Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 1 of 8 PageID #:
`9917
`
`
`
`
`
`
`
`
`
`
` RHYU
`EXHIBIT 12
`
`
`

`

`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 2 of 8 PageID #:
`9918
`[CANCER RESEARCH54, 2771-2777, May 15. 1994]
`Sensitivity of HER-2/neu Antibodies in Archival Tissue Samples: Potential Source of
`Error in Immunohistochemical Studies of Oncogene Expression1
`
`Michael F. p@88,2 Gene Hung, William Godolphin, and Dennis J. Slamon
`
`Department of Pathology, University of Southern California School of Medicine, Los Angeles, California 9fX@33[M. F. P., G. H.J; Division of Clinical Chemistry, Vancouver
`General Hospital@ Vancouver, British Columbia, Canada [W. G.J; and Division ofHematology-Oncology,
`UCLA.
`School ofMedicine,
`Los Angeles, California 90024 fD. J. S.]
`
`ABSTRACT
`
`HER-2/neu oncogene amplification and overexpression ofbreast cancer
`tissue has been correlated with poor prognosis in women with both
`node-positive and node-negative disease. However, several studies have
`not confirmed this association. Review of these studies reveals the pres
`eace of considerable methodological variability including differences In
`study size, follow-up time,
`techniques and reagents. The inaJonty of
`papers with clinical
`follow-up information
`are immunohistochemical
`stud
`ies using archival, paraffin-embedded breast cancers, and a variety of
`HER-2/neu
`antibodies
`have
`been used
`In these
`studies
`Very
`little
`infor
`mation, however, is available about the ability of the antibodies to detect
`overexpressioa
`following tissue processing
`for paraffin-embedding.
`Therefore, a series of antibodies, reported in the literature or commer
`daily available, were evaluated to assess their sensitivity and specificity as
`immunohistochemical
`reagents. Paraffin-embedded samples of 187 breast
`cancers,
`previously
`characterized
`as
`frozen
`specimens
`for HER-2lneu
`amplification
`by Southern
`blot and for overexpression
`by Northern
`blot,
`Western
`blot,
`and immunohistochemlstry,
`were used. Two multitumor
`paraffin-embedded tissue blocks were prepared from the previously an
`alyzed breast cancers as a panel of cases to test a series of previously
`studied and/or commercially available anti-HER-2/neu antibodies. Immu
`nohistochemical staining results obtained with 7 polyclonal and 21 mono
`clonal antibodies
`in sections from paraffin-embedded
`blocks of these
`breast
`cancers were compared.
`The ability of these antibodies
`to detect
`overexpression was extremely variable, providing an important explana
`tioa for the variable overexpression rate reported In the literature.
`
`INTRODUCTION
`
`encodes a membrane receptor with
`The HER-2/neu protooncogene
`homology to the epidermal growth factor receptor (1—3). Amplifica
`tion and overexpression of HER-2/neu in human breast cancer has
`been correlated with a shorter disease-free interval and shorter overall
`survival
`in some studies
`(4—16) but not
`in others
`(17—24). Some
`studies
`find a correlation with clinical outcome in only select sub
`groups of women with breast cancer
`(25—27). A major question has
`been: Why do these studies come to different conclusions? Differ
`ences in the conduct of these investigations may be responsible
`for
`these conflicting data. Among published reports, there is considerable
`variability including differences in study size, techniques, reagents,
`and follow-up time. Problems with study design in any of these areas
`has the potential
`to lead to erroneous conclusions with regard to
`HER-2/neu
`prognostic utility. The importance of sample size has
`already been demonstrated and discussed (9, 16) as has the importance
`of the sensitivity of various analytical techniques (5). The majority of
`
`Received 6/2/93; accepted 3/17/94.
`The costs of publication of this article were defrayed in part by the payment of page
`charges. This article must therefore be hereby marked advertisement in accordance with
`18 U.S.C.Section1734solelyto indicatethisfact.
`from the
`and CA50589
`CA36827,
`1 Supported
`in part by Grants
`CA48780,
`CA32737,
`from the National Institute of Child
`National Cancer Institute; Grant N01-HD-3-3175
`Health and Human Development; Grant 114-21-291003and PDT-411 from the American
`CancerSociety;the Revlon/UCLAWomen'sCancerResearchProgram;andthe Molec
`ular Core LaboratoryFacilitiesof the GeneralClinicalResearchCenters(National
`institutesof HealthNationalCenterfor ResearchResourcesof the GeneralClinical
`Research Centers MOl RR-43). D. J. S. has a Faculty Research Award from the American
`CancerSociety.
`2 To whom requestsfor
`reprints shouldbe
`addressed,
`at DepartinentofPathology—HMR91O,
`USC SchoolofMedicIne,2011ZonalAvenue@LosAngeles,CA90033.
`
`2771
`
`the published studies in the literature have been performed on paraffin
`blocks and use immunohistochemistry
`as the only analytic technique
`for assessing HER-2/neu alteration.
`In the current study, we investi
`gate the importance of an essential
`study reagent, HER-2/neu anti
`bodies,
`in the context of immunohistochemical
`analysis of paraffin
`embedded tissue samples.
`Review of the published studies shows a wide range of overexpres
`sion rates, varying from 9% at
`the lowest
`to 38% at
`the highest
`(21,28).However,awidevarietyofdifferentantibodieshavebeen
`used in these studies. Sensitivity and specificity of only a few of these
`antibodies has been characterized in this type of material
`(16, 17). In
`this
`study,
`a group of 187 archival
`breast
`cancers with known
`HER-2/neu amplication and expression levels were used as a panel to
`characterize
`the sensitivity of 28 separate anti-HER-2/neu
`antibodies
`which have been used or are currently available as immunohistochem
`of
`ical
`reagents. This approach was facilitated by the preparation
`multitumor
`tissue blocks, which proved to be an effective quality
`control strategy (29, 30).
`
`MATERIALS
`
`AND METHODS
`
`Breast Cancers. All breast cancer specimens used in this study had known
`HER-2/neu amplification and expression levels. Amplification was determined
`by Southern blot using DNA extracted from frozen tumor specimens
`(5, 31),
`while expression was determined by Northern analyses, Western immuno
`blotting,
`and immunohistochemistry
`using total RNA,
`total protein,
`and
`histological
`sections,
`respectively,
`from the
`same
`frozen
`tumor
`tissue
`(5, 31). Amplification and overexpression were found in 27% of the breast
`cancers
`used in this study (5, 31). Overexpression
`without
`amplification
`was found in an additional 10% of cases, giving a total overexpression rate
`of 37% in this cohort. Paraffin-embedded tissue blocks from these same
`cases were
`obtained
`from institutional
`archives.
`In order
`to test
`each
`antibody
`in a consistent
`and reproducible
`fashion,
`two multitumor
`tissue
`blocks were prepared containing small strips of each of the 187 breast
`cancer
`specimens. The blocks were prepared
`according
`to the methods
`of
`Battifora and Mehta (30) as summarized below.
`used in this
`cancers
`Preparation
`of Multitumor
`Blocks. The breast
`study had been originally
`fixed in a routine
`fashion with buffered
`formalin
`for varying periods of time in the surgical pathology laboratory. The
`specimens
`had been dehydrated
`through
`graded alcohols
`and xylene
`and
`placed in warm (60°C)paraplast in an automatic tissue processor (Tech
`nicon Co.). The
`tissue was
`then placed
`in a mold
`and cooled
`slowly
`to
`embed the tissue in solidified wax.
`The multitumor blocks were prepared by slowly warming archival tumor
`specimens from each breast cancer to 60°Cuntil the embedding medium
`(paraplast) melted. A 2.0-mm wide segment of tissue was removed from one
`edge of each breast carcinoma and mounted in a rectangular, clear, plastic box.
`The breast
`cancer
`strips were arranged in rows of 10 to 12. Each row was
`separated from the next row by a layer of paraflim. After 10 rows had been
`arranged in this way,
`the specimens were reembedded
`in paraffin to form a
`multitumor block (Fig. 1). Two such blocks were prepared. The coordinates of
`each breast cancer were recorded and stored in a computerized database for
`later comparison with amplification and expression data.
`Tissue sections from these multitumor blocks were immunostained in a
`blinded fashion as described below. This approach permitted the use of very
`small quantities ofantibody
`and, more importantly, provided equal exposure of
`each breast cancer to the antibody being tested.
`
`

`

`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 3 of 8 PageID #:
`9919
`HER-2/aeuIN PARAFFIN-EMBEDDEDTISSUE
`
`@,7.
`
`B
`Fig@1. MultitumortissueblocLA, grossphotographshowingarrangementof breastcancersin a multitumortissueblock.B, histologicalsectionfromthe multitumorblock.
`Hematoxylin and rosin stain. X 4.
`
`HER-2/neu Antibodies. Seven polyclonal rabbit antibodies and 21 mouse
`monoclonal antibodies were tested for HER-2/neu immunostaining using mul
`titumor tissue sections (Table 1). Each of these antibodies has been reported to
`specifically recognize the HER-2/neu receptor protein either in the published
`literature or in commercial
`descriptive
`information. Polyclonal
`rabbit antisera
`were all directed against synthetic peptides based on the HER-2/neu receptor
`sequence (R60, 21N, D485, pAbi, PAB, R347, and A8010; Table 1).
`Immunohistechemical Technique. For uniformity, all antibodies were
`evaluated using the peroxidase/antiperoxidase technique. Optimal dilutions of
`the antibodies with excess antibody compared to antigen were determined by
`serial antibody dilutions with immunohistochemical testing and/or use of
`optimal conditions
`as described in the literature (Fable 1). In brief,
`the method
`involved the sequentialapplicationof threeantibodiesto tissue sections (16):
`(a) primary HER-2/neu antibody; (b) a secondary or bridging antibody,
`recognizing the immunoglobulin type of both the primary antibody and the
`tertiary pesoxidase
`anti-peroxidase
`antibody; and (c) a peroxidase/antiperoxi
`dase antibody complex of the same species and immunoglobulin subclass as
`
`Table1 HER-2/neuantibodies
`
`or
`AntibodyDilutionconcentrationSourceLiterature
`
`communication9C210
`@&g/mlAmgen,
`Inc.RK@R601:2000UCLA31A80105
`
`citation or
`
`personal
`
`the primary antibody. When the primary antibody was a rabbit polyclonal
`antibody, a goat anti-rabbit
`IgO antiserum (1:50 dilution; Sternberger-Mayer,
`Inc.) was used as the secondary antibody, and a rabbit pesoxidase/antiperoxi
`dma antiserum (1:50 dilution; Sternberger-Mayer, Lnc.)was used as the tertiary
`antibody. When the primary antibody was a mouse monoclonal
`antibody,
`a
`rabbit anti-mouse immunoglobulin
`antiserum (1:50 dilution; Zymed,
`Inc.) was
`used as the secondary antibody, and a mouse monoclonal peroxidase/antiper
`oxidase
`antibody (1:50 dilution; Sternberger-Mayer,
`Inc.) was used as the
`tertiary antibody. One of the antibodies, TAB25O, was reported to provide
`better immunostaining results when the tissue sections were pretreated with
`0.1%protease(Nagarse;SigmaChemicalCo., Inc., St. Louis, MO)for 20 rain
`at room temperature (10). TAB25Owas used with and without this protease
`pretreatment. Another antibody, A8010, was available both as fresh antiserum
`and lyophilized antiserum. These preparations were tested separately.
`Each of the primary HER-2/neu antibodies was incubated overnight at 4°C,
`and secondary and tertiary antibodies were incubated at room temperature
`for
`one-halfof an h. After treatmentwith each antibody,the tissue sections were
`washed with phosphate-buffered
`saline. The immunoprecipitates were identi
`fled microscopically
`after incubation with the chromogen
`diaminobenzidine.
`Membrane staining was interpreted as HER-2/neu oncoprotein expression,
`and
`the amount of immunostaining was scored in a blinded fashion as low/
`negative, intermediate, or high as described and illustrated elsewhere (16). The
`sensitivity and specificity ofthe immunostaining
`procedure was determined as:
`Sensitivity = True PositivesiTrue Positives + False Negatives; Specificity =
`True NegativeslFrue Negatives + False Positives.
`
`RESULTS
`
`Inc.W?'11G510
`@g/mlOncor,
`pg/edAmgen,
`Inc.RI(°DA4851:100Dako,
`Corp.TAB25O'@10
`g@gImlBerleLBiosciences,
`Inc.10pAbi1:15Bcrlex
`Biosciences,
`Inc.403H410
`pg/mlGenentech,
`Inc.384D510
`pg/mIGenentech,
`Inc.3836B710
`@&g/edBerlex Biosciences,
`Inc.135C610
`g@g/m1Univ. of
`Chicago437F310
`@g/edGenentech,
`Inc.38CB111:15Signet
`Inc.393B51:5000Oncogene
`Labs,
`Science,
`Inc.419G65
`@Lg/mlOncogene Science,
`Inc.4121N
`Hospital
`
`R3471:200WJI'H6Q3810
`
`5 g.@g/mlHammersmithOncor, Inc.42
`
`pg/mIUniv.Chicago436E910
`
`of
`pg/mIGenentech,
`Inc.38145WW10
`@.&gfmlUniv.of
`Chicago432H1110
`@g/mlGenentech,
`Inc.383E810
`j.tglmlGenentech,
`Inc.38TA!100
`g@gfmlOncogenctics
`Inc.255B810
`pg/edGenentech,
`Inc.387C210
`@&g/mIGenentech,
`Inc.382C410
`pg/mIGenentech,
`Inc.387D310
`@g/mlGenentech,
`Inc.38PAB'1:50Oncogene
`
`Partners,
`
`Science, Inc.
`
`Inc.
`Amgen,
`Ph.D.,
`a R. @c, Ray Koski,
`b w.
`j., wiuis@
`James,
`Ph.D.,
`Oncor,
`
`Inc.
`
`C TAB251
`as well
`as other monoclonal
`antibodies
`(TAB26O
`series)
`not
`provided
`because“theimmunoreactivitiesof theseantibodiesare similarto or lessthanTAB25O,
`which was providedâ€(cid:157) (personal communication; Berlex Biosciences, Inc.).
`
`corre
`cell membranes
`of breast carcinoma
`Only iminunostaining
`lated with elevated levels of HER-2/neu expression (overexpression),
`while staining of tumor cell cytoplasm or nuclei and staining of
`noncarcinoma cells did not correlate with HER-2/neu overexpression
`and was, therefore,
`regarded as nonspecific. Membrane immunostain
`ing was demonstrated with the various antibodies in 2 to 30% of the
`breast cancers in this cohort, depending on the antibody used (Fig. 2;
`Table 2). Each of the antibodies
`tested detected at least some of the
`breast cancers with known HER-2/neu overexpression,
`but none de
`tected all of the formalin-fixed,
`paraffin-embedded
`specimens with
`known HER-2/neu
`overexpression.
`Three
`of the six antibodies
`which
`identified 70% or more of the breast cancers with overexpression were
`polyclonal
`rabbit antibodies
`(Fig. 3; Table 2). All six of these anti
`bodies
`showed distinct
`and strong membrane
`immunostaining
`of
`tumor cells with minimal nonspecific background staining (Fig. 2, A,
`B, and C). Immunohistochemical staining with most of the other,
`predominantly monoclonal, antibodies similarly demonstrated distinct
`membrane immunostaining
`of tumor cells with little background but
`detected fewer of the breast cancers known to have overexpression
`2772
`
`

`

`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 4 of 8 PageID #:
`9920
`HER-2lneu IN PARAFFIN-EMBEDDEDTISSUE
`
`>5-Fold
`
`2 to 5-Fold
`
`-
`
`l@
`
`,.@
`__
`,
`-@ -
`
`@- _
`
`-
`
`__t*\_
`
`/
`
`.“@ - ‘a
`
`-@
`
`@—.-
`
`@L@:
`
`:
`
`C
`
`@..
`
`4D5
`
`Low
`R60
`
`4D5
`
`TA-I
`
`E
`
`H
`
`TA—I
`
`F
`‘@;@ I@'
`
`‘PA—I
`
`‘I
`
`‘
`
`a
`
`,-
`
`a@
`
`@,
`
`:
`
`@@:=-‘@•
`i-@:@@@--@
`
`@
`
`@
`
`@
`
`@
`
`
`
`@@
`
`
`
`@@@
`
`@
`
`G
`
`Fig. 2. Immunostaining patterns observed with various HER-2lneu antibodies in breast cancers of the multitumor tissue block. A-I, comparison of membrane immunostaining
`obtained with three different HER-2fneu antibodies (1(60,4D5, and TA-i) in the same three breast cancers of a multitumor tissue block. The three breaMcancers were known to have
`greater than 5-fold amplification with overexpression, 2- to 5-fold amplification with overexpression, and no amplification with low expression ofHER-2/neu. The antibodies illustrate
`therangeof immunostainingobservedineath ainplification/overexpressioncategnry.A-C@membraneimmunostainingis observedwithR60inbreastcancerswithoverexpressionand
`>5-fold(A)and2- to 5-fold(B) amplificationofthe gene,butnostainingis observedinthebreastcancerlackinggeneamplificationandlackingoverexpresaion(C).1)-F, membrane
`immunostaining is observed with 4D5 antibody in a breast cancer with overexpression and >5-fold amplification (D) but not in a breast cancer with overexpression and 2- to 5-fold
`amplification (E) and, as expected, not in a breast cancer with neither amplification nor overexpression (F). G-4 lack of immunostaining with TA-i antibody in breast cancers with
`overexpression and >5-fold (G) or 2- to 5-fold (H) amplification of the gene. Coarse granular cytoplasmic staining is observed with TA-i in a breast cancer lacking overexpression
`(I). X 14-40.
`
`(Table 2; Fig. 2, D, E, and F; Fig. 3). In fact, eight antibodies detected
`less than 30% of breast cancers with known overexpression. Seven of
`these eight were mouse monoclonal antibodies (Table 2; Fig. 2, G, H,
`and I). The eighth, a polyclonal rabbit antibody (PAB°),detected the
`lowest percentage of positive cases (Table 2).
`Protease pretreatment of the tissue section substantially improved
`sensitivity of the TAB25O antibody from 36% with no pretreatment to
`70%with proteasepretreatment(Table2). Lyophilization diminished
`the sensitivity of A8010 polyclonal
`antibody after reconstitution
`in
`distilled
`water
`as compared
`to the
`nonlyophilized
`preparation
`of
`the
`material
`(Table 2).
`With 20 of the 28 antibodies evaluated, membrane staining was
`restricted to breast cancers known to have HER-2/neu overexpression
`from previous Northern and Western blot analyses (5, 31). Eight of
`the antibodies, however, showed membrane immunostaining, not only
`in breast cancers with overexpression, but also in breast cancers with
`low expression of HER-2lneu. The antibodies with false positive
`immunostaining
`were
`as
`follows
`(specificity):
`1105
`(92%),
`36B7
`(96%), TAB25O(97%), TA-i
`(98%), 6E9 (99%), 3H4 (99%), D485
`(99%),andPAB(99%).
`Twenty-three of the 28 antibodies identified a higher percentage of
`breast cancers which have >5-fold amplification
`of the gene and
`overexpression than those which have either 2- to 5-fold amplification
`
`or overexpressionwithout gene amplification (Fig. 3). Only 3E8,
`TA-i, 5B8, 201, and PAB immunostained a higher percentage of
`breast cancers with moderate levels (2- to 5-fold gene amplification)
`thanwith high levelsof overexpression(>5-fold geneamplification).
`Varying
`amounts
`of nonspecific
`staining,
`not associated with
`HER-2/neu overexpression, were observed with selected antibodies in
`a variety of staining patterns in the breast carcinoma cells.
`i35C6 had
`coarse
`granular
`cytoplasmic
`staining
`in 9i% of all breast
`cancers
`in
`the cohort. The 3B5 monoclonal antibody produced varying degrees
`of course granular cytoplasmic immunostainingin 62% of breast
`cancers with overexpression
`and in 68% of breast
`cancers with low
`expression. Delicate,
`irregular
`intracytoplasmic,
`linear staining was
`observed
`with
`one
`antibody,
`3E8,
`in both
`high
`and low expression
`breast cancers in a pattern consistent with staining of internal mem
`branes
`(i9% of cases). TA-i
`had both coarse
`and fine granular
`cytoplasmic
`staining in 14% of cases (Fig. 2!). Uniform,
`fine cyto
`plasmic staining was obtained with another antibody, 2Hii,
`in 7% of
`cases, and the majority of these were breast cancers without known
`overexpression. Nuclear
`immunostaining was observed in a low per
`centage of breast cancers with five different antibodies
`(A80i0, 19%;
`i35C6,
`8%;
`2C4,
`3%;
`3E8,
`3%;
`and
`7F3,
`2%). With
`TAB
`250
`and
`36B7,
`luminal membranes of endotheial
`cells were also immuno
`2773
`
`

`

`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 5 of 8 PageID #:
`9921
`FIER-2/neuIN PARAFFIN-EMBEDDEDTISSUE
`
`Table 2 Comparison of inununostaining obtained with various HER-2/neu antibodies
`
`Antibody
`“Idealâ€(cid:157)resulta
`9C2
`R60â€(cid:157)
`A8010â€(cid:157)
`llG5c
`A485@'
`TAB25O pret.â€(cid:157)
`p@u,1#@e
`3H4
`4D5
`36B7
`135C6
`7F3
`CR11
`3B5
`906
`2iNb
`R347
`H6G38
`TAB25Oâ€(cid:157)
`6E9
`A8010 (lyophil.)L@f
`145WW
`2Hll
`3E8
`TA-l
`5B8
`7C2
`2C4
`7D3
`p@@ob.g
`
`Breast cancer
`Immunostained (%)
`37
`30
`29
`28
`38
`26
`26
`24
`22
`21
`24
`20
`19
`19
`20
`18
`17
`17
`16
`16
`12
`I 1
`Il
`10
`9
`9
`7
`5
`3
`3
`2
`
`Sensitivity (%)
`100
`82
`80
`75
`80
`70
`70
`65
`59
`57
`54
`53
`51
`51
`50
`47
`47
`47
`42
`36
`31
`30
`30
`28
`24
`22
`18
`14
`8
`7
`6
`
`Specificity (%)
`100
`100
`100
`100
`92
`99
`100
`100
`99
`100
`96
`100
`100
`100
`98
`100
`100
`100
`100
`97
`99
`100
`100
`100
`100
`98
`100
`100
`100
`100
`99
`
`@
`
`is based on analysis of the breast cancer cohort as frozen specimens
`‘Ideal' resultâ€(cid:157)
`a
`by Southern hybridization, Northern hybridization, Western immunoblot, and immuno
`histochemistry.Overexpressionwasobservedin 18,9, and 10%of breastcancerswith,
`respectively, >5-fold amplification, 2- to 5-fold amplification, and no amplification of
`HER-2/neu gene, yielding a total overexpression rate of 37% (5).
`b Indicates
`polyclonal
`antibodies.
`All
`others
`are monoclonal
`antibodies.
`
`by scoring moderate
`
`(+ +)
`
`C Optimal
`specificity
`and sensitivity
`for 1 1G5 was obtained
`or @reaterimmunostaining as overexpression.
`TAB25O pret. and TAB25O summarize results obtained with TAB25O antibody,
`respectively, with and without protease pretreatment. TAB251 as well as other monoclo
`nal antibodies (TAB26O series) not provided “becausethe immunoreactivities of these
`antibodies are similar to or less than those that were providedâ€(cid:157) (personal communication;
`Berlex Biosciences, Inc.).
`C Berlex
`Biosciences,
`Inc.
`1A8010 and A8010 (lyophil.) summarize findings using, respectively, fresh and
`lyophilized antiserum.
`g Oncogene Science,
`
`Inc.
`
`surfaces of benign breast ducts (data not
`
`stained, as were luminal
`illustrated).
`
`DISCUSSION
`
`In a previous study using frozen tissue from this same cohort of
`breast cancers,
`immunohistochemical
`staining was compared with
`Northern
`blot and Western
`immunoblot
`analyses
`as measures
`of
`gene expression. A high level of concordance (91%) was observed
`among these three assays (5). Immunohistochemistry,
`using one of
`the polyclonal
`antibodies
`assessed in this study (R60), provided the
`highest
`level of concordance with the other assays of gene expres
`sion (99%). The level of concordance
`obtained with Northern blot
`analysis
`was
`similar
`to that obtained
`with
`immunohistochemistry
`(98%). Although Western immunoblot usually agreed with the
`other
`two assays of gene expression
`(94%),
`it provided discordant
`results more often than either Northern blot analysis or immuno
`histochemistry.
`In addition, when immunostaining
`in frozen tissue
`sections was compared with immunostaining
`in formalin-fixed,
`paraffin-embedded
`tissue sections
`from the same breast cancers,
`the number of positive cases was reduced for both the R60 poly
`clonal antibody (5, 16) and the 3B5 monoclonal
`antibody (16).
`Because of this, demonstrable change in sensitivity which occurs
`
`2774
`
`antibodies when
`to a greater or lesser degree with anti-HER-2/neu
`using formalin-fixed,
`paraffin-embedded
`tissue and because of the
`marked variability
`in frequency
`of overexpression
`as determined
`by immunostaining reported in the literature, we decided to attempt
`to address
`this potentially
`important
`issue in a systematic
`fashion.
`In this study, a cohort of breast cancers with known levels of
`HER-2/neu gene amplification and expression was used as a panel to
`demonstrate
`variable detection rates of immunostaining
`in archival
`tissue for a large series of HER-2/neu-specific
`antibodies, many of
`which have been reported in published studies of HER-2/neu overex
`pression in breast cancer. The results demonstrate that, while all of the
`antibodies did show some degree of membrane-associated
`immuno
`staining in the molecularly characterized study cohort, each antibody
`failed to detect some of the cases known to have overexpression. This
`was expected since it was known from direct comparison of iminu
`nostained frozen tissue sections with immunostained
`paraffin-embed
`ded tissue sections from the same breast cancer cases that there is a
`reduction in immunostaining with tissue fixation and paraffin-embed
`ding on a case-by-case basis (5). More importantly, the ability of
`different
`antibodies
`to detect HER-2/neu
`protein overexpressed
`in
`archival
`tissue sections is quite variable.
`This variability in sensitivity in archival tissues is very likely to
`be one (if not
`the major)
`factor
`contributing
`to the differing
`immunostaining
`frequencies
`reported
`in the literature.
`If one as
`sumes
`that
`the HER-2/neu
`detection
`rates observed
`here are rep
`resentative
`of percentages
`found in tissue processed
`for formalin
`fixation and paraffin embedding of breast tissues in general,
`then
`we can estimate the “trueâ€(cid:157)overexpression rate of reported cohorts
`(Table 3). Reported HER-2/neu
`immunostaining
`percentages
`are
`summarized in Table 3 for various antibodies. Although the re
`ported immunostaining
`percentages,
`ranging from 14% to 25%, are
`considerably lower than the overexpression rate observed in frozen
`specimens,
`“correctionâ€(cid:157) for
`false-negative
`cases
`in paraffin-em
`bedded tissue using the values observed here provide a range of 26
`to 39% for seven of the eight antibodies
`(Table 3). These “cor
`rectedâ€(cid:157) expression rates more closely approximate
`the rate reported
`in cohorts
`analyzed as frozen specimens
`(5).
`TA-i was the only antibody whose reported immunostaining per
`centage did not “correctâ€(cid:157)to approximately
`the 30% range for over
`expression using the sensitivity rate observed for paraffin-embedded
`samples in this study. The reason for this discrepancy is unclear. The
`commercial distributor of TA-i does not recommend this antibody for
`analysis of paraffin-embedded
`tissues because it does not successfully
`stain
`this
`type of
`specimen.3
`Despite
`this, TA-i was
`tested
`in this
`study because it has been used in a previous
`immunohistochemical
`study of paraffin-embedded
`breast cancers (25). Our findings with this
`antibody support
`the recommendations
`of the distributor
`that it is not
`very sensitive for identifying paraffin-embedded
`breast cancers with
`HER-2/neu overexpression. Since three of the twenty tumors showing
`membrane immunostaining with this antibody were breast cancers
`known to have low expression,
`this antibody has the potential
`for
`significant
`false-positive misclassification
`of cases.
`The proportion of false-positive breast cancers included as having
`membrane staining was relatively small for nearly all of the antibodies
`and probably has had only a very limited impact on errors made in
`classification
`of breast cancers by immunostaining. However,
`the
`nonspecific
`cytoplasmic
`staining observed with some of the anti
`bodies could be an important compounding factor if this staining was
`assessed as “positiveâ€(cid:157) staining in some studies.
`In this study, we
`
`3 Personal
`
`communication,
`
`Oncogenetics
`
`Partners,
`
`Inc.
`
`

`

`@
`@@@@
`
`@@@
`
`@
`
`@
`
`@@
`@
`@
`
`@
`
`@r
`@:
`@;
`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 6 of 8 PageID #:
`9922
`HER-2/neuIN PARAFFIN-EMBEDDEDTISSUE
`
`l@
`
`-
`
`100
`90@
`80
`7Ø@
`60
`50
`40
`30•
`20
`10@
`0@
`
`c'@
`
`ii-t
`
`lrt
`
`@r
`
`I L@
`
`1$
`
`‘@1-@ft -
`
`>5-fold
`
`2 to 5-fold
`
`iHigh
`
`a
`
`LikLLhn: i@
`
`@o 0
`
`O\
`
`@.
`
`—
`
`00
`
`‘—‘ 00
`
`C-@I
`
`c@@t
`
`F@
`
`@d,@@l
`
`p@ 2'
`
`‘
`
`p@
`
`N
`
`e1.@
`ZN
`
`@,o ‘t@
`
`E
`Fig. 3. Bar graph comparing the percentage of breast cancers immunostained in each of 4 HER-2lneu expression categories by 28 different antibodies. One of the antibodies,
`TAB25O, was used with (FAB25Op) and without (TAB25O)protease pretreatment of tissue sections@Another antibody, A80i0, was used either as fresh antiserum (A8010) or as
`lyophilized antiserum that was reconstituted (A80i0[l]). The four expression categories were based on HER-2/neu gene amplification and expression results obtained by analysis of
`DNAwithSouthernhybridization,RNAby Northernhybidization,totalproteinbyWesternimmunoblot,andimmunohistochemistryusingfrozentissuesfromthesamebreastcancers
`analyzed in this report as paraffin-embedded tissues. The figure legend identifies three groups of breast cancers with overexpression, separtely categorized by gene amplification levels
`(>5-fold,2- to 5-fold,andiH), anda groupof breastcancerswithneitheramplificationnoroverexpressionof HER-2lneu(1 Low).Immunostainedcasesarepresentedas a percentage
`of the breast cancers identified in each breast cancer expression group (ordinate).
`
`-f-'nNc@N
`
`c@tcn
`
`once again confirm that only membrane staining is associated with
`HER-2/neu overexpression.
`in
`for immunohistochemistry
`The lack of antibody characterization
`the literature has undoubtedly contributed to confusion regarding
`potential prognostic
`utility of HER-2/neu
`overexpression
`in breast
`cancers. Previously,
`only one other group (17) besides ourselves
`(5, 16) has characterized
`antibodies
`for immunostaining
`in paul
`finized breast cancers using breast cancers with known HER-2/neu
`gene amplification from Southern blots (32). However, the previously
`reported complete sensitivity of 3B5 and 9G6 for breast cancers with
`gene amplification (17) is at odds with our own findings reported here.
`In the previous study (17), thirty-eight breast cancer specimens were
`used as calibration standards, 10 with and 28 without amplification.
`Although the 10 breast cancers with amplification were all reported to
`show 2- to 5-fold amplification,
`they actually appear to have much
`higher
`levels of gene amplification (see Fig. IA in Ref. 32). Each of
`the breast cancers with amplification
`had a HER-2/neu
`signal on
`Southern blot that was as great as or greater than the amplification
`signal of the positive control cell
`line, SK-BR-3, which is known to
`show 6- to 8-fold amplification (33). Therefore,
`it appears that the
`overexpressing cases which were used to characterize the 3B5 and
`
`Table 3 Projected (or “correctedâ€(cid:157))breast cancer overexpression rates forvariousstudy
`
`
`cohorts based on reported percentagessensitivitiesReported
`and observed antibody
`
`Observed“Correctedâ€(cid:157)overexpression
`
`overexpressionAntibody
`antibody
`sensitivities (%)
`ReferencesR60
`Rate (%)
`15,16TAB25O(pret)
`25
`79
`23
`70
`10pAbl#
`16
`65
`40CB11
`20
`51
`393B5
`14
`50
`41906
`14
`47
`17,412iN
`18
`47
`22TA-i
`15
`22
`
`rate (%)
`33
`33
`26
`39
`28
`30
`38
`68
`
`17,
`
`7,9,12—14,21,
`25
`
`2775
`
`tissue were limited to 10
`in paraffin-embedded
`906 immunostaining
`highly amplified, overexpression
`breast cancers (32).
`Quality control
`is as important
`an issue for immunohistochem
`istry as it is for other procedures
`of potential
`diagnostic
`clinical
`utility. However,
`careful
`characterization
`and quality
`control
`of
`reagents by either the distributors
`of anti-HER-2/neu
`antibodies
`for
`diagnostic
`purposes or the investigators
`using these antibodies
`in
`immunohistochemical
`studies of paraffinized
`tissue has been very
`limited (34, 35). The Biological
`Stain Commission
`convened
`a
`workshop
`in 1988 to formulate
`recommendations
`for quality con
`trol and standardization
`of immunoreagents
`in immunohistochem
`istry (35). One of the recommendations,
`that each antibody should
`pass a “performance testâ€(cid:157) for example with a multitissue
`“sausage
`type blockâ€(cid:157), has,
`to our knowledge,
`not been performed previously
`with any of
`the HER-2/neu
`antibodies
`used for
`immunohisto
`chemistry
`in paraffin-embedded
`tissues.
`Differences
`in antibody sensitivity, while clearly existing, are not
`the sole reason for variability
`in immunohistochemical
`results ob
`tamed with paraffin-embedded
`tissues. Published studies support our
`impression that tissue processing also accounts for at least some of the
`observed variability in the literature. Gusterson reports that methacarn
`fixation provides better immunostaining for HER-2/neu than does
`formol saline fixation (36). In addition,
`the same antibody, 21N, has
`been used in several
`reports from different
`institutions
`for immuno
`staining paraffin sections. The percentage of breast cancers scored as
`positive in these reports varies from 9 to 24% (7—9, 12—14, 21, 22).
`Since the breast cancers originate from different
`institutions,
`tissue
`processing,
`especially
`tissue fixation, may be different
`and could
`account
`for some of the variability in the percentage of positively
`immunostained
`cases.
`The sensitivity and specificity of antibodies used for immunohis
`tochemical assays can also have an impact on the sample size needed
`to demonstrate a statistically significant difference in prognostic sub
`
`

`

`Case 1:18-cv-01363-CFC Document 82-12 Filed 03/22/19 Page 7 of 8 PageID #:
`9923
`
`HER-2/neu IN PARAFFIN-EMBEDDED TISSUE
`
`diminishing sensi
`groups. For example, antibodies of progressively
`tivity would be expected to have an increasing fraction of misclassi
`fication of the study population and therefore require a progressively
`larger cohort
`to show significant
`differences
`between HER-2/neu
`immunostaining
`groups.
`of various antibodies has been
`Finally, although the sensitivities
`characterized under the described routine conditions permitting more
`rational
`selection of immunostaining
`reagents,
`it is anticipated that
`there will be further
`improvements
`in HER-2/neu sensitivity in par
`affin-embedded
`tissues as more antibodies become available for test
`ing and as techniques for immunostaining improve. For example, the
`sensitivity
`of several of these antibodies
`is enhanced by “antigen
`retrievalâ€(cid:157) with microwave
`pretreatme