classic chordomas were aneuploid. One example of both a classic and a chondroid chordoma was tetraploid .

Discussion

In 1973, Heffelfinger et al. were the first to describe chondroid chordoma,' noting a more favorable clinical

Table 3. Summary of Flow Cytometric Data

Unclassified/ Chordoma insufficient subtype No. Diploid Aneuploid Tetraploid nuclei

Classic 14 6 3 1 4 Chondroid 16 8 0 1 7

100

8 8ot L (1) Chondroids ~ 4 0 years old (n = 13)

Chondroids 240 years old (n= 12)

(2) (2)

2o 1 P=O.O1

01 I I I I I 0 2 4 6 8 10

Years

Figure 5. Kaplan-Meier survival curves demonstrating that young age (5 40 years) at presentation is a statistically significant variable for improved survival in the following groups: (top left) all chordomas, classic and chondroid; and (top right) chondroid chordomas. The statistical significance of young age is not demonstrable in the classical chordoma group (bottom left), although a trend is present.

outcome in association with what they considered a unique chordoma subtype. In that series, patients with chondroid chordoma survived an average of 15.8 years from the time of diagnosis, as compared to 4.1 years for patients with classic chordomas, a degree of survival difference never reported subsequently. The authors also noted the tendency of chondroid chordoma to occur in a younger age group (mean age, 35 years) than classic chordoma (mean age, 42 years).

Some authors dispute the concept of chondroid chordoma, preferring to regard such cartilage-containing neoplasms of the skull as low-grade chondrosarcomas.' A basis of arguments both for and against the existence of chondroid chordoma has been the results of immunostaining studies for epithelial markers (Table 4). The immunoprofiles of classic chordoma and chondrosarcoma are not disputed. The former stain for keratin and often other epithelial markers and are, with variable frequency, S-100 protein reactive. In contrast, chondrosarcomas lack reactivity for epithelial markers and are nearly always 5-100 protein-positive. Thus, studies in which some or all cartilage-containing skull base neoplasms stained for epithelial markers have led their authors to conclude that chondroid chordomas did indeed represent a clinicopathologic entity. Con- versely, those studies in which this group of neoplasms lacked staining for epithelial markers, most notably the study of Brooks et al., led the authors to conclude that they represent chondrosarcoma.6~7~'3 The small number


Table 4. Summary of Immunohistochemical Profile Studies of Chondroid Chordomas ~ _ _ _ _ ~ ~~~

No. of cases with positive epithelial

Study marker* Conclusion

Chondroid chordoma does not exist; it is a chondrosarcoma variant

Chondroid chordoma is a variant of chordoma


Chondroid chordoma is a variant of chordoma

Chondroid chordoma is distinct from chondrosarcoma


Chondroid chordoma exists as a variant of chordoma; epithelial marker-negative examples represent chondrosarcoma

Bottles and Beckstead (1984)7 O / l t

Abenoza and Sibley (1986)6 3/3

Brooks et al. (1987)5 0/7

Salisbury (1987)'' 2/2

Meis and Giraldo (1988)' 1 /3

Walker et al. (1991)" 0/5

Mitchell et al. (1992) 8/25

* Positive for one or more of the following: AE1-AE3, CAM 5.2, MAK 6, EMA, and CEA. t This tumor was negative for the enzyme 5's nucleotidase, which was present in all 4 of 4 classic chordomas studied by the authors.

of chondroid chordomas studied to date may explain the paucity of reported epithelial marker-positive chondroid chord~mas.' ,~,~ When these reports are "pooled," the 28% rate of epithelial marker positivity is similar to the 32% rate observed in our own series.

Thus, we conclude that cytokeratin immunoposi- tivity in chondroid lesions is essentially confined to the subset of true chondroid chordomas. Supportive of this concept is our own search for cytokeratin positivity in 12 extracranial chondrosarcomas; none were positive. S ~ a n s o n , ' ~ in his review of the literature regarding cytokeratin staining in soft tissue sarcomas, found only 1 of 12 dedifferentiated chondrosarcomas and 0 of 44 chondrosarcomas and variants to be immunoreactive for cytokeratins. We believe that the argument rejecting chondroid chordoma on the grounds of epithelial marker-negativity is unfounded in that fully one-third of cartilaginous skull base tumors in our study exhibited an epithelial immunotype.

Points in favor of the interpretation that keratin- negative chondroid tumors represent chondrosarcoma are the largely cartilaginous nature of most examples and the identical immunoprofile of such tumors to ordi- nary chondrosarcoma. In practice, the distinction of chondroid chordoma and chondrosarcoma appears to be of no use, as the survival rates for patients with epithelial marker-positive and immunonegative chondroid lesions are nearly identical (Fig. 4, top right).

Our observations obviate the prognostic necessity of performing immunohistochemical studies in search of keratin positivity in both classic chordomas and cartilage-containing tumors of the skull base. Clearly, as

originally shown by Forsyth et al.,13 their behavior is primarily dictated by the clinical parameter of age at diagnosis. In light of presently available treatment mo- dalities, demonstrating the presence or absence of keratin or other markers of epithelial differentiation pro- vides no useful therapeutic or prognostic information.

The results of ploidy analysis performed on both chordoma subsets were unrevealing. Both classic and chondroid lesions exhibited essentially the same pro- portions of diploid and tetraploid examples. Of interest is the occasional finding of aneuploidy in classic chordomas, whereas none of the chondroid tumors were aneuploid. The significance of this finding is unclear in that the behavior of the aneuploid lesions was similar to that of diploid tumors.

In our opinion, the concept of chondroid chordoma is valid. It represents a distinct neoplasm that, given its tendency to occur in younger patients than classic chordoma, is associated with more favorable survival. Un- like conventional chondrosarcomas, such tumors are immunoreactive for epithelial markers.

References

1. Dahlin DC, MacCarty CS. Chordoma: a study of fifty-nine cases. Cancer 1952; 5:1170-8.

2. Heffelfinger MJ, Dahlin DC, MacCarty CS, Beabout JW. Chor- domas and cartilaginous tumors at the skull base. Cancer 1973;

3. Volpe R, Mazabrand A. A clinicopathologic review of 25 cases of chordoma (a pleomorphic and metastasizing neoplasm). Am Surg Pathol 1983; 7:161-70.

4. Adson AW, Kernohan JW, Woltman HW. Cranial and cervical

32:410-20.

Skull Base Chordomas and Chondroid Neoplasms/MitcheZZ et al. 2949

chordomas: a clinical and histologic study. Arch Neurol Psychia- try 1935; 33:247-61.

5. Brooks JJ, LiVolsi VA, Trojanowski JA. Does chondroid chordoma exist? Acta Neuropathol 1987; 72:229-35.

6. Abenoza P, Sibley RK. Chordoma: an immunohistologic study. Hum Pathol 1986; 17:744-7.

7. Bottles K, Beckstead JH. Enzyme histochemical characterization of chordomas. A m ] Surg Pathol 1984; 8:443-7.

8. Meis JM, Giraldo AA. Chordoma: an immunohistochemical study of 20 cases. Arch Pathol Lab Med 1988; 112:553-6.

9. Miettinens M. Chordoma: antibodies to epithelial membrane antigen and carcinoembryonic antigen in differential diagnosis. Arch Pathol Lab Med 1984; 108:891-2. Salisbury JR. Demonstration of cytokeratins and an epithelial membrane antigen in chondroid chordoma. ] Pathol 1987;

11. Salisbury JR, Isaacson PG. Demonstration of cytokeratins and an epithelial membrane antigen in chordomas and human fetal notochord. A m Surg Pathol 1985; 9:791-7.

10.

153 :3 7-40.

12. Walker WP, Landas SK, Browley CM, Maher Sturm MT. Immu- nohistochemical distinction of classic and chondroid chordomas. Mod Pathol 1991; 4:661-6.

13. Forsyth PJ, Cascino TL, Shaw EG, Scheithauer BW, O’Fallon JR, Dozier JC, et al. Intracranial chordomas: a clinicopathological and prognostic study of 51 cases. ] Neurosurg 1993; 78:741-7.

14. Hedley D, Friedlander M, Taylor I, Pregg C, Musgrove E. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. ] Histochem Cy- tochem 1983; 31:1333-5.

15. Vindelov L, Christensen I, Nissen N. A detergent-trypsin method for the preparation of nuclei for flow cytometric DNA analysis. Cytometry 1983; 3:323-7. Gonchoroff N, Ryan J, Kimlinger T, Witzig TE, Greipp PR, Meyer JS, et al. Effect of sonication on paraffin-embedded tissue preparations for DNA flow cytometry. Cytometry 1990; 11:

Swanson PE. Heffalumps, jagulars, and chesire cats: a commen- tary on cytokeratins and soft tissue sarcomas. A m ] Clin Pathol

16.

642-6. 17.

1991; 95(S~ppl):2-7.

chordoma and chondroid neoplasms of the spheno-occiput: an immunohistochemical study of 41 cases...

Documents