CD44s expression correlated with the International Neuroblastoma Pathology Classification (Shimada system) for neuroblastic tumours

LTaylor&Francis

Pathology (2003) 35, pp. 125-129

• healthsciences

ANATOMICAL PATHOLOGY

CD44s expression correlated with the International Neuroblastoma Pathology Classification (Shimada system) for neuroblastic tumours M.

1.

l uLlA MUNCHAR ,

N. A.

SHARIFAH,

R.

lAMAL· AND

L.

M.

LOOIt

Departments of Pathology alld * Paediatrics, Faculty of Medicine, Ul1illersiti Kebangsaan Malaysia and

t Department of Pathology, University of Malaya, Malaysia

Summary Aim: CD44 is a cell adhesion molecule that plays an important role in the cascade of metastasis and progression of human malignant tumours. A large family of variants or isofarms, generated by alternative splicing of a single gene, has been reported to be involved in the malignant process by conferring metastatic potential to non-metastatic cells. The objective of this study was to compare the expression of CD44 standard molecule with the International Neuroblastoma Pathology Classification (INPC) for neuroblastic tumours, a histological grading syslem based on the Shimada system for predicting the clinical outcome in neuroblastic tumours. Methods: Paraffin blocks of primary neuroblastic tumour

were graded according to the INPC system into favourable and unfavourable histological types. Tumour tissues were then analysed with immunostaining using monoclonal antibodies against the CD44 epitope. Results: In this retrospective study, 32 cases of primary neuroblastic tumour were collected. Based on the IN PC system, 13 cases (40.6%) had a favourable histology while 19 cases (59.4%) were of unfavourable histology. CD44 slandard (CD44s) expression was detected in 69.2% of neuroblastic tumours with favourable histological type while 89.5% of tumours with unfavourable histological type did not express CD44s staining. Statistically, there was significant correlation belween morphological grading of neuroblastic tumours based on the INPC and the presence of CD44 staining (Fisher's exact test, P < 0.05). Conclusion: This stUdy shows that there is a Significant correlation between CD44s expression and tumour histology based on the INPC in which CD44s non-expression is correlated with an unfavourable histological type and the expression of CD44s with a favourable type. Therefore, the analysis of CD44s expression should be recommended as an additional biological marker in the initial routine staging of the disease. Key Ilv'ords: CD44, neuroblastoma, Shimada's histological classification . Received 14 June, revised 22 August. accepted 26 August 2002

INTRODUCTION Neuroblastic tumours are tumours derived from the sympathoadrenal system in children. They exhibit a broad range of differentiation, from the malignant neuroblastoma at one end of the spectrum to the benign ganglioneuroma ISSN 0031-3025 printed/ISSN 1465- 3931 online/03/020 t 25-05 DOl: 10.1080/0031302031000082214

at the other end, with ganglioneuroblastoma in between. Neuroblastoma represents the most common extracranial solid tumour in infancy. 1 The overall prognosis of this tumour is poor; however, the prediction of surviva l rate can be difficult due to the biological heterogeneity of the tumour. In the past decades, various prognostic factors have been reported [or neuroblastoma, which include age at diagnosis, sex of the child, site, tumour histology, tumour stage and catecholamine pattern. 2 Among these, clinical staging and histological differentiation have been the commonly used criteria in determining the prognosis of this tum our.' Shimada's histological grading, proposed by Hiroyuki Shimada in 1984, has been known to be valid and reproducible in distinguishing between favourable (2-year survival rate of 87%) and unfavourable (2-year survival rate of 7%) prognostic types of neUIoblastic tumours.' The International Neuroblastom a Pathology Classification (INPC), which was proposed by the International Neuroblastoma Pathology Committee in 1999, is part of the international co-operative effort to develop a complete set of international neuroblastoma risk groupS. 3 The INPC was convened with the objective of proposing a prognostically significant and biologically relevant cla ssification. It is a morphological classification of neuroblastic tumours and is based on Shimada's classification, risk group cla ssification and the modified risk group (Joshi) classification. Statistical analysis has demonstrated that the Shimada classification system has a 3-year event-free survival rate of 85% for the favourable histology and 41 % for the unfavourable histology groups· Therefore, the INPC has been proposed for international use in assessing neuroblastic tumours. With the advance of diagnostic techniques such as trucut biopsy and fine needle aspiration, smaller tissue samples are obtained for histological evaluation. Therefore, histological grading may not be possible in determining the prognosis of neuroblastic tumours. Furthermore, these tumours are known to have the ability to differentiate or spontaneously regress, 5 therefore, the initial histological grading may not be sufficient to allow reliable prediction of disease outcome. In the past, there has been a search for more discriminatory criteria to predict the biological heterogeneity of neuroblastic tumours in certain subgroups of patients. New prognostic markers that have been described include partial monosomy of chromosome 1,6 amplification of the N-myc oncogene,7-9 tumour cell pioidylO,t and cell surface molecule CD44 expression. 12

© 2003 Royal College of Pathologists of Australasia

126

MUNCHAR "

al.

The CD44 molecule is a hyaluronate receptor and a cell adhesion molecule. It was first described by DaJchau el. al. on a variety of cells of haemopoietic, epithelial and mesothelial origins. i3 CD44 molecule has been associated with diverse physiological functions such as cell-cell adhesion, cell- matrix interaction and lymphocyte homing and circulation. It consists of a family of immunologically related integral membrane glycoproteins with molecular weights ranging from 80 to 250 kDa and is derived from a single gene containing 20 exons on chromosome 11.14 Alternative splicing of the 10 exons located between exons 5 and 16 of the CD44 gene may produce variant isoforms (CD44v). IS-18 The most abundant form of CD44 in tissues is the standard form (CD44s) in which there is no expression of the 10 variant isoforms,19.20 Both clinical and experimental studies have linked the CD44 molecule with tumorigenesis. A number of studies in various malignancies have shown that the CD44 standard and variant isoform proteins are overexpressed in certain tumours and down-regulated in others. In this study we correlated the expression of the cell surface molecule CD44s with the histological grading of the tumour based on the INPC.

MATERIALS AND METHODS Case selection Tissue specimens were obtained from 32 untreated cases of neuroblastic tumour from 1974 to 1996. A total of 14 suitable tissue blocks were retrieved from the Pathology Department, National University Malaysia, and the additional 18 cases were collected from the Pathology Depart· meDt, University of Malaya. The selection of patients for the study was based on the availability of sufficient amounts of tumour tissue for histological and immunohistological analyses. Inclusion criteria included all neuroblastomas and ganglioneuroblastomas from the primary site (the adrenal and any of the sympathetic chain). Ganglioneuroma cases were not included in the study as they were not included in the original Shimada classification for neuroblastic tumours. On reviewing the medical records, any patients who had received either chemotherapy or radiotherapy prior to tumour sampling were also excluded from the study, Diagnosis and histological grading of neuroblastic tumours were performed according to the INPC. Based on the morphological criteria as defined for the INPC,4 neuroblastic tumours were classified into four categories and their sUbtypes: (1) neuroblastoma (Schwannian stromapoor), undifferentiated, poorly differentiated, and differentiating; (2) ganglioneuroblastoma, intermixed (Schwannian stroma-rich); (3) ganglioneuroma (Schwannian stroma-dominant), maturing and mature; and (4) ganglioneuroblastoma, nodular (composite Schwannian stromarich, stroma-dominant and stroma-poor). Specific features, such as the mitosis-karyorrhexis index (MKI). the mitotic rale and calcification were evaluated. Overall, two prognostic groups were distinguished in this classification. The favourable histology group included: (a) neuroblastoma with low or intermediate MKI diagnosed at age I.S years; (b) neuroblastoma, differentiating subtype with low MKl diagnosed between ages I.S and S years; (c) ganglioneuroblastoma, intermixed at any age; and (d) ganglioneuroma at any age. The unfavorable histology group included: (a) neuroblastoma with high MKl at any age; (b) neuroblastoma, undifferentiated and poorly differentiated sUbtype diagnosed between ages 1.S and S years; (c) neuroblastoma with intermediate MKl diagnosed between ages I.S and S years; (d) neuroblastoma diagnosed at age > S years; and (e) ganglioneuroblastoma, nodular, diagnosed at any age. The 3-year event-free survival rate was 8S% for the group with favourable histology and 41% for the group with unfavourable histology,4

Pathology (2003), 35, April Detection of cell-surface CD44s expression by immunohistochemical slaining

The monoclonal CD44s (mouse IgG 1 isotype) is derived from the A3D8 hybridoma (Sigma Immunochemicals, USA). Immunohistochemical staining was performed using an indirect three-stage immunoenzymatic procedure with diaminobenzidine as described below. Deparaffinised sections were brought to water before being treated with Target Retrieval solution and microwaved for 20 min. Slides were then left to cool at room temperature before washing in running water and treatment with 3% hydrogen peroxide for 10 min at room temperature. This step was mainly to block the endogenous peroxidase activity. After several washes with Tris-buffered saline (TBS), the sections were incubated with the primary antibody for CD44 (1:S0 dilution) at room temperature for 30rnin, followed by link antibody and streptavidin at ISmin each. Between these steps, the tissue sections were washed with three changes of TBS each. Finally, a drop of diaminobenzidine was placed on each slide for 7 min before they were washed in water and counterstained with haematoxylin, Sections were dehydrated in increasing concentrations of alcohol, cleared with xylene and mounted with cover slips. In the interpretation of results, the samples were classified as positive for CD44 expression when greater than 10010 of tumour cells showed positive plasma membrane reactivity with anti-CD44 monoclonal antibody. However, in almost all tumour samples, staining was strong and homogenous in more than 90'% of the population .

Statistical analysis Statistical comparisons between subgroups were performed using the Fisher's exact test.

RESULTS Thirty-two tumour specimens from newly diagnosed neuroblastic tumour patients were collected since 1974. There were 13 cases (40.6%) with favourable histology and 19 cases (59.4%) with unfavourable histology based on this classification. Positive immunostaining was observed on 11 samples (34.4%). Table 1 shows the correlation between CD44 cell expression and the histological grading of neuroblastic tumours based on the INPC. Of the 13 cases (40.6%) of favourable histology, nine cases (69.2%) showed CD44s cell surface expression and four cases (30.8%) were negative for CD44s. Of the 19 cases (59.4%) classified as unfavourable type, two cases (10.5%) were positive and 17 cases (89.5%) were negative for CD44s expression. Figure 1 shows the expression of CD44s cell surface on immunohistochemical staining. Statistically, there was significant correlation between Shimada's histological grading and the presence of CD44s cell surface expression. TABLE I Prognostic type of neuroblastic tumours based on the International Neuroblastoma Pathology Classification for neuroblastic tumours in relation to CD44 staining CD44

Positive (n = I I) Negative (n=21) Total

Favourable type (Shimada's grading) (%)

Unfavourable type (Shimada's grading) (%)

9 (69.2)

2 (10.5) 17 (89.5) 19 (100.0)

4 (30.8) 13 (100.0)

Fisher's exact test, P
(

l CD44 AND NEUROBLASTOMA

Fig. I

127

Immunohistochemistry of CD44s shows a strong membranous pattern of staining in most tumour cells.

DISCUSSION The role of the CD44 adhesion molecule in the evolution and progression of cancer has received significant attention

expression or down-regulation of CD44v has been shown to be co-resfonsible for the progression of squamous cell carcinomas2 and endometrial carcinomas. 27 This suggests that tumour invasion and metastasis may be enhanced by both cell- matrix interaction in tumours that show CD44

differentiation (P0.09) . In this study, we analysed the expression of CD44s in 32 cases of neuroblastic tumours collected over a span of 20 years . Using the INPC system, 13 of the 32 neuroblastoma tumours (40.6%) were classified as favourable prognostic type and 19 cases (59.4%) were of unfavourable prognostic type. Our results demonstrate that CD44s-negative tumours (approximately 89.5%) are associated with the unfavourable prognostic type based on the INPC (with a poorer event-free survival probability) in comparison with CD44spositive tumours (approximately 65.5%) (P < 0.05). The results from this study were comparable with other

show CD44 down-regulation. CD44 surface molecule is expressed in many foetal and adult tissues, and in foetal sympathoblasts of the adrenal medulla (tenth week of gestation)l' An earlier study by Terpe et al. in 1993 on CD44s and CD44v expression in 20 primary neuroblastomas showed that a subgroup of these tumours failed to express CD44s, and that none of the tumours expressed CD44v." In 1995, the same authors" showed that primary neuroblastomas [52 diffuse-stroma rich (DSR) tumours, 82 stroma-poor differentiated (SP-D) tumours and 71 stroma-poor undifferentiated (SP-V) tumours]' exhibited an inverse correlation between CD44s expression and tumour progression. All the investigated tumours were CD44v negative. The results showed that CD44s-negative tumours (approximately 14%) are associated with a poorer event-free survival probability in comparison with CD44s-positive tumours (approximately 86%) (P<0.05). Studies by Gross et al.3o have also concluded that CD44s, but not CD44v, is a prognostic marker in neuroblastoma, which Terpe et al.'9 also found to correlate significantly with the grade of tumour cell

inverse correlation with the expression of CD44 isoforrns in contrast to other malignant tumours."-38 The presence of CD44s on neuroblastic cells, in particular, might confer them decreased tumorigenic and metastatic properties. Studies have shown that the expression of CD44s by neuroblastic cells is linked to tumour maturation and differentiation. 3o Therefore, CD44s expression is seen in the more mature tumour, which corresponds with the better prognostic type, based on the INPC system. In this study, we have used the International Neuroblastoma Pathology Classification (INPC) as the gold standard for prognostic evaluation and comparison with a new prognostic marker, CD44s. Shimada et al. 4 demonstrated statisticaUy that the Shimada classification system (90% consensus; 3-year event-free survival, 85% for the group with favourable histology and 41% for the group with unfavourable histology; P=0.31 x 10- 9) had a significantly stronger prognostic effect than other prognostic groups. We would like to recommend the use of CD44s as a prognostic marker in place of morphological evaluation,

in recent years. Enhanced expression or up-regulation of

CD44s or CD44v has been related to tumour progression in malilSnant non-Hodgkin's lym~hornas,2 1 gastric carci-

nomas, 2 colorectal carcinomas? brain tumours 24 and carcinomas of the uterine cervix. 25 In contrast, decreased

overexpression and cell-cell interaction in tumours that

studies6,7, 14,29- 32 on neuroblastic tumours, indicating an

128

MUNCHAR el al.

which requires a significantly large amount of tumour tissue for assessment. The INPC requires tumour tissue containing at least 5000 viable neuroblastic cells from mUltiple microscopic fields for the assessment of MKI and grade of neuroblastic differentiation. The analysis of CD44s expression by immunohistochemistry, however, is a sensitive technique that requires minimum amounts of tumour material and can be easily standardised for laboratory routine. Unlike the analysis of genetic abnormalities, immunohistochemical technique is rapid and more cost effective. In our study, the high specificity of this test (approaching 90%) indicates that it is reliable in detecting true negative cases. Thus, the detection of non-expression of CD44s is a predictor of an unfavourable clinical outcome and a more aggressive approach to treatment with extensive chemotherapy and bone marrow transplantation can be advocated. Our results were comparable with those of Terpe et al. 29 in 1995 comparing CD44s expression with the original Shimada histological grading for neuroblastic tumours. Their study on 205 cases of neuroblastic tumours showed that all investigated tumours were CD44v-negative, while the subpopulation of CD44s-negative tumours (approximately 14%) is associated with a poorer event-free survival probability in comparison with CD44s-positive tumours (approximately 86%) (P<0.05). We were unable to study the impact of survival in our 32 cases due to non-standardised treatment protocols as the cases selected were from different institutions and were managed differently. Furthermore, in some cases there was unavailability of data as the patients were lost to follow-up. However, a study in 1993 by Favrot et al. on 52 patients with neuroblastic tumour treated with standard treatment protocols, demonstrated that in the CD44-positive group the I-year survival rate was 92% but in the CD44-negative group the survival rate was 29%.31 The CD44 expression was determined using a monoclonal antibody (JI73) directed against an epitope common to all CD44 isoforms. The role of the CD44 molecule in neuroblastoma has been extensively researched by Gross et al. 30.32 The authors have shown that most, if not all, of CD44s functions are attributed to its ability to recognise hyaluronate. They have also proven that the absence of CD44-mediated hyaluronate binding, as demonstrated by modifications on the adhesive properties of CD44s, could be related to the increased malignancy in human neuroblastomas. 32 In the study by Combaret et al. in 1996,7 a multivariate analysis of four well-known parameters of neuroblastoma (i.e., stage of disease, age at diagnosis, N-myc amplification, and tumour histology) and of CD44 expression was performed. In their study, comprising 52 tumour samples of neuroblastoma, the CD44 expression was determined by immunostaining using the monoclonal antibody (JI73) directed against an epitope at the CD44 constant region. Their results showed that disease stage and CD44 expression were the only statistically independent prognostic factors for patient survival. Expression of CD44 by malignant cells strongly correlated with the absence of N-myc amplification. However, the analysis of CD44 affords significant prognostic discrimination in subgroups of patients with or without N-myc amplified tumours.

Pathology (2003). 35. April

In conclusion, the non-expression of CD44s in neuroblastoma cells is associated with an undifferentiated phenotype as observed in the unfavourable prognostic group based on Shimada's histological classification. CD44s is thus a marker of differentiation and represents a highly valuable prognostic marker. Therefore, the analysis of CD44 expression must be recommended as an additional biological marker in the initial staging of neuroblastic tumours. Address for correspondence: Dr J. Munchar bt. Munchar Jajuli, Pathology

Department, F aculty of Medicine, Hospital Universiti Kebangsaan Malaysia, Jatan Yaacob Latif, Bandar Tun Razak, 56000 Kuala Lumpur. Malaysia. E-mail: julia@mail .hukm.ukm.my

References 1. Grosfeld JL. Neuroblastoma: a 1990 overview. Paediatr Surg 1nl 1991; 6, 9~13. 2. Sltimada H, Chatten J, Newton WA, et al. Histopathologic prognostic factors in neuroblastic tumours: Definition of subtypes of gangIioneuroblastoma and an age-linked classifications of neuroblastomas. J Natl Cancer Inst 1984; 73: 405-13. 3. Shimada H, Amborse IM, Dehner LP, Hala J. Joshi VV, Roa1d 8. Terminology and morphologic criteria of neuroblastic tumors: recommendations by the International Neuroblastoma Pathology Committee. Cancer 1999; 86; 349-63. 4. Shimada H, Amborse IM, Dehner LP, Hata J, 10shi VV, Roald 8. International Neuroblastoma Pathology Classification. Cancer 1999; 86, 364-72. 5. Niethamme D , Handgretinger R. Clinical strategies for the treatment of neuroblastoma. Ellr J Cancer 1995; 31A: 568-7\. 6. Combaret V. Lasset C, Frappez D, et af. Evaluation of CD44 prognostic value in neuroblastoma: Comparison with the other prognostic factors. Ellr J Cancer 1995; 31A: 545-49. 7. Combaret V, Gross N, Lasset C, et af. Clinical relevance of CD44 cell-surface expression and N-myc gene amplification in a multicentric analysis of 121 paediatrics neuroblastoma. J Clill Ollcol 1966; 14: 25- 34. 8. Brodeur GM. Seeger RC, Schwab M, et al. Amplification of N-myc in untreated human neuroblastoma correlated with advanced disease stage. Science 1994; 224: 1121-4. 9. Seeger RC, Brodeur G, Sather H, et af. Association of multiple copies of the N-myc oncogene with rapid progression of neuroblastomas. New Engl J Med 1995; 313: lllJ - 6. 10. Look AT, Hayes FA Nitschke R, McWiUiams NB, Green AA . Cellular DNA contents as a predictor of response to chemotherapy in infants with unresectable neuroblastoma. New Engl J Med 1994; 311 , 231~5 . It. Christiansen H. Lampert F. Tumour karyotype discriminates between good and bad prognostic outcome in neuroblastoma. Br J Cancer 1988; 57, 1 21~6. 12. Terpe JH, Stark H. Prehm P, Gunther U . CD44 variant isoforms are preferentially expressed in basal epithelia of non-malignant human fetal and adult tissues. Histochemistry 1994; 101: 79- 89. 13. Da1chau R. Kirkley J, Fabre J. Monoclonal antibody to a human brain-granulocyte antigen probably homologous to the W31l3 antigen of the rat. Eur J Immun 1980; 10: 745-9. 14. Shtivelman E. Bishop IM. Expression of CD44 is repressed in neuroblastoma cells. Mol Ceil 8io1 1991; 11: 5446-53. 15. Screaton GR, Bell MY, Jackson DG. et al. Genomic structure or DNA encoding the lymphocyte homing receptor CD44 reveals at least 12 alternatively spliced exon. Proc Natl Acad Sci USA 1992: 89, 12160-4. 16. Screaton GR . Bell MY, Bell Jl. et al. The identification of a new alternative exon with highly restricted tissue expression in transcripts encoding the mouse Pgp-l (CD44) homing receptor: comparison ofal! 10 variant exons between mouse. human. and rat. J 8iol CIJem t 993; 268, 12235-8. 17. Tolg C, Hofmann M, Herrlich P, et al. Splicing choice from ten variant exons establishes CD44 variability. Nucleic Acids Res 1993; 2L 1225- 9. 18. Jackson DG, Buckley J, Bell J1. Multiple variants of the human lymphocyte homing receptor CD44 generated by insertions at a single site in the extracellular domain. J Biof Chem 1992; 267: 4732- 9.

CD44 AND NEU ROBLASTOMA 19. Stamenkovic I, Amiot M, Prcsando LM , et u1. A Iymphocyte molecule implicated in lymph node homing is a member of the cartilage link protein family. Cefl 1989; 56; 1057- 62. 20. Goldstcin LA, Zhou DF, Picker LJ , et al. A Iymphocyte homing rcceptor, the heme antigen, is related to cartilage proteoglycan core and link protein family. Cell 1989: 56: 1063-72. 21. Koopman G. Heider KH , Horst E, et 01. Activated human Iymphocytes and aggressive non-Hodgkin's lymphomas express a homologue of ral metastasis-associated variant of CD44. J Exp Med 1993; 177: 897- 904. 22. Mayer B, Jauch KW, Gunthert U. et al. De-novo expression of CD44 and survival in gastric cancer. Lancet 1993; 342: 1019- 22 . 23. Wielenga YJ, Heider KH . Offerhaus Gl . et al. Expression of CD44 varian t proteins in human colorectal cancer is related to tumour progression. Can cer Res 1993; 53: 4754-6. 24 . Li H, Hamou MF, Tribolet de N, et al. Variant CD44 adhesion molecules are expressed in human bntin melaslases bu t not in glioblastomas. C(lncer Res 1993; 53: 5345-9. 25. Dall p, Heider KH, Hekele A. et al. Surface pro tein expression and messanger RNA-splicing ana lysis of CD44 in uterine cervical cancer and normal cervical epithelium. Cancer Res 1994; 54; 3337-41. 26. Salmi M, Gron-Virta K. Soin[u P. et al. Regulated expression of exon v6 containing isoforms of CD44 in man: downregulation during malignant transfonnat ion of tumours of squamoceliular origin. J Cell BioI 1993: 122: 431 -42. 27. Fujita N, Yaegashi N, Ide Y, et af. Expression of CD44 in normal human versus tumour endometrial tissues: possible implication of reduced expression of CD44 in lymph-vascular space in volvement of cancer cells. Callcer Ri'.f 1994; 54: 3922- 28.

129

28. Terpe HJ, Franke F, Stark H, et al. CD44 und seinen Isoformeuunter orthologischen und pathologischcn Bedingungen. Verh Dtsch Ges Pat}wl 1993; 77: 276-8 1. 29. Terpe HJ, Christiansen H. Gonzales M, et al. Differentiation and prognosis of neoroblastoma in correlation to the expression of CD44s. Ellr J Cancer 1995: 31A: 549- 52. 30. Gross N, Berelta C, Perruissca u G, et al. CD44H expression by human neuroblastoma cells in relation to N-myc amplifica tion and lineage differentiation . Cancer Res 1994: 54: 4238-42. 31. Favrot MC, Combaret V, Lasset C. CD44 a new p rognostic marker ror neuroblastoma. New ElIg J Med 1993; 329: 1965. 32. Gross N. Balmas K. Beretta Brognara C. Role or CD44H carbohydrate structure in neuroblastoma adhesive properties. Med Pediatr ancol 2001; 36: 139-4 1. 33. Gunthert U, Hofman M, Rudy W. et af. A new variant of glycoprotein CD44 confers metastatic potential to tat carcinoma cells. Cell 1991; 65: 13- 24. 34. Hofmall M, Rudy W, Zoller M, et Ill. CD44 splice variants confer metastatic behavior in rats; homologous sequences are expressed in human tumour cell lines. Cuncel' Res 1991: 51: 5292- 7. 35. Matsam ura Y, Tarin D. Signific ance of CD44 gene products fo r cancer diagnosis and disease evaluation . Lancet 1992; 340; 1053- 8. 36. Fox SB, Gatter KC, Jackson DG, et al. CD44 and cancer screening. Loncet 1993; 342: 548- 9. 37. Tanabe K . Ellis L. Saya H . Expression of CD44RI adhesion molecule in colon carcinoma mctastases. L(lnce! 1993; 34 1: 725- 6. 38. Birch M, Mitchell S. Hart JR . Isolation and characterization of human melanoma cell variants expressing high and low levels of CD44. o,Ilcer Res 1991: 51: 6660- 7.