Prognostic Markers in Chondrosarcoma: Evaluation of Cell Proliferation and of Regulators of the Cell Cycle

Purpose. The prognosis, treatment principles and prediction of clinical outcome of patients with chondrosarcoma currently rest on histologic grading which is somewhat ambiguous due to difficulty in pathologic interpretation of this neoplasm. Immunohistochemistry, flow cytometry and oncogene/tumor suppressor gene expression have been examined as alternative indices to predict the biologic behavior of these tumors. Because of partial successes obtained with flow cytometry and because of the improvement in predicting recurrence offered by examining the S-phase fraction, we undertook the current study to determine if expression of specific regulators of the cell cycle would act as prognostic indicators for these patients. Subjects/methods. We examined archival pathologic specimens from 39 patients with at least 2 years' clinical follow-up for the presence of p53, Rb, src and MIB-1 by immunohistochemistry and correlated this with clinical histories and incidence of recurrence. Results. While Rb, p53 and src gene products were identified to a variable extent in these specimens, there was no prognostic significance to their expression. In contrast, MIB-1, an epitope expressed only during semiconservative replication and an accepted marker of cell proliferation, served as a significant prognostic indicator. MIB-1 staining was present in 14.5% of tumor cells in all specimens (range 0–59%). When MIB-1 staining was examined with respect to disease recurrence, there was a statistically significant association between staining and histologic grade (p < 0.05) as well as event-free survival (p < 0.02). Comparing survival curves stratified by MIB-1 expression, there was a significant decrease in event-free survival associated with increasing MIB-1 indices (p < 0.003). Covariates that were associated with event-free survival include histologic grade (p = 0.025) and stage (Musculoskeletal Tumor Society) (p = 0.014). There was no statistical association with patient age (p = 0.15), tumor size (p = 0.47), tumor histology (p = 0.62) or anatomic location (p = 0.316). Discussion. These results indicate that determination of the proliferation index by MIB-1 immunostaining may serve as a useful adjunct to current histopathologic classification. Patients with a high proliferation index may benefit from established adjuvant therapies or experimental approaches including immunotherapy or biologic modulation.


Introduction
C hondrosarcom a rem ains a challenge for all disciplines of oncology. Histologic grading system s have traditionally been used for prediction of clinical behavior. 1 H owever, the pathologist is not infrequently hum bled in an attem pt to distinguish a benign cartilaginous lesion from a m alignant one. T he radiation therapist and the m edical oncologist have been unable to provide any signi® cant advances in local or system ic adjuvant treatm ent. 2 T he patient can be confronted with either local or system ic recurrence for as long as 10 years after the prim ary resection. Taken in conjunction, these factors highlight the need to predict which patients are at highest risk to fail, and the need for novel approaches to system ic adjuvant therapies.
Surgical resection rem ains the prim ary m odality for the treatm ent of chondrosarcom a and survival is m ost directly related to the adequacy of the surgical procedure. 3,4 Patients w ith low -grade chondrosarcom a that are adequately treated exhibit 5-year survival rates that approach 90% . 5 In contrast, those with inadequate surgical treatment and those with high-grade lesions faired m uch m ore poorly with 5-year survival rates of 43% . 6± 8 Several lines of evidence suggest that increased rates of cell proliferation are associated with poor prognosis in patients with chondrosarcom a. T hese include alterations in m itotic index,¯ow cytom etry determ ined by ploidy and S-phase, Ki-67 staining and oncogene expression which have been identi® ed in these patients. W e undertook the current study 1357-714 X/97/020079± 09 $9.00 Ó 1997 Carfax Publishing Ltd in an effort to com pare clinical outcome and histologic grade with determination of src, Rb, p53 and M IB -1 m arkers determ ined by im m unohistochem ical analysis. T hese m arkers were selected because they either re¯ect or in part control the proliferative activity within the cell cycle. MIB-1 (K i-67) is expressed in all phase s of the cell cycle except for G0 and some portions of G 1. As such, it represents a good m arker of the overall proliferative activity in a population of neoplastic cells. p53 is thought to act as a rate lim iting protein affects cell passag e through the G 1 phase of the cell cycle. H ence, its absence or a m utation m ay lead to a transform ed and non-functional phenotype. W hile the native form of p53 is short lived, the m utant form s are thought to persist and be detectable by imm unostaining. T he p53 gene is located on chrom osome 17 and appears to cause cell arrest in the G1 phase . 9± 11 This protein, the deletion of both w ild-type alleles or the presence of a dom inant m utation in one allele can lead to unregulated cell replication. Several m utations of the p53 gene are known and are associated w ith defective protein products. T hese protein products lack the normal proliferation-suppressing properties and have prolonged half-liv es. T he prolonged half-life allows accum ulation of the protein and its detection in abnorm ally large am ounts by quantitative im munohistochem istry. 12 The src gene product is a 60-kD a phospho protein located on the cytoplasm ic side of the plasm a m embrane that functions as a tyrosine kinase. It has been im plicated in the developm ent of avian sarcom as and m ay act through phosphorylation of cytoskeletal proteins such as vinculin. T he src oncogene produces a protein product which, along w ith ras and the myc oncogenes, affe ct cell proliferation and D N A synthesis. 13 The serine/threonine protein kinase p34 is the m ajor regulator of progression through the m amm alian cell cycle and it app ears to activate src. 14 T he precise role of src is unknown but it m ay be asso ciated with cytoskeletal alterations w hich occur at m itosis and is required for cell division. 15 Rb-1 is a nuclear protein which is thought to act as a tum or suppressor gene. T his is based on an absence of expression in retinoblastom a and the ability to restore a norm al phenotype to osteosarcom a cells w ith transfection of this gene. The role of this gene in the etiology of chondrosarcom a is unknown. The RB (RB-1) gene encodes for a protein, p105 R B , w hich has hom ology with a region of protein p107. p107 is known to bind to the T antigen on E1A. 16 pl05 R B is known to bind with doublestranded D N A in a non-sequence-speci® c m anner and app ears to inhibit the stim ulation of D N A polym erase d . p105 RB also app ears to repress myc transcription in som e tumor cells. 14 H ence, RB appears to be an im portant suppressor gene for the cell cycle.
This report correlates markers of cell cycle abnorm alities in patients diagnosed w ith chondrosarcom a with histologic grading and vigilant clinical followup. T he 39 patients had adequate archival pathology specim en, non-m etastatic disease at the time of presentation and were treated at a single institution by a single surgeon over a 21-year period. All patients underw ent resection w ith negative m argins indicating adequate initial therapy. The data suggest that sim ilar to histologic grading, proliferation index derived from M IB -1 staining was a signi® cant prognostic indicator of clinical outcom e in these patients.

Subjects
Sixty-six patients were treated at D uke University M edical Center between 1972 and 1993 for prim ary non-m etastatic chondrosarcom a. O f these, 39 had adequate clinical history and suf® cient pathologic m aterial available to perm it evaluation of im m unohistochem ical m arkers for prognostic signi® cance. W ithin this group of 39 patients, biopsies were perform ed and clinical staging com pleted with radiographs, radionucleotide, computed tomograph y and m agnetic resonance im aging scans when available. T he patients each underwent a w ide or radical resection depending on the speci® cs of the clinical situation and this w as carried out by a single surgeon. All surgical m argins were negative indicating adequate initial treatment. T he patients were then followed with clinical exam ination and serial radiographs for an average of 6.7 years with a m inim um follow-up of 24 m onths. T here were two patients who presented with local failures and seven patients who presented with distant failures. Both patients with local recurrences w ere patients w ho subsequently developed distant disease. T hree patients received radiation therapy and seven received chem otherapy post-operatively follow ing either local or distant recurrence. All patients with distant failure succum bed to the disease process.

Archival specimens
Surgical specim ens obtained at the time of initial resection were routinely ® xed in 10% neutral buffered form alin and subsequently paraf® n em bedded. The histologic sections for the patients included in this study were obtained from the app ropriate paraf® n blocks and were reviewed by one pathologist (LJL).

Immunoh istochemistry
Im m unohistochemistry was perform ed in the D uke U niversity M edical Center C om prehensive Cancer Facility as previously described. 14,17± 19 Histologic m aterial from all cases was exam ined by a single pathologist for sam ple adequacy. U pon tissue block retrieval, serial 5 m m sections were m ounted on F isher 1 Plus 1 slides (Fischer, Charlotte, N C, U SA). The slides were dried overnight at 65°C in an Im perial oven (Baxter, McGraw Park, IL, U SA). Specimens were deparaf® nized in three successive xylene baths and were cleared in abso lute ethanol. Slides were then gradually brought to hydration. All antibodies were diluted in phosphate-buffered saline w ith 2% bovine serum albumin (Sigm a Chem ical C o., St Louis, M O , U SA), pH 7.4 (PBS/BSA). As negative controls, m urine IgG1 (Coulter Source, M arietta, GA, USA) was diluted 1; 100 and normal rabbit serum (Grand Island Biochem ical C o., G rand Island, N Y, U SA) w as diluted 1; 1000 in 2% PBS/BSA.
M IB-1. The prim ary antibody used tor M IB -1 determ ination is a m urine m onoclonal IgG 1 that reacts w ith the Ki-67 nuclear antigen (345-and 395-kD a double band in W estern blot analysis) expressed by proliferating cells. 20 D eparaf® nized slides were incubated in 10 m M citrate buffer, pH 6.0 (Sigm a) for antigen retrieval. Antigen retrieval w as carried out using a 700-W m icrowave (Q uasar, m odel M Q 7677BW , Elk Grove, IL , USA) as recomm ended by the antibody m anufacturer (Imm unotech). Slides w ere placed in 10 m M citrate solution, pH 6.0, and m icrow aved at full pow er for 5 min. T hey were then allow ed to cool and m icrow aved again for 5 m in and w ere slow ly brought to water, trying to avoid any drastic temperature change. Assay slides were rinsed with PBS and then placed in 5% norm al goat serum (NGtS) for 20 min. Excess N GtS w as blotted from the slides and the prim ary antibodies (IgG1 , M IB -1) diluted 1; 100 w ere applied and incubated overnight at 4°C for approxim ately 18 h. All of the assay incubations w ere carried out in a humidity cham ber to prevent solution evaporation. After bringing the slides to room temperature, they were rinsed in PBS, three times for 5 min, and incubated with goat anti-mouse biotinylated antibody (BioG enex, San Ram on, C A, U SA) for 35 m in. The slides w ere rinsed in PBS three times for 5 m in followed by application of peroxidase-conjugated strepavidin label. The slides w ere then developed for 5 m in with chrom agen 3,39 diam inobenzidene (Sigm a) (0.5% diaminobenzidine in 0.05 M Tris buffer and 0.6% hydrogen peroxide). Finally, the assay slides w ere rinsed in running tap water for 10 m in, counter-stained with 1% m ethyl green (Sigma), dehydrated in acetone and cover-slipped in Protexx (Baxter).
Rb, p53, src. Prim ary antibody directed against p53 (PAb 1801) was obtained from O ncogene Science, M anhasset, N Y, USA (Ab-2) w hich is an af® nity-puri® ed IgG 1 m onoclonal antibody that recognizes a denaturation-resistant epitope in the hum an p53 protein located between am ino acids 32 and 79. 21 Im m unostaining was perform ed with 1.0 m g m l 2 1 of primary antibody on deparaf® nized slides. T he secondary antibody was biotinylated af® nity-puri® ed horse anti-m ouse IgG (Vector). Sections were developed in diam inobenzidine solution, rinsed and counter-stained in 1% m ethyl green in sodium acetate buffer, pH 5.2.
The antibody to src is a puri® ed IgG (O ncogene Science, Uniondale, N Y, U SA). T he antibody (titer of 1; 100) w as prepared and run in a fashion described for M IB -1. T he antibody directed against Rb protein product was obtained from O ncogene Science (Uniondale, N Y, USA) and run as just described using a titer of 1; 50.
The quantitation of im m unostaining for each of the markers w as determined using a CAS 200 Im age Analysis System (Becton D ickinson Cellular Im aging System s, San Jose, CA, USA) in com bination with the Quantitative Proliferation Index C AS Software Program (Becton D ickinson Cellular Im aging System s). Q uantitation of proliferative index has been described previously for tumor tissue sections as the percentage of total nuclear area that stains positively with antibody. A sim ilar app roach was applied for p53 and Rb. 22 D ata obtained by com puterized static im age analysis for src were expressed as a percentage of cellular area positively stained with m onoclonal antibody relative to the total cellular area. Fifteen ® elds of the tum or were analyzed at a m agni® cation of 3 400. T he autom ated m ean for staining was scored as the representative index of the neoplasm or it' s com ponents. C ontrol sections, stained with norm al m urine IgG 1 were prepared in each case, observed by standard m icroscopy and analyzed to establish background im m unostaining thresholds. Intra-observer reproducibility w as determined as described previously. 23

Statistical analysis
Patients were divided into low and high expression groups based on the m edian expression value for each epitope. For M IB -1 this was . 6.25% , for Rb, src and p53 . 2% . T he segregation into low and high expression was then used to exam ine the correlation of these param eters with clinical outcom e.
All statistical tests were two-sided with a 5 0.05.
T he analysis of varian ce test w ith T ukey' s honest signi® cant difference procedure or two sam ple t-test were perform ed to assess the association of M IB-1 with other prognostic factors. T he Kaplan± M eier 24 procedure was used to determ ine survival curves and the curves were subsequently com pared using the W ilcoxon rank sum test. 25  The immunostaining is presented as the percentage of patients who demonstrated positive staining for the epitope and also quantitatively as the percentage of cellular area with positive staining normalized to total cellular area. The prognostic signi® cance was determined by survival analysis.

Results
T wenty-seven patients underwent a prim ary attempt at lim b salvage w hile 12 patients underwent a prim ary am putation, all w ith wide or radical surgical m argins. There were two patients who developed a local recurrence and were treated with radiation therapy. C hem otherapy w as adm inistered in the seven patients who developed distant m etastasis. Both of the patients who developed local recurrences also developed distant m etastasis and expired.

Immunostaining
T he qualitative results of im m unostaining of chondrosarcom a specimens for the cell cycle regulators, p53, src and Rb, and for the proliferation index and their prognostic signi® cance are sum m arized in T able 1. T he specim ens were further studied by quantitative im age analysis in an attempt to m ake the evaluation sim ilar to that of the cell proliferation index. T here was no statistically signi® cant correlation between im m unostaining results for Rb, p53 and src with histologic type, grade or stage in these patients.
Tw enty-four of the 39 patients dem onstrated positive staining with the Rb antibody (range 0± 23% w ith a m edian of 4.3% ). Survival analysis indicated that there was no prognostic signi® cance to Rb expression, p 5 0.49 (Fig. 1). N ineteen of 39 patients demonstrated a positive im m unostaining with the src antibody (range 0± 20% w ith a m edian of 2.6% ). Kaplan± M eier survival analysis indicated that the survival in low src expression patients was equivalent to that in patients expressing higher am ounts of the gene product (Fig. 2). W ilcoxon rank sum analysis indicates that this difference in event-free survival is not statistically signi® cant, p 5 0.62. Sixteen of 39 patients demonstrated an abnorm ality in p53 structure on the basis of imm unostaining (range 0± 47% with a m edian of 4.7% ). Patients with p53 abnorm alities demonstrated a shorter event-free survival than those with undetectable p53; however, this difference was not statistically signi® cant, p 5 0.67 (Fig. 3). However, it should be noted that in four of the six patients with a dedifferentiated chondrosarcom a, strong p53 staining w as observed.    The correlations between histologic grade and cell proliferation index are displayed. Histologic grade was determined following resection of the tumor. The mean of the proliferation index and the standard error (SE) of the m ean are indicated. The number of patients for each group is also indicated. The proliferation index is examined for patients who remained disease free and for those who experienced a recurrence. The mean and the standard error (SE) of the m ean for proliferation indices are displayed. The number of patients in each group is also indicated.

T able 3. Proliferation index and event-free survival
M IB -1 im m unostaining was used to quantitate a proliferation index. T he m ean of the proliferative index was 14.5% w ith a range of 0± 59%. T here was a statistically signi® cant asso ciation between MIB-1 level and histologic grade, p , 0.05 (Table 2), and w ith disease status, p , 0.02 (Table 3). C om paring survival curves strati® ed by MIB-1 staining, there w as a signi® cant decrease in survival with increasing M IB-1 indices (p 5 0.003; F ig. 4). Cox analysis was perform ed and the proliferation index was an independent predictor of event-free survival. Covariates that were asso ciated with survival time included histologic grade (p 5 0.025) and Enneking stage (p 5 0.0142). There was no asso ciation with age (p 5 0.1556), tum or size (p 5 0.4671), tum or type (p 5 0.6211) or anatom ic location (p 5 0.3160).
The m ean proliferation index was exam ined with respect to histologic type of chondrosarcom a. T here w ere 22 classic intram edullary chondrosarcom as, six dedifferentiated chondrosarcom as and 11 chondrosarcom as including those arising in osteochondrom as (extraskeletal, clear cell and m yxoid variants w ere grouped to facilitate comparison). The mean proliferation index for the intram edullary chondrosarcom as w as 0.106, that for the dedifferentiated chondrosarcom as was 0.396 and that for the grouped varian ts w as 0.086 (Table 4). The dedifferentiated chondrosarcom as had a proliferation index higher than the m ean index and also had a low er event-free survival; however, survival analysis could not be perform ed due to limited numbers.

D iscussion
N eoplastic cell proliferation, invasio n and m etastases represent independent and im portant variables determ ining the aggressiveness and behavior of neo-  The proliferation index for each of the histologic subtypes of chondrosarcoma is displayed. The proliferation index is displayed as the mean and the standard error (SE) of the m ean for each group. The number of patients with each diagnosis is also displayed. but is probably not a m ajor factor contributing to disease progression.

p53
Recent studies have detected m utations in the p53 gene in som e soft tissue sarcom as, 31,32 and ® brom atosis. 33 King reported the developm ent of chondrosarcom a in patients with Li-Fraum eni syndrom e im plicating p53 m utation in the m alignant transform ation of cartilage tissue. 34 T oguchida et al. exam ined a group of sarcom a patients and found a 33% incidence of p53 m utation. 32 W adayam a et al. identi® ed 25% of sarcom as having a p53 mutation, with ® ve of 20 chondrosarcom as having m utations. 34 This ® nding w as asso ciated predom inantly with a dedifferentiated phenotype in each case. 35 D obash i et al. exam ined p53 expression in 16 chondrosarcom a specimens and found only two patients with over-expression, however, both of these were high-grade (III) lesions. 36 Yang et al. found p53 over-expression to be of no prognostic signi® cance in a series of 54 m alignant ® brous histiocytomas. 37 N awa et al. reported a correlation of p53 staining with patient survival; how ever, in m ultivariate analysis this did not emerge as an independent predictor. 38 Lastly, Simm s et al. reported strong staining in the non-cartilaginous com ponent of dedifferentiated chondrosarcom a patients in all of the eight patients exam ined and only light staining in the cartilaginous com ponent of the lesion in two patients. 39 The experience of these investigators is consistent with the data presented in the current study which show s that four of six dedifferentiated chondrosarcom as stained positively for p53. As mentioned, this subset of patients faired poorly from a clinical standpoint. So while it seem s that in these patients p53 m utations m ay be associated w ith the dedifferentiated phenotype, they are not a useful prognostic indicator of chondrosarcom a recurrence in general. It also suggests that the dedifferentiated com ponent of chondrosarcom a should be considered a separate disease process.

rb-1
T he retinoblastoma gene (RB) codes for a nuclear phospho protein involved in cell cycle control. 40 T he RB gene is m utated or absent in a variety of m alignancies. 15,24,38,41± 47 M onoclonal antibodies have been developed for identi® cation of the Rb protein in paraf® n-em bedded tissues. Studies correlating Rb protein status and clinical outcom e have demonstrated that the absence of the Rb protein can be associated with a poor prognosis 35,48 as well as resistance to som e chem otherapy agents. 49 In the patients exam ined in the current study, 27 of 39 patients stained positively for Rb protein. T here was no statistical correlation between Rb-1 staining and event-free survival. vascular invasion and histologic analysis of host/ tum or boundary interactions. Recently, a variety of oncogenes and their protein products have been recognized as regulators of cell behavior. A variety of these protein products are known to affe ct the cell cycle and hence represent control substances for cell proliferation. Others like M IB -l appear to be useful as m arkers of overall cell proliferation. The regulation of cell proliferation by oncogene protein products is com plex and at present is only partially understood. W e chose to study the predictive value of the cell proliferation index as determined by M IB-1 as well as the predictive value of quantitation of certain oncogene products known to have regulatory functions in the cell proliferation cycle.
src It is known that cells transfo rm ed by src will be inhibited from entering into the S-ph ase when antiras antibodies are injected into cell cultures. Such observations suggest that src oncogene transfo rm s cells via a pathw ay involving ras. O ther tumor suppressor genes including p53 and RB-1 activate pathw ays that converge downstream of the ras oncogene point of control. T he src fam ily of genes codes for proteins w ith tyrosine kinase activity. 26 This activity is norm ally inhibited in vivo by phosph orylation of T yr 527. In intact cell, the targets of src involved in regulating cell proliferation have not been identi® ed 27 but src m ay play an indirect role in cell cycle regulation by its ability to phospho rylate p34. 28 In studies with m ice, src app ears to be essential for bone form ation, thus im plicating a physiologic role in norm al m esenchym al tissue. Early reports had im plicated myc and src oncogenes in the proliferation and synthesis of speci® c matrix com ponents in chondroblasts. 29 Zhu et al. were unable to detect src gene transcription using N orthern hyb ridization. 30 W e w ere able to identify 19 of 39 patients with im munohistochemical evidence of src gene product translation but there was no statistical correlation w ith clinical outcom e. W e conclude that the role of the src gene product in the etiology and clinical behavior of chondrosarcom a is currently unknown M IB-1 T he cell proliferation index has been determ ined in the past with thym idine labeling, 50 5-brom odeoxyuridine labeling, 51¯o w cytometry 52± 54 and by quantitation of nucleolar organizer regions. 55 Assessm ent of the growth fraction m ay also be perform ed on frozen sections by Ki-67 im m unostaining w hich reacts with a proliferation associated antigen expressed in all cells not in G 0 or portions of the G 1 phase of the cell cycle. 56,57 M ore recently, several antibodies have been described that demonstrate proliferating cells in form alin-® xed tissue including K i-67, PCN A and M IB -1. 58± 60 M IB-1 staining has been demonstrated to have prognostic in¯uence in patients with breast carcinom a, 61,62 ovarian cancer, 63 but a recent study exam ining benign and m alignant ® brohistiocytic lesions demonstrated only a lim ited role for M IB -1 in distinguishing between these lesions. 64 In contrast, Yang et al. did show that the proliferation index as determ ined by K i-67 imm unostaining had prognostic value in ® brous lesions. 65 O ur data supported the usefulness of the proliferation index (M IB -1) as a signi® cant prognostic indicator for chondrosarcom a: a higher proliferation index was asso ciated with a lower event-free survival (p . 0.01).
D N A analysis by¯ow cytometry has been explored in the past as a means of predicting biologic behavior of chondrosarcom a. 66 Alho et al. reported a statistically signi® cant asso ciation between m alignant behavior of the tum or and aneuploid content. 67 A later report by the sam e senior investigator extended this analysis over a larger patient population w ith similar results. 67 In analysis by¯ow cytom etry, it can be dif® cult to differentiate D N A ploidy abnorm alities from changes in D N A content asso ciated w ith the cell cycle. M ore recently, an independent investigator has reported that the S-phase fraction rather than aneuploidy correlates m ost closely with biologic behavio r. 69 T he num ber of tum ors examined was small (four benign and four m alignant cartilaginous tum ors), but the data suggest that the proportion of cells that have entered the cell cycle w as prognostically im portant. An alternative m eans of determining the fraction of cells entering the cell cycle is to exam ine the expression of the Ki-67 epitope. N awa et al. have recently reported a correlation of M IB -1 staining with histologic grade and event-free survival. 38 N awa et al. found a mean proliferation index of 3.2% for grade I lesions, 14.7% for grade II lesions and 16.4% for grade III lesions. This is consistent with the distribution of proliferation indices that w as observed in the current study and indicates that the proliferative index serves as a prognostic indicator for chondrosarcom a.
It is interesting that the cell proliferation rate appears to be an important prognostic factor in patients with chondrosarcom a yet two of the proteins identi® ed in cell cycle control do not ap-pear to be involved. T his leads one to speculate that there are additional mechanism s regulating the proliferative rate in cartilaginous tissues.

Concluding rem arks
T he data presented in this report indicate that the proliferation rate of chondrosarcom a cells serves as a prognostic indicator for disease recurrence and clinical outcom e. Tum ors w ith a high rate of cell proliferation have a worse prognosis and patients af¯icted with these tum ors dem onstrate a shorter event-free survival. M IB -1 m ay serve as a useful clinical adjunct to histopathologic grading and¯ow cytom etry in the staging and treatment of patients with chondrosarcom a. Prognosis was not correlated to the level of regulatory protein products (oncogenes, Rb-1, src, p53) in our experience. Perhaps quantitation of other regulatory oncogenes of cell proliferation as well as m arkers of cell± cell cohesion or invasive ness might yield results m ore predictive of tum or behavior and correlate with the proliferative index. A high proliferation index at the time of biopsy m ay lead clinicians to consider existing adjuvant therapies as these tum ors m ay be m ore responsive because of their increased rate of DN A synthesis. Alternatively, future adjuvant therapies such as im m unotherapy or biologic m odulation m ay be appropriate for these patients who are at high risk of recurrent disease.