`
`
`
`
`
`
`PHIGENIX
`PHIGENIX
`Exhibit 1002
`Exhibit 1002
`
`
`
`
`
`inhibits uptake of the ligand. In our case, however,
`chloroquine primarily affected degradation of ”’1-
`labeled Mot-MESA as shown from the following
`observation. Hamster peritoneal macrophages were
`incubated at 37°C in medium containing 'z’I-la-
`beled Mtx-MBSA (10 pg per milliliter of protein).
`The cellular content of radioactivity and that re-
`leased into medium (acid-soluble) were measured as
`a function of time in the presence and absence of
`chloroquine (3 uM). Cellular content of radioactiv‘
`ity continued to increase in chloroquine-treated
`cultures but release of acid-soluble radioactivity in
`the medium was arrested. These results suggested
`that at this concentration chloroquine inhibited the
`13 September 1988; accepted 20 February 1989
`degradation of Mot-MBSA and not its uptake.
`
`
`18. M. Rabinowitch, v. Zilberfarb, c. Ramazeillet, }.
`Exp. Med. 163, 520 (1986).
`.
`'
`19. S. Sehgal and S. K. Arora, Ind. j. Med. Res. 82, 202
`(1985).
`20. B. Phillips and I. C. Garret, Nature 220, 1140
`(1968).
`21. I. D. Berrmn and D. I. Wyler, ]. Infect. Dis. 142, 83
`(1980).
`22. We thank V. K. Kalra and G. C. Mishra for critically
`reviewing the manuscript, and the Council of Scien-
`tific and Industrial Research and Indian Council of
`Medical Research for award of fellowships to A.M.
`and G.C.
`
`,
`
`i3.
`
`Studies of the HER-Z/mem Prom-oncogene in Hmnan
`Breast and Gvarian Cancer
`
`
`DENNIS J. SLAMON,* WILLIAM GODOLPHIN, ILOVELL A. JONES,
`IOHN A. HOLT, STEVEN G. WONG, DUANE E. KEITH, WENDY J. LEVIN,
`SUSAN G. STUART, IUDY UDOVE, AXEL ULLRICH, MICHAEL F. PRESS
`
`Carcinoma of the breast and ovary account for one-third of all cancers occurring in
`women and together are responsible for approximately oneaquarter of cancer-related
`deaths in females. The I-IER-Z/neu prom-oncogene is amplified in 25 to 30 percent of
`human primary breast cancers and this alteration is associated with disease behavior.
`In this report, several similarities were found in the biology of I-IER-Z/nzu in breast
`and ovarian cancer, including a similar incidence of amplification, a direct correlation
`between amplification and over-expression, evidence of tumors in which overexpres-
`sion occurs without amplification, and the association between gene alteration and
`clinical outcome. A comprehensive study of the gene and its products (RNA and
`protein) was simultaneously performed on a large number of both tumor types. This
`analysis identified several potential Shortcomings of the various methods used to
`evaluate HER-Z/neu in these diseases (Southern, Northern, and Western blots, and
`immunohistochemistry) and provided information regarding considerations that
`should be addressed when studying a gene or gene product in human tissue. The data
`presented fiirther support the concept that the I-IER-Z/neu gene may be involved in the
`pathogenesis of some human cancers.
`
`was observed. Finally, the conjugated drug
`caused 90% reduction in the size of the
`
`lesion 11 days after the initiation of drug
`treatment. The greatest effect on the regres-
`sion of the lesions by the conjugated drug
`was observed at a dose of 1 mg/kg per
`footpad. The lack of effect at higher concen-
`trations probably reflects saturation of the
`receptor-mediated uptake process for Mtx-
`MBSA. The footpad regressed to nearly
`normal size when Mtx—MBSA was used. In
`contrast, administration of free Mtx did not
`significantly afiect the footpad lesion. The
`lesions did not reappear even 4 weeks after
`the last injection of Mtx-MBSA. During the
`experimental period all
`the animals
`re-
`mained healthy with no apparent weight
`loss. No antibody against MBSA or Mtx-
`MBSA was detectable in these animals after
`3 weeks as determined by the Ouchterlony
`immunodiflirsion technique.
`In conclusion, our results show that effec-
`tive delivery of drug to macrophages can be
`achieved by using the “scavenger” receptor-
`mediated endocytic pathway to achieve se-
`lective killing of intracellular parasites resid-
`ing in macrophages, both in vitro and in
`vivo. A similar approach may be useful for
`effective delivery of drugs in the treatment
`of other diseases in which macrophages are
`the primary target, including tuberculosis,
`leprosy, monocytic leukemia, and heavy
`metal storage diseases.
`
`
`REFERENCES AND NOTES
`
`Int. Rev. Cytal (Suppl) 14, 267
`
`,-
`M. L. Chance, Br. Med. 1. 283, 1245 (1981).
`2. 1. A. Walsh and K. 5. Warren, N. Eng. 1. Med. 301,
`967 (1979).
`3. K. P. Chang,
`(1983).
`4. I. D. Berman, in Leishmaniasi‘s, K. P. Chang and R.
`S. Bray Eds. (Elsevier, Amsterdam, 1985), vol. 1,
`pp. 111—138.
`5. I. I. Mart, in Parasitic Diseases, I. M. Mansfield, Ed.
`(Dekker, New York, 1984), vol. 2, pp. 201—227.
`6. C. D. V. Black, G. I. Watson, R. J. Ward, Trans. R.
`Soc. Trop. Med. Hyg. 71, 550 (1977).
`7. ,C. R. Alving, E. A. Steck, W. 1.. Hanson, P. S.
`Ioizeaux, W. L. Chapman, Ir., Lie Sci. 22, 1021
`(1978).
`8. R. R. C. New, M. L. Chance, 5. C. Thomas, W.
`Peters, Nature 272, 55 (1978).
`9. R. R. C. New, M. L. Chance, S. Heath, 1. Antimi-
`crob. Chemoihrr. 8, 371 (1981).
`in Liposame
`10. C. R. Alving and G. M. Swartz, Ir.,
`Technology, G. Gregoriadis, Ed. (CRC Press, Boca
`Raton, FL, 1984), vol. 2, pp. 55—68.
`11. C. R. Alving er al., Prat. Natl. Amd Sci. U.S.A. 75,
`2959 (1978).
`12. I. L. Goldstein, Y. K. Ho, S. K. Basu, M, S. Brown,
`ibid. 76, 333 (1979).
`13. M. S. Brown, S. K. Basu, C. R. Falk, Y. K. Ho, I. L.
`Goldstein,_]. Supramol. Strurr. 13, 67 (1980).
`14. O. Stein and Y. Stein, Biochim. Biophys. Add 620,
`631 (1980).
`15. A. N. Glazier, R. S. Delange, D. S. Sigman, Eds,
`Chemical Modification of Proteins
`(North-Holland,
`Amsterdam, 1975). pp. 79—81.
`16. H. I. P. Ryser and W. C. Shen, Proc. Natl. Acad. Sci.
`U.S.A. 75, 3867 (1978).
`17. A. Mukhopadhyay, G. Chaudhuri, S. K. Basu, un-
`published data. In some receptor systems chloro-
`quine, especially at relatively high concentrations,
`
`12 MAY 1989
`
`ROTO-ONCOGENES
`
`REPRESENT
`
`A
`
`P family of normal cellular genes that
`
`were identified on the basis of their
`
`similarity to genetic sequences with known
`tumorigenic or transforming potential (1).
`Considerable circumstantial evidence now
`exists that alterations in either the structure,
`
`copy number, or expression of one or anoth—
`er of these genes may play a role in the
`pathogenesis of some human malignancies
`(2). One such gene, called HER-Zlneu or c-
`erb B2, was first identified by transfection
`studies in which NIH 3T3 cells were trans—
`
`formed with DNA from chemically induced
`rat neuroglioblastomas (3). The gene en-
`codes a protein that has extracellular, trans-
`mcmbrane, and intracellular domains (4)
`which is consistent with the structure of a
`
`growth factor reception.
`Recently, we found a 28% incidence of
`amplification of HER-Zlneu in 189 primary
`human breast cancers (5). Patients with mul-
`tiple copies of the gene in DNA from their
`tumors had a shorter time to relapse as well
`as a shorter overall survival indicating that
`
`gene amplification was prognostic for dis-
`ease behavior in these individuals. More-
`
`over, multivariate survival analysis showed
`HER-Z/neu amplification to be more predic»
`tive for clinical outcome than all other
`
`known prognosticators with the exception
`of positive lymph nodes (5). Since that
`initial report, a number of studies have been
`published on the amplification of this gene
`
`D. Slamon, S. G. Wong, D. E. Keith, W. I. Levin,
`Division of Hematology-Oncolo , Department of
`Medicine, and the Ionsson Compre ensive Cancer Cen-
`ter, U.C.L.A. School of Medicine, Los Angeles, CA
`90024.
`t of Clinical Chemistry, Van-
`W. Godolphin, De
`couver General Hospital, Vancouver, Canada VSZIM9,
`L. A. Jones, Department of7G;necology, M. D. Ander-
`son Hospital, Houston, TX 030.
`l. A. Holt, Department of Obstetric and Gyneco
`,
`University of Chicago Medical Center, Chicago, L
`60637.
`S. G. Smart, Triton Biosciences, Inc, Alameda, CA
`94501.
`I. Udovc and M. F. Press, Department of Pathology,
`University of Southern Califorrua, School of Medicine,
`Los Angeles, CA 90033.
`, Genen-
`A. uunch, D artment of Molecular Biology
`
`tech, Inc., Sou
`San Francisco, CA 94080.
`
`'To whom correspondence should be addressed.
`
`WIS MATERIAL MAY BE PROTECTED
`1811 COPYRIGHT LAW (17 US. CODE)
`
`REPORTS 707
`
`.__,.
`
`PHIGENIX
`
`Exhibit 1002-01
`
`
`
`:r‘
`
`/
`
`in human breast cancer and the association
`
`of gene amplification with clinical behavior
`(6—8). There is considerable variability in
`both the reported incidence of amplification
`and the correlation of gene amplification
`with patient outcome (5—10). Some groups
`have found amplification rates as low as
`10% and no correlation to outcome data
`
`while others have found rates as high as
`33% and a strong association with outcome
`(7, 10). Given the variable natural history
`arid heterogeneity of human breast cancer,
`all studies published to date suffer from a
`similar problem, which is small numbers of
`
`evaluated cases (5—10). Perhaps a more sig-
`nificant shortcoming of most prior studies
`of oncogenes in human tumors including
`our own is that only one aspect of the gene
`in question (DNA, RNA, or protein status)
`is evaluated (5, 11, 12). The potential errors
`introduced by dilution of tumor cell macro-
`molecules with macromolecules from sur-
`
`rounding normal vascular, stromal, or in-
`flammatory cells is a general problem in
`human tumor tissue and a particular prob-
`lem in breast cancer where these non-malig-
`nant cells can account for more than 50% of
`
`the tissue. All solid matrix—blotting tech-
`
`SOUTH ERN
`
`
`
`NDH‘I’HCHN
`"VIM HIS?
`
`Flg. 1. Examples of the correlation between HER-Z/neu gene amplification and expression. Southern
`blot analyses show the 12.5-kb HER-2/neu band seen with Eco RI cut DNA. All DNA samples were
`checked for integrity of high molecular weight species and samples showing evidence of DNA
`degradation were not evaluated (9). Southern blots and HER-Zineu copy number detemiinau'ons were
`as described (5, 9). Hybridizations were done with a 1.4-kb, 3' Eco RI fragment of the human HER-
`2/neu cDNA clone. Northern blot analyses show the 4.5-kb HER-Zlneu transcript. All RNA samples
`were checked for integrity of the 28S and 185 ribosomal RNA species. Total RNA (20 ug) from
`samples with intact RNA were run on an agarose gel, transferred to a nylon filter, and hybridized with a
`32P-labeled HER—Zlneu probe as described (11). Samples with intact DNA but showing some
`degradation ofthe RNA were evaluated by slot-blot analysis by loading 12 ug of total RNA on the filter
`and hybridizing as above. Samples showing degradation of both DNA and RNA were not used for
`RNA analysis. The relative optical density (O.D.) of bands was detemiined by soft laser densitometry
`scanning and ranged from a low of 0.1 O.D. units to a high of 3.8 O.D. units. Tumors were grouped
`into RNA expression categories as follows: 0.1 to 0.5 O.D. units, 1+; >0.5 to 1.0 O.D. units, 2+;
`>10 to 1.5 O.D. units, 3+; and >15 0_.D. units, 4+. Western blot analyses show the 185-kD HER-
`2/neu protein band. The relative O.D. of bands ranged from a low of 0.1 O.D. units to a high 4.5 O.D.
`units. Tumors were grouped into protein expression categories as follows: 0.1 to 0.5 O.D. units, 1+;
`>0.5 to 1.0 O.D. units, 2+; >10 to 1.5 O.D. units, 3+; >l.5 O.D. units, 4+. Immunohistochemical
`analysis was done as described (20) with the anti-HER-Z/neu specific antibody and frozen sections (14).
`Tissues were scored and placed in one of the four staining categories shown on the basis of the relative
`level of specific staining as judged by microscopic examination as follows: negative to weak, 1+, 2+,
`3+. The five samples analyzed here are arranged in identical order from left to right in each panel. The
`HER-21m copy numbers (from left to right) were >20, 5 to 20, 2 to 5, l, and 1, respectively. The
`corresponding O.D. readings from the Northern blots were 4.3, 1.4, 0.9, 0.2, and 2.8, respectively. The
`corresponding O.D. readings from the Western blots were 2.0, 1.1, 0.6, 0.12, and 2.1, respectively.
`The corresponding immunohistochemistry readings were 3+, 2+, 1+, weak, and 3+, respectively.
`
`niques (Southern, Western, or Northern)
`are susceptible to these errors. Similarly,
`these techniques cannot determine; whether
`an observed alteration is specific for only the
`malignant cells in the tissue.
`Studies of the rat neu gene isolated from
`the chemically induced neuroglioblastomas
`revealed it to contain a single mutation in ’
`the transmembrane domain that difierentiat-
`
`ed it from the nontransforrning neu gene
`found in normal rat tissues. This mutation is
`
`critical in the conversion of the normal gene
`into a transforming gene (13). To address
`
`
`
`Fig. 2. Comparison of irnmunoperoxidase stain- _-
`ing with Western blot on stroma-rich breast
`cancer. The inset (upper left) is a hematoxylin-
`eosin stain of a breast tumor rich in stromal tissue.
`Note the absence of significant numbers of tumor
`cells. Large middle panel
`is
`the immunoper-
`oxidase staining of tumor cells (TC) found in this
`tissue. The staining is 3+, placing this tumor in
`the highest category of HER-Z/neu expression as
`judged by immunostaining. Southem analysis re-
`vealed two to five copies of the gene in the DNA
`and Northem blot analysis gave an O.D. reading
`of 0.6 (2+). Western blot of protein from this
`sample is shown in lane A of the lower right inset.
`The O.D. reading for this sample was 0.18 (1+),
`while the Western blot of protein from another
`specimen with amplified HER-2/neu and greater
`numbers of tumor cells is shown in lane B. The
`O.D. reading for this tumor (lane B) was 3.2
`(4+). Eight of the 11 tumors found to have
`inappropriately low Western blot data in compar-
`ison to other data were similar in that they were
`stromal-rich tumors.
`
`
`
`Fly. 3. Comparison of immunohistochemical
`staining of HER-Zlncu protein in the same breast
`cancer specimen evaluated with frozen tissue and
`'fonnalin-fixed, paraffin-embedded u'ssue. Tissue
`shown in panels A and B are from the same tumor
`which was found to have 5 to 20 copies of the
`gene and a 2+ expression level by Northern and
`Western blot analyses. Panel A is the frozen
`section and shows 2+ immunostaining. Panel B is
`the formalin-fixed, parafiin-embedded section of
`the same tumor and shows negative immuno-
`staining.
`
`SCIENCE, VOL. 244
`
`
`
`PHIGENIX
`
`Exhibit 1002-02
`
`
`
`
`
`the question of whether or not a similar
`change had occurred in an amplified HER-
`2/neu gene found in human breast cancer,
`cDNA clones from tumor tissue rather than
`
`lines were generated by means of the
`cell
`Okayama-Berg vector (14, 15). Tissue was
`used to circumvent acquired genetic changes
`which can occur in vitro. Analysis of the
`transmembrane domain of eight clones from
`two separate tumors shoWed the identical
`sequence. There was no glutamic acid for
`valine substitution as reported in the trans-
`forming neu gene from the chemically in-
`duced rat tumors (13). There was, however,
`a neutral change of isoleucine for valine at
`position 655 in the transmembrane domain,
`which is similar to the sequence found in
`breast cancer cell lines (16). Analysis of the
`entire coding sequence of full-length clone
`and comparison with the published placen-
`tal sequence (3) showed no other significant
`changes (14). These data are consistent with
`the concept that overexpression of a normal
`HER-Z/neu gene product rather than muta-
`tion to an abnormal gene may be an impor-
`tant pathogenic event for some rumors.
`Having obtained and sequenced a full-
`length cDNA clone from a human breast
`tumor, we next wanted to generate antisera
`to the human gene product. The generation
`and characterization of this antiserum is
`
`described elsewhere (14) and the antibody is
`capable of identifying the gene product both
`by Western blot analysis of tissue homoge-
`nates and immunohistochemically in tissue
`sections.
`
`In the current study, we collected a total
`of 668 human breast cancer specimens. Of
`these, 526 had sufficient clinical foUow-up
`to allow for evaluation of an association
`
`between gene amplification and disease out-
`come. As in our initial study (5), we per-
`formed Southern analysis on samples with-
`out knowledge of the clinical outcome. All
`DNA blots were stripped and reprobed with
`both p53 and myeloperoxidase probes to
`evaluate the relative loading of DNA in each
`lane and to exclude the possibility that “am-
`plification” was caused by partial or com-
`plete duplication of chromosome 17 (9).
`Blots were scanned by soft laser densitome-
`try and the level of HER-Z/neu amplification
`was determined by the ratio of the HER-
`2/neu signal relative to the single-copy p53
`signal.
`We evaluated 345 patients with node-
`positive disease in a blinded fashion (Table
`1). Of these, 101 (27%) had evidence of
`HER-2/neu amplification. Univariate surviv-
`al analysis
`showed amplification of the
`HER-Zlneu gene to be a significant predic-
`tor of both disease-free survival and overall
`
`survival for these patients (Table 1). Tumor
`size was slightly better than HER-Zlneu
`
`[2 MAY 1989
`
`Table 1. Univariate and multivariate survival analyses comparing disease—free (relapse) and overall
`survival to prognostic factors in 345 node-positive breast cancer patients. Statistical analyses were
`performed by the X2 test and by Cox’s partially nonparametric regression analysis to evaluate the
`predictive power of various combinations and interactions of prognostic factors in a multivariate
`manner as described (5). Prognostic parameters evaluated include number of nodes (Nodes), HER-Zl
`neu gene amplification (HER-Zineu), estrogen receptor (ER), progesterone receptor (PGR), size of
`primary tumor (Size), and age of patient at diagnosis (Age). The median follow up was 57 months (60
`months for those still alive).
`Disease free survival
`Overall survival
`
`Uni-
`Multi-
`Uni-
`Multi-
`
`variate (P)
`variate (P)*
`variate (P)
`variate (P)*
`
`<0.0001 [0.0912 : 0.0346]
`<0.0001
`<0.0001 [0.0818 1- 0.0214]
`<0.0001
`Nodes
`0.045 [0.0864 1. 0.0288]
`0.041
`0.006 [0.1142 1 0.0413]
`0.01
`HER-Zlneu
`0.157
`0.091
`0.60
`0.235
`ER
`0.24
`0.20
`0.07
`0.045
`PGR
`0.16
`0.006
`0.15
`0.003
`Size
`
`
`
`
`0.92 0.96 0.20Age 0.11
`
`*Regression coefficients t SE are shown in square brackets.
`
`amplification in the univariate analysis but
`lost its significance on multivariate analysis,
`which indicates that it was not independent
`of nodal status (Table 1). Multivariate analy—
`sis showed HER—Z/neu amplification to be
`an independent predictor of both disease
`relapse and overall survival (P é 0.006 and
`P = 0.045, respectively) and superior to all
`other known prognostic factors with the
`exception of a number of positive lymph
`nodes for this group of patients (Table 1).
`We also evaluated DNA from tumors of 181
`
`node-negative patients with a median fol-
`low-up of 59 months (62 months for those
`still alive). Of these, 45 (25%) had amplifi-
`cation of the HER—Z/neu gene. Univariate
`and multivariate analysis did not show an
`association between gene amplification and
`disease outcome in this group of patients.
`There were 187 tumor samples of sulfi-
`cient size and integrity to allow for multiple
`studies in the same specimen. This group of
`specimens was representative of the overall
`group in that lesions from both node—nega-
`tive as well as node-positive patients were
`included as well as tumors of varying sizes
`(<1 cm to >25 cm). The availability of a
`cDNA clone for HER-Z/neu from a human
`
`tumor as well as antisera that could identify
`the protein in both Western blots and tissue
`sections allowed for a comprehensive evalu-
`ation of the gene and its products (RNA and
`protein) in these tissues. Such a study ad-
`dresses
`several
`critical
`issues
`regarding
`HER-Z/neu in human breast cancer. First,
`the correlation between a given level of
`amplification and relative expression for
`both RNA and protein is important. Some
`genes that are amplified in breast cancer,
`such as erb A (6), are not expressed; these
`genes may serve as usefill markers but are
`unlikely to be involved in the pathogenesis
`of the disease. Second, it should be possible
`to address the issue of whether amplification
`and over-expression of HER-2/neu is specif-
`
`ic to tumor cells in these specimens. When
`blotting techniques alone are used, there is
`the risk of dilutional effects, which make it
`impossible to distinguish signals from tu-
`mor cells versus those from normal cells.
`
`Third, this approach should give some indi—
`cation of the relative strengths and weak-
`nesses of the various techniques used in
`assessing the status of the HER-Z/neu gene
`and its products in the same primary human
`tissue. Studies at both the RNA and protein
`levels are important since there are examples
`of human tumors in which transcripts of a
`particular gene are present but no protein
`can be detected (17). Fourth,’ if gene expres-
`sion correlates closely with amplification, a
`separate assessment of the incidence of am-
`plification can be made.
`The samples used for the comprehensive
`analysis were between 0.5 and 1 g in size
`and had been stored for various periods of
`time at —70° to — 140°C. Tumor tissue was
`
`fractured in liquid nitrogen to obtain a
`representative piece suitable for cryostatic
`sectioning and immunohistochemical analy-
`sis. The remainder of the entire specimen
`was ground to powder in liquid nitrogen
`with a mortar and pestle. This process al-
`lowed for an even distribution of tumor cells
`
`in all subsequent analyses. A portion of the
`tissue powder (50 mg) was stored for West-
`ern blot analysis and the remainder was
`extracted for DNA and RNA simultaneous-
`
`ly. As in our clinical correlation studies, each
`procedure was conducted separately and
`without knowledge of the results obtained
`by other modalities of evaluation. Southern
`blots with appropriate p53 and myeloperox-
`idase controls were performed and HER—
`2/neu copy number determined as described
`(Fig. 1). Northern blots were performed on
`20 ug of total RNA extracted from the
`sample and were analyied for the presence
`and relative intensity of the 4.5-kb HER-
`2/neu messenger RNA using soft laser densi-
`REPORTS 709
`
`PHIGENIX
`
`Exhibit 1002-03
`
`
`
`
`
`‘1‘.
`
`.
`
`I
`
`I
`
`'
`
`“
`
`J
`
`.1
`
`/
`
`i
`
`a". ’-
`
`tomen'y and grouped into one offour RNA
`expression categories depending on the den-
`sitometry results (Fig. I); Western blots
`were performed on .100 pg of total protein
`extracted from each sample and were ana-
`lyzed for the presence and relative intensity
`of the 185—kD HER-2/neu gene product by
`soft
`laser densitometry and grouped into
`one of four protein expression categories
`(Fig. I). Immunohistochernical analysis was
`performed by an irnmunoperoxidase stain-
`ing technique on frozen sections of the
`tumor tissue. The immunohistochemical
`
`specificity of the antibody was previously
`shown (14).
`After completion of testing of all samples
`by each modality, the study was unblinded
`and a direct correlation analysis between
`gene amplification and expression was per-
`formed. Ofthe 187 samples evaluated in this
`manner there was almost complete concor—
`dance of the data, with a few notable excep-
`tions. Fifty-one samples (27%) of the cases
`were called amplified by the Southem blot
`analyses, and in every instance the amplified
`samples were shown to over—express the
`HER-2/neu gene product relative to nonam-
`plified cases. In 46 of the amplification-
`positive samples (90%) there was complete
`concordance of all tests, that is, amplifica-
`tion correlated with overexpression as deter-
`mined by Northern, Western and immuno-
`histochemistry. In all 51 cases with ampli-
`fication, two of the three measures of ex-
`pression were concordant in showing over-
`expression.
`The Western blot analysis was most dis-
`cordant, being inconsistent with data ob—
`tained by Southern blot, Northern blot, and
`irnmunohistochemistry in 11 of the 187
`cases (6%). A possible explanation for this
`phenomenon is
`that samples showing a
`weak signal by Western blot may do so
`because of a relatively large amount of stro-
`ma in the tissue. In 8 of the 11 cases in
`
`which the Western was discordant, histolog-
`ic examination confirmed the presence of
`excessive stromal elements (Fig. 2). DNA
`and RNA analyses were less sensitive to this
`problem since there was evidence of amplifi-
`cation and increased transcript levels in these
`tumors. The increased susceptibility of the
`Western blot to dilutional effects is likely
`due to the fact that large amounts of noncel-
`lular connective tissue can substantially con-
`tribute to the total protein in a sample by
`adding significant amounts of extracellular
`matrix proteins such as collagen to the ly-
`sate. The stroma, however, is relatively poor
`in cellularity and will make only minimal
`contributions to the total DNA or RNA
`
`extracted from the same specimen.
`The Northern blot analysis was discor-
`dant in only four cases (2%) judged ampli—
`
`710
`
`fied at the DNA level, and the immunohis-
`tochemical data were inconsistent in only
`two cases (1%). In each of these cases, the
`other two modalities used showed over-
`
`expression. The correlation between amplifi-
`cation and overexpression (P < 0.0001)
`shown in the comprehensive analysis confirms
`previously published amplification rates of 25
`to 30% for HER-2/neu in human breast
`
`mncer. There were 18 cases (10%) which
`were judged as single copy by DNA analysis
`but which showed clear over-expression at the
`RNA and protein levels (Fig. 1, lane 5). These
`cases may represent examples of alterations
`that occur in connol mechanisms for gene
`expression, rather than increases in gene copy
`number. Altematively, these cases may repre-
`sent instances of true gene amplification at
`low levels (two- to fivefold) which are missed
`because of dilution of the tumor cell DNA
`
`with DNA from nonmalignant tissue.
`Some studies have used only immuno-
`staining of formalin-fixed, paraffin-embed-
`ded tissue to assess the status of the HER-
`
`2/neu product (8, 10, 18). This approach
`presents some problems, since fixation and
`embedding of tissue frequently decreases or
`totally destroys reactivity for many antigens
`(19). Furthermore, the degree of loss can
`vary considerably depending on the dura-
`tion of fixation with the same fixative (19,
`20). For 42 of the 187 cases in the compre-
`hensive analysis,
`formalin-fixed, parafiin-
`embedded tissue was available. This allowed
`
`for a comparison of immunohistochemical
`staining of the HER-Zlneu protein in the
`same specimen prepared in different ways.
`Evaluation of HER-2/neu immune-reactiv-
`
`ity showed that in virtually every case there
`was some decrease in immunohistochemical
`
`staining with the rabbit polyclonal antise—
`rum when comparing fixed to frozen tissue.
`In tumors expressing very high levels (3 + to
`
`4+), the protein was visible by immuno-
`staining in tissue prepared by either method.
`The problem was more significant in sam-
`ples expressing moderate levels of protein
`(2+)
`since many completely lost
`their
`immunohistochemical reactivity when fixed
`in formalin and embedded in paraffin (Fig.
`3). The majority of these tumors were am-
`plified for the HER-Z/neu gene at levels of
`two— to fivefold. However, occasional
`tu-
`mors with 5- to 20-fold amplification that
`expressed moderate to high levels of protein
`appeared negative in fixed tissue.
`To determine if the same or similar alter-
`
`ations in the HER—Z/neu gene found in
`breast cancer occurred in other human ma-
`
`lignancies, we screened DNA from 60 non—
`small cell lung cancers, 40 colon carcinomas,
`35 neuroblastomas, and 18 ovarian cancers. .
`There was no significant incidence of HER-
`2/neu gene amplification or rearrangement
`in any of these groups except for ovarian
`cancer, where gene amplification was found
`in 6 of the 18 tumors analyzed. There are
`several similarities between carcinomas of
`
`the ovary and breast. Like breast cancer, the
`vast majority (>90%) of ovarian malignan-
`cies are of epithelial rather than stromal
`origin (21). Also, like breast cancer, steroid
`hormone receptors are found in the tumor
`cells of many ovarian malignancies (22).
`«Finally, epidemiologic studies suggest that
`some carcinomas of the breast and ovary
`may share common etiologic factors since
`women with breast cancer have twice the
`
`expected incidence of ovarian cancer and
`women with ovarian cancer have a three- to
`
`fourfold increased risk of developing carci-
`noma of the breast (21). These similarities
`and the findings of HER-Z/neu amplifica-
`tion in the initial cases examined prompted
`us to evaluate a larger number of ovarian
`cancers for incidence of gene amplification,
`
`Table 2. Median survival time of patients with ovarian cancer as well as various degrees of HER-2/neu
`gene amplification and expression. All samples were analyzed by both the Kaplan-Meier method and the
`Cox proportional hazards regression model to provide an assessment of the association of the various
`assays with survival. The median follow-up was 75 months for those patients still alive. Subgroup
`categories were as described (Fig. 1).
`
`Ass
`
`ay
`
`Sub-
`group
`
`DNA
`
`RNA
`
`Protein
`
`1 copy
`2 to 5
`>5
`1+
`2+
`3+ to 4+ .
`1+
`2+ to 4+
`
`Num-
`bet
`of
`cases
`
`64
`17
`6
`44
`16
`7
`27
`10
`
`Median
`survival
`.
`time
`(days)
`
`1879
`959
`243
`1960
`919
`531
`1960
`959
`
`Log
`rank
`test
`(P)
`
`.
`COX proportional
`hazards regressron
`model (P)*
`
`<0.0001
`
`<0.0001 [0.1479 1 0.0320]
`
`0.0710
`
`0.0244 [0.4537 x 0.1969]
`
`0.1866
`
`0.0628 [0.5437 1- 0.2944]
`
`1960
`36
`Histochem. Neg. to 1+
`<0.0001 [0.4276 2 0.1163]
`0.0126
`1093
`32
`2+
`417
`4
`>2+
`
`”Re
`ion coefficients : SE are shown in
`uare brackets.
`‘
`
`sq
`
`SCIENCE, VOL. 244.
`
`
`
`PHIGENIX
`
`Exhibit 1002-04
`
`
`
`
`
`
`
`‘(p< 0.0126) association between HER-
`2/neu expression and survival was found
`(Table 2 and Fig. 4). There were a smaller
`number of cases (67 total) for which both
`intact RNA and follow-up were available
`and even fewer (37 total) for which both ,
`Western blot data and follow-up were avail-
`able. Median survival time of these cases
`
`showed the negative relationship between
`survival and increased expression (Table 2)
`but the data did not reach statistical signifi-
`cance for Western analysis (most likely due
`to the small number of samples in the
`group).
`Several conclusions regarding HER-Zlneu
`in human breast and ovarian cancer can be
`
`drawn from these data. For breast cancer,
`we confirmed our earlier conclusions that
`
`amplification of the gene is a predictor of
`both relapse and survival
`in node-positive
`patients and is superior to all other prognos-
`ticators with the exception of positive lymph
`nodes (5). The data, however, do not ad-
`dress the possible influence of therapy in this
`group. To do this would require a study of
`an even larger series of patients given the
`various methods of treatment used for the
`
`disease. In 181 node-negative patients, there
`was no evidence of an association between
`
`gene amplification and outcome. This may
`be due to some fundamental difference in
`
`the basic biology of node-negative disease or
`alternatively may represent a statistical arti-
`fact introduced by the relatively small num~
`ber of patients studied. The overall
`inci—
`dence of disease recurrence in node-negative
`patients is only 15 to 25% and, as a result,
`there may not be a suflicient number of
`events, that is, relapse or death in our co-
`hort, to reveal an association. Recent data
`from another study indicate that over-
`expression of HER-Zlneu is predictive of a
`poor clinical outcome in node-negative
`breast cancer (10).
`Similarities in the biology of the HER-
`2/neu gene in human breast and ovarian
`cancer are considerable and include the fact
`
`that amplification of the gene occurred in 25
`to 30% of both malignancies, amplification
`was uniformly associated with over—expres-
`sion in both nimor types, approximately
`10% of both breast and ovarian cancers had
`
`evidence of a single gene copy with over-
`expression of the gene product, and (in both
`malignancies) there was an association be-
`tween amplification and clinical outcome.
`Few of the 187 breast cancer cases studied in
`
`the‘comprehensive analysis had significant
`follow-up and an even smaller number
`showed amplification/over-expression, malt-
`ing it impossible to perform the necessary
`statistical analyses to evaluate the association
`of gene expression with outcome. However,
`the correlation between amplification and
`
`REPORTS 7" .
`
`PHIGENIX
`
`Exhibit 1002-05
`
`A
`
`T
`
`'
`
`1 copy
`2-5 copies
`>5 copies
`
`l -..
`
`I
`
`I
`._——.A_y.
`-A____A_l_..
`4000
`2000
`____.___ _-_
`
`6000
`
`—— Negative1+ B
`
`L
`
`'.
`‘
`
`-
`
`‘
`
`"
`
`
`
`o
`
`3,1
`.
`":
`---.
`
`0.80
`
`0.60
`
`0.40
`
`2
`e 0.20
`«:v
`_
`g 0
`U}
`0
`->-
`
`'0
`
`g
`E3
`
`O
`
`“
`
`correlation between amplification and over-
`expression, and an association between am-
`plification/over-expression and clinical be“
`havior of the malignancy.
`DNA from 120 primary ovarian malig—
`nancies of epithelial origin were evaluated
`.. by Southern blot analysis using the same
`criteria and techniques described for the
`breast cancer cohort. Identical cont