SEARCH

SEARCH BY CITATION

Keywords:

  • CT-antigen;
  • NY-ESO-1;
  • antigen expression;
  • ES121;
  • tumor-associated antigen

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

NY-ESO-1, a member of the CT (cancer/testis) family of antigens, is expressed in normal testis and in a range of human tumor types. Knowledge of NY-ESO-1 expression has depended on RT-PCR detection of mRNA and there is a need for detecting NY-ESO-1 at the protein level. In the present study, a method for the immunochemical detection of NY-ESO-1 in paraffin-embedded tissues has been developed and used to define the expression pattern of NY-ESO-1 in normal tissues and in a panel of human tumors. No normal tissue other than testis showed NY-ESO-1 reactivity, and expression in testis was restricted to germ cells particularly spermatogonia. In human tumors, the frequency of NY-ESO-1 antigen expression corresponds with past analysis of NY-ESO-1 mRNA expression e.g., 20–30% of lung cancers, bladder cancers and melanoma, and no expression in colon and renal cancer. Co-typing of NY-ESO-1 antigen and mRNA expression in a large panel of lung cancers showed a good correlation. There is great variability in NY-ESO-1 expression in individual tumors, ranging from an infrequent homogeneous pattern of staining to highly heterogeneous antigen expression. © 2001 Wiley-Liss, Inc.

Cancer/testis (CT) antigens are expressed in a range of malignant neoplasms, but with the exception of testis, not in normal tissues.1, 2 CT antigens elicit cellular3 as well as humoral4–6 immune responses. Ten genes or gene families have been identified that code for CT antigens.7 The first members of this class of tumor antigens, MAGE, BAGE and GAGE, were identified as targets for CD8 T cells in a patient with melanoma.8–10 Other CT antigens have been recognized by the cloning technique called SEREX, involving serological screening of cDNA expression libraries of tumors with patient sera as the source of antibody11 and by representational-difference analysis.12, 13 NY-ESO-1 is a classic CT antigen discovered during a SEREX analysis of an esophageal cancer.14 NY-ESO-1 is a 22 kD hydrophobic protein14 coded for by a gene in the Xq28 region.15 In normal tissues, expression of NY-ESO-1 mRNA as detected by RT-PCR is predominantly restricted to the testis, whereas in cancer NY-ESO-1 is found in a variable proportion of a wide range of different malignancies, including melanoma, cancers of the breast, ovary and lung.14 The strong immunogenicity of NY-ESO-1 is another distinguishing feature of the antigen; 40–50% of patients with advanced NY-ESO-1 mRNA positive tumors have humoral6,16,17 or CD8 T cell18 response to NY-ESO-1.

Current knowledge about CT antigen expression in normal and malignant tissues is mainly based on RT-PCR analysis. Mouse monoclonal antibodies (MAbs) to MAGE gene products have been generated,19–23 and these have been used to analyze the pattern of MAGE expression in tissues.24, 25 Mouse MAbs recognizing NY-ESO-1 antigen have been generated against NY-ESO-1 recombinant protein.6 In the present study, these NY-ESO-1 MAbs have been used to establish the expression pattern of NY-ESO-1 in normal and malignant tissues.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Antibodies

Two NY-ESO-1 monoclonal antibodies were described previously.6 Clone E978 (IgG1) was generated against a 23 kD NY-ESO-1 recombinant protein and clone ES121 (IgG1) was generated against a shorter, 14 kD NY-ESO-1 recombinant protein. Initial analysis with testis and selected tumor tissues indicated a similar immunohistochemical staining pattern with these two MAbs. Because ES121 showed both the best and most consistent results in immunohistochemical staining, it was chosen as the reagent for our study.

Tissues

Frozen and formalin-fixed paraffin-embedded tissues were provided by the Departments of Pathology of Memorial Sloan-Kettering Cancer Center and New York Hospital Cornell University Medical School. Frozen and 10% neutral buffered formalin-fixed paraffin-embedded tissues were used. The frozen specimens were embedded in O.C.T. (Tissue Tek, Torrance, CA) and snap-frozen in pre-cooled (−70°C) isopentane. Five μm cuts were performed from frozen or paraffin blocks and applied to histology slides for immunohistochemistry (Superfrost Plus, Fisher Scientific, Pittsburgh, PA). Frozen sections were fixed in cold acetone for 10 min and air-dried before use. Slides of paraffin cuts were dried overnight at 60°C.

A panel of normal tissues (Table I) and series of neoplastic lesions were examined. The tumors consisted of 52 lung tumors (Table II) and a variety of other malignant tumors (Table III), including 14 breast carcinomas, 11 metastatic melanomas, 10 carcinomas of the head and neck, 10 colonic carcinomas, 12 sarcomas, 9 carcinomas of the urinary bladder, and 10 renal cell carcinomas.

Table I. NY-ESO-1 Expression in Normal Tissues: Immunohistochemical Staining with MAB ES121 in Normal Tissues
TissueES121 immunoreactivity
Salivary gland
Pancreas
Liver
Esophagus
Stomach
Adipose tissue
Small bowel
Large bowel
Thyroid
Adrenal
Lung
Spleen
Lymph node
Kidney
Urinary bladder
Prostate
Breast
Ovary (adult)
Vagina
Placenta
Cervix
Testis++++ germ cells
Skin
Table II. NY-ESO-1 mRNA and MAB ES121 Immunoreactivity in Lung Tumors1
Diagnosis/histological type (total)RT-PCR pos/totalIHC pos/totalRT-PCR pos/IHC posRT-PCR pos/IHC negRT-PCR neg/IHC posRT-PCR neg/IHC neg
  • 1

    ACC, adenocarcinoma; LCC, large cell carcinoma; SQCC, squamous cell carcinoma.

ACC (33)6/336/3342225
LCC (9)2/94/9225
SQCC (9)4/92/9225
Carcinosarcoma (1)1/11/11
Total13/52 (25%)13/52 (25%)9 (17%)4 (8%)4 (8%)35/52 (67%)
Table III. NY-ESO-1 Expression in Human Cancers with NY-ESO-1 MAB ES121
Histological typeTotal testedES121 positive
Metastatic melanoma114
Breast carcinoma142
Urinary bladder carcinoma92
Synovial sarcoma32
Carcinomas of the head and neck100
Colonic carcinoma100
Leiomyosarcoma40
Liposarcoma50
Renal cell carcinoma100

Immunohistochemistry

Titration and reactivity assessments were done on frozen and paraffin-embedded testicular specimens. Testis was also used as a positive control tissue in all assays. For frozen tissues, several fixatives were tested including acetone and formaldehyde with variable fixation times. For paraffin-embedded specimens, several heat-based antigen retrieval methods were examined, including different retrieval solutions such as citrate buffer (pH 6.0, 10 mM), EDTA buffer (1 mM, pH 8.0) and commercial retrieval solutions like DAKO-TRS (DAKO, Carpinteria, CA), and DAKO high pH. Detection of the primary antibody was performed with a biotinylated horse-anti-mouse-secondary reagent (1:200; Vector Laboratories, Burlingham, CA) followed by an avidin-biotin-complex system (Vector, Elite) using diaminobenzidine tetrahydrochloride (DAB, Biogenex, San Ramon, CA) as a chromogen. Negative control slides were included in all tests.

RT-PCR

RT-PCR was done as previously described.14 Briefly, total RNA was extracted from 20 μm sections of frozen tissue samples of lung cancers. Testicular tissue was used as a positive control tissue. RNA was reverse transcribed and PCR-amplified with AmpliTaq Gold (Perkin Elmer, Norwalk, CT) for 30 cycles in a thermal cycler (Perkin Elmer) at an annealing temperature of 60°C. The following oligonucleotide primers were used: Sense 5′-ATG GAT GCT GCA GAT GCG G-3′; Antisense 5′-GGC TTA GCG CCT CTG CCC TG-3′. Both primers were synthesized commercially (Operon Technologies, Alameda, CA). RT-PCR products were visualized with ethidium bromide.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

In contrast to the poor reactivity of ES121 with frozen tissue sections, a reproducible immunoreactivity was observed in formalin-fixed, paraffin-embedded tissues using antigen retrieval techniques. The best staining was achieved with EDTA or the DAKO high pH-solution as retrieval solutions. Table I summarizes the immunohistochemical staining properties of ES121 for normal tissues. Testis was the only normal immunoreactive tissue, with germ cells in the seminiferous tubules showing strong staining. No staining was present in the interstitial tissue (Fig. 1a). In the seminiferous tubules, the staining was most intense in spermatogonia and in primary spermatocytes; no staining could be seen in the cells of the innermost tubular part. The staining intensity showed some intertubular variability and the pattern was primarily cytoplasmic. A nuclear staining component, however, was observed in some spermatogonia. Spermatids and Sertoli cells were immunonegative (Fig. 1a). No immunoreactivity was found in ovary or uterine tissues.

thumbnail image

Figure 1. Immunohistochemical staining of NY-ESO-1 antigen with MAb ES121, ABC method, DAB chromogen. Normal testis (a) and tumor tissues (b–f). (a) Seminiferous tubules with strong intratubular staining of mostly early spermatogenic cells, no reactivity with spermatids ([UPWARDS ARROW]) or Sertoli cell ([UPWARDS ARROW]), negative interstitial tissue (*). (b) Large cell lung carcinoma, low magnification showing staining of all carcinoma areas (*). (c) Homogeneous ES121 staining of malignant melanoma. (d) Heterogeneous ES121 staining in a malignant melanoma showing mostly negative tumor areas and focal staining of melanoma cells with strong nuclear reactivity. (e) Homogeneous staining of a synovial sarcoma. (f) High magnification of a breast carcinoma metastatic to a lymph node; negative lymphatic tissue (*) and negative carcinoma infiltrates (+) with focus of strongly ES121-reactive cells.

Download figure to PowerPoint

Table II summarizes the results of immunohistochemical staining and RT-PCR analyses of 52 lung tumors. Thirteen tumors (25%) showed ES121 immunoreactivity. The staining pattern varied from focal to homogeneous, with only 2 cases showing staining in more than 50% of the tumor (Fig. 1b). The remaining 11 positive cases showed a highly heterogeneous staining. A comparison of antigen and mRNA expression, revealed concordance in 44 cases. Among the disparate cases, 4 mRNA positive tumors showed no detectable ES121 staining and 4 ES121-positive cases showed no NY-ESO-1 mRNA. Because of the heterogeneity of NY-ESO-1 expression, we consider these discrepancies to be the result of sampling variation. With regard to the pattern of staining in lung cancer, reactivity was primarily cytoplasmic with nuclear staining being found only in focal areas. No staining was present in adjacent normal cells or tissues.

The typing results of neoplasms other than lung cancer is shown in Table III. In metastatic melanoma, 4/11 cases were positive for ES121 (Fig. 1c,d), with 1 tumor showing a homogeneous immunoreactivity (Fig. 1c). In carcinomas in of the urinary bladder and breast (Fig. 1f), 2/9 and 2/14 cases were ES121-positive respectively. No ES121 reactivity was found in carcinomas of the head and neck, renal carcinomas or colon carcinomas. With regard to the panel of 12 sarcomas, no NY-ESO-1 antigen expression was found in 4 leiomyosarcomas or 5 liposarcomas. Two of 3 synovial sarcomas, however, showed strong and homogeneous ES121 immunoreactivity (Fig. 1e). As in lung carcinomas, the staining was mostly cytoplasmic and in some cases also strongly nuclear (Fig. 1d).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

In this initial survey of human tumors, the overall frequency of NY-ESO-1 expression in human tumors as detected by MAb ES121 was comparable to our previous analysis of NY-ESO-1 mRNA expression by RT-PCR, e.g., 20–30% of lung cancers, melanoma, and bladder cancers with no expression in colon and renal cancer.14 In a direct comparison of RT-PCR and immunohistochemical typing of a panel of lung cancers, the NY-ESO-1 phenotype was concordant in 44 specimens and discordant in eight. Because of the heterogeneity of NY-ESO-1 expression (see below), it seems likely that this lack of correspondence between RT-PCR and immunohistochemistry in 8 cases resulted from sampling different areas of the tumor with disparate NY-ESO-1 expression, but these discrepancies need further study. This discrepancy has been observed in other studies involving the co-typing of CT antigen and mRNA expression.26, 27

Because of the strong and consistent staining of germinal cells in the testis, reactivity with spermatogonia can be used to standardize new serological reagents against NY-ESO-1 and other CT antigens. The NY-ESO-1 MAb used in our study, ES121, showed a predominantly cytoplasmic staining of spermatogonia, although some nuclear reactivity was also evident. Parallel studies with MAbs against MAGE gene products have shown strong nuclear staining of spermatogonia with the polyvalent MAGE -reagent 57B25 and the MAGE-4 reagent MAb R5,21 whereas MA454, a MAGE-1 specific MAb,19 showed an exclusively cytoplasmic reactivity.27 As the biological functions of CT antigens such as NY-ESO-1 and MAGE are unknown, the meaning of these different staining patterns of the male germ cells remains obscure. In our original report,14 NY-ESO-1 mRNA was found in ovary and low levels in endometrium. No ES121 staining, however, was detected in these organs. Future studies are necessary to determine whether female germ cells express NY-ESO-1 or other CT antigens.

As demonstrated in our study with NY-ESO-1 and in previous studies with MAGE,25, 26, 28 heterogeneous expression in tumors seems to be a characteristic feature of CT antigens. The expression of NY-ESO-1 in individual tumors can vary from single positive cells or small nests of cells to uniform staining of all tumor cells. The basis for this extensive intra and intertumoral variation in NY-ESO expression is unknown, and it will be important to extend the analysis of heterogeneous expression to the mRNA level with in situ hybridization and laser dissection microscopy. The central issue with regard to heterogeneous expression of CT antigens is whether the CT phenotype of an individual tumor type is a stable trait or a variable one. In other words, if CT-positive or CT-negative cells in a tumor showing heterogeneous CT-expression were examined over a period of hours or days, would there be a fidelity of expression or a modulation of the CT phenotype? One approach to this question would involve the cloning and propagation of cells from tumors showing different proportions of CT-positive and CT-negative cells and determine whether the in vivo CT+/CT− ratio changes or stays the same. The mechanism for gene activation or de-repression leading to aberrant CT antigen expression in malignant somatic cells is also an unresolved issue. Abnormal patterns of gene methylation associated with neoplasia have been related to MAGE-gene expression in cancer,29 but some cancers such as colon cancer, also known to have abnormal methylation, rarely express CT antigens.3, 25 Another possibility to account for aberrant expression in cancers is chromosomal translocation involving the coding region for CT antigens on the X chromosome, leading to gene activation and expression. The finding that a percentage of synovial cell sarcomas express NY-ESO-1, as well as MAGE antigens, offers a direct approach to testing this idea that chromosomal translocation is involved in CT expression. Synovial sarcoma show a characteristic t(X;18) translocation involving the X chromosome29,30 and different subtypes of translocations have been described and some of them assigned to particular histological subtypes.31–33 The relation between presence/type of translocation and NY-ESO-1 expression in individual synovial sarcomas is the object of current study.

Because of their high degree of tumor specificity, CT antigens, such as MAGE and NY-ESO-1, are promising targets for therapeutic cancer vaccines. With the generation of serological reagents detecting CT antigens, the antigenic phenotype of individual tumors can now be assessed at both the protein as well as the transcriptional level. Information about antigen expression, expression levels, and homogenous or heterogeneous antigen expression in tumors is an essential component of proper protocol design and patient selection, and is critical for the interpretation of results of clinical vaccine trials. The implication of heterogeneous antigen expression also needs consideration in the design of cancer vaccines, because of the expectation that antigen negative tumor cells will escape immune attack. For this reason, emphasis should be placed on polyvalent vaccines containing multiple CT antigens and other types of tumor antigens to prevent cancer cells from escaping the consequences of immune recognition.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  • 1
    Boon T, Coulie PG, Van den Eynde B. Tumor antigens recognized by T cells. Immunol Today 1997;18: 2678.
  • 2
    Old LJ, Chen YT. New paths in human cancer serology. J Exp Med 1998;187: 11637.
  • 3
    Van den Eynde BJ, Van der Bruggen P. T cell defined tumor antigens. Curr Opin Immunol 1997;9: 68493.
  • 4
    Sahin U, Tureci O, Pfreundschuh M. Serological identification of human tumor antigens. Curr Opin Immunol 1997;9: 70916.
  • 5
    Chen YT, Gure AO, Tsang S, Stockert E, Jager E, Knuth A et al. Identification of multiple cancer/testis antigens by allogeneic antibody screening of a melanoma cell line library. Proc Natl Acad Sci USA 1998;95: 691923.
  • 6
    Stockert E, Jager E, Chen YT, Scanlan MJ, Gout I, Karbach J, et al. A survey of the humoral immune response of cancer patients to a panel of human tumor antigens. J Exp Med 1998;187: 134954.
  • 7
    Chen YT, Scanlan MJ, Obata Y, Old LJ. Identification of human tumor antigens by serological expression cloning (SEREX). In: RosenbergSA. Principles and practice of biologic therapy of cancer, 3rd edition. Philadelphia: Lippincott, Williams & Wilkins, 2000: 55770.
  • 8
    Van der Bruggen P, Traversari C, Chomez P, Lurquin C, de Plaen E, Van den Eynde B, et al. A gene encoding an antigen recognized by cytolytic T-lymphocytes on a human melanoma. Science 1991;254: 16436.
  • 9
    Boel P, Wildmann C, Sensi ML, Brasseur R, Renauld JC, Coulie P, et al. BAGE: a new gene encoding an antigen recognized on human melanomas by cytolytic T lymphocytes. Immunity 1995;2: 16775.
  • 10
    Van den Eynde B, Peeters O, De Backer O, Gaugler B, Lucas S, Boon T. A new family of genes coding for an antigen recognized by autologous cytolytic T lymphocytes on a human melanoma. J Exp Med 1995;182: 68998.
  • 11
    Sahin U, Tureci O, Schmitt H, Cochlovius B, Johannes T, Schmits R, et al. Human neoplasms elicit multiple specific immune responses in the autologous host. Proc Natl Acad Sci USA 1995;92: 118103.
  • 12
    Lethe B, Lucas S, Michaux L, De Smet C, Godelaine D, Serrano A, et al. LAGE-1, a new gene with tumor specificity. Int J Cancer 1998;76: 9038.
  • 13
    Gure AO, Stockert E, Arden KC, Boyer AD, Viars CS, Scanlan MJ, et al. CT10: a new cancer-testis (CT) antigen homologous to CT7 and the MAGE family, identified by representational-difference analysis. Int J Cancer 2000;85: 72632.
  • 14
    Chen YT, Boyer AD, Viars CS, Tsang S, Old LJ, Arden KC. Genomic cloning and localization of CTAG, a gene encoding an autoimmunogenic cancer-testis antigen NY-ESO-1, to human chromosome Xq28. Cytogenet Cell Genet 1997;79: 23740.
  • 15
    Chen YT, Boyer AD, Viars CS, Tsang S, Old LJ, Arden KC. Genomic cloning and localization of CTAG, a gene encoding an autoimmunogenic cancer-testis antigen NY-ESO-1, to human chromosome Xq28. Cytogenet Cell Genet 1997;79: 23740.
  • 16
    Jager E, Chen YT, Drijfhout JW, Karbach J, Ringhoffer M, Jager D, et al. Simultaneous humoral and cellular immune response against cancer-testis antigen NY-ESO-1: definition of human histocompatibility leukocyte antigen (HLA)-A2-binding peptide epitopes. J Exp Med 1998;187: 26570.
  • 17
    Jager E, Stockert E, Zidianakis Z, Chen YT, Karbach J, Jager D, et al. Humoral immune responses of cancer patients against “Cancer-Testis” antigen NY-ESO-1: correlation with clinical events. Int J Cancer 1999;84: 50610.
  • 18
    Jager E, Jager D, Karbach J, Chen YT, Ritter G, Nagata Y, et al. Identification of NY-ESO-1 epitopes presented by human histocompatibility antigen (HLA)-DRB4*0101–0103 and recognized by CD4(+) T lymphocytes of patients with NY-ESO-1-expressing melanoma. J Exp Med 2000;191: 62530.
  • 19
    Chen YT, Stockert E, Chen Y, Garin-Chesa P, Rettig WJ, Van der Bruggen P, et al. Identification of the MAGE-1 gene product by monoclonal and polyclonal antibodies. Proc Natl Acad Sci USA 1994;91: 10048.
  • 20
    Schultz-Thater E, Juretic A, Dellabona P, Lüscher U, Siegrist W, Harder F, et al. MAGE-1 gene product is a cytoplasmic protein. Int J Cancer 1994;59: 4359.
  • 21
    Takahashi K, Shichijo S, Noguchi M, Hirohata M, Itoh K. Identification of MAGE-1 and MAGE-4 proteins in spermatogonia and primary spermatocytes of testis. Cancer Res 1995;55: 347882.
  • 22
    Carrel S, Schreyer M, Spagnoli G, Cerottini JC, Rimoldi D. Monoclonal antibodies against recombinant-MAGE-1 protein identify a cross-reacting 72-kDa antigen which is co-expressed with MAGE-1 protein in melanoma cells. Int J Cancer 1996;67: 41722.
  • 23
    Jurk M, Kremmer E, Schwarz U, Forster R, Winnacker E-L. MAGE-11 protein is highly conserved in higher organisms and located predominantly in the nucleus. Int J Cancer 1998;75: 7626.
  • 24
    Kocher T, Schultz-Thater E, Gudat F, Schaefer C, Casorati G, Juretic A, et al. Identification and intracellular location of MAGE-3 gene product. Cancer Res 1995;55: 22369.
  • 25
    Jungbluth AA, Busam KJ, Kolb D, Iversen K, Coplan K, Chen YT, et al. Expression of MAGE-antigens in normal tissues and cancer. Int J Cancer 2000;85: 4605.
  • 26
    Fischer C, Gudat F, Stulz P, Noppen C, Schaefer C, Zajac P, et al. High expression of MAGE-3 protein in squamous-cell lung carcinoma. Int J Cancer 1997;71: 111921.
  • 27
    Jungbluth AA, Stockert E, Chen YT, Kolb D, Iversen K, Coplan K, et al. Monoclonal antibody MA454 reveals a heterogeneous expression pattern of MAGE-1 antigen in formalin-fixed paraffin embedded lung tumors. Br J Cancer 2000;83: 4937.
  • 28
    Cheville JC, Roche PC. MAGE-1 and MAGE-3 tumor rejection antigens in human germ cell tumors. Mod Pathol 1999;12: 9748.
  • 29
    De Smet C, de Backer O, Faraoni I, Lurquin C, Brasseur F, Boon T. The activation of human gene MAGE-1 in tumor cells is correlated with genome-wide demethylation. Proc Natl Acad Sci USA 1996;93: 714953.
  • 30
    Limon J, Mrozek K, Mandahl N, Nedoszytko B, Verhest A, Rys J, et al. Cytogenetics of synovial sarcoma: presentation of ten new cases and review of the literature. Genes Chromosomes Cancer 1991;3: 33845.
  • 31
    Gure AO, Tureci O, Sahin U, Tsang S, Scanlan MJ, Jager E, et al. SSX: a multigene family with several members transcribed in normal testis and human cancer. Int J Cancer 1997;72: 96571.
  • 32
    Lasota J, Jasinski M, Debiec-Rychter M, Szadowska A, Limon J, Miettinen M. Detection of the SYT-SSX fusion transcripts in formaldehyde-fixed, paraffin-embedded tissue: a reverse transcription polymerase chain reaction amplification assay useful in the diagnosis of synovial sarcoma. Mod Pathol 1998;11: 62633.
  • 33
    Argani P, Zakowski MF, Klimstra DS, Rosai J, Ladanyi. Detection of the SYT-SSX chimeric RNA of synovial sarcoma in paraffin-embedded tissue and its application in problematic cases. Mod Pathol 1998;11: 6571.
  • 34
    Skytting B, Nilsson G, Brodin B, Xie Y, Lundeberg J, Uhlen M, et al. A novel fusion gene, SYT-SSX4, in synovial sarcoma [letter]. J Natl Cancer Inst 1999;91: 9745.