`
`
`
`
`
`
`PHIGENIX
`PHIGENIX
`Exhibit 1023
`Exhibit 1023
`
`

`

`Acad. Sci. USA
`Proc. N
`Vol. 93,’ p. 8618—8623, August 1996
`Medical Sciences
`
`Eradication of large onion tumor xenografts by targeted
`delivery of maytansinoids
`(immunoconjugate/colon cancer xenografts)
`CHANGNtAN Ltu", B. MlTRA TADAYONl, LIZABETH A. BOURRET, KRISTIN M. MATTOCKS, SUSAN M. DERR,
`WAYNE C. WlDDISON, NANCY L. KEDERSHA, PAMELA D. ARlNlELLO, VICTOR S. GOLDMACHER, JOHN M. LAMBERT,
`WALTER A. BLATI'LER, AND RAVI V. J. CHARt
`
`.
`
`lmmunchn, Inc., 148 Sidney Street, Cambridge, MA 02139—4739
`Communicated by Stuart Schlossman, Dana—Father Cancer Institute, Boston, MA, May 22, 1996 (received for review March 29, 1996)
`attempts using monoclonal antibodies and the anticance;
`drugs doxorubicin (6), methotrexatc (7), and Vinca alkaloid:
`(8), have been largely unsuccessful. These antibody-drug con
`jugatcs were only moderately potent and 'usually less cytotoxii
`than the corresponding unconjugated drugs. In fact, antigen
`specific cytotoxicity toward cultured tumor cells was rarel'
`demonstrated (6-8). In vivo therapeutic effects with tit-:5
`conjugates in tumor xenograft animal models were, in genera
`observed only when the treatments were commenced befor
`the tumors werc‘well established (8) or when exceedingly larg
`doses (up to 90 mg/kg, drug equivalent dose) were used (6
`1t is. therefore. not surprising that in human clinical trials, 11
`significant antitumor effects were observed with these agen
`(9, 10). Indeed, the peak circulating scrum conccntrationsr
`conjugates were only in the same range as their in vitro 1C
`values and, thus, capable of eliminating at best only about 5'1"
`of tumor Eclls.
`These observations have led us (11. 12) and others (1?.
`to conclude that the previous attempts at delivering therapcz
`tic doses of cytotoxic drugs via monoclonal antibodies ha‘
`met with little success in clinical trials because of inappropria
`choices of drug. We concluded that immunoconjugatcs mu
`be composed of drugs possessing much higher potency than ti
`clinically used anticancer agents if therapeutic levels of co
`jugate at the tumor sites are to be achieved in patients. We ha
`recently described antibody conjugates with CC-1(165 analt~
`and with maytansinoids. that are 100-
`to lUtltl-fold 1..
`cytotoxic, than the chemotherapeutic agents doxortmr.
`methotrexate, and Vincu alkaloids (11. 12). Herein, we rcp
`the results of preclinical efficacy tests with C242-DM1.
`'maytansinoid drug (DMl) linked to the monoclonal antibt
`C242 directed against human colorectal cancer.
`
`The maytansitmoid drug DMl is 1100- to 1000?
`ABSTRACT
`fold more cytotoxic than anticancer drugs that are currently
`in clliniml ~u'seli'The immunoconjugate C242-DM1 was pre-
`pared by conjugating DMll to the monoclonal antibody C242,
`which recognizes ,--a :mucin-type giycoprotein expressed to
`various extents by human collomectal cancers. C242-BM] was
`found to be highly cytotoxic toward cultured coion cancer cells
`in an antigen-specific manner and showed remarkable anti-
`tumor eff'mcy in viva. C242-0M1 cured_mice hearing subcu-
`taneous COMO 205 human colon tumor xenografts (tumor
`size at time of treatment 65—130 films), at doses that showed
`very little toxicity and were well below the maximum tolerated
`dose. CMZ-DMI could even effect complete regressions or
`cures in animals with large (260- to SM—mm’) COIJO 205
`tumor xenografts. Further, C242-D311 induced complete re-
`gressions of subcutaneous LoVo and HT—Z9 colon tumor
`xenografts that express the target antigen in a heterogeneous
`manner. C242-0M1 represents a new generation of immuno-
`conjugates that may yet fulfill the promise of effective cancer
`therapy through antibody targeting of cytotoxic agents;
`_____________’_._.———————————
`Colorectal cancer is one of the most common malignancies and
`is among the leading‘causcs of death from cancer. Surgical
`resection is the primary treatment modality for these tumors,
`but about half of all patients will die of disseminated disease
`(1). Because of the high incidence and poor prognosis of
`patients with metastatic disease, successful treatment of colo-
`rectal cancer requires effective systemic therapy in addition to
`surgery, either as adjuvant treatment to surgery or for primary _
`treatment of those 25% of all patients for whom surgery alone
`cannot achieve a complete :ré'épdnse (2). Unfortunately, the
`conventional systemic treatment options for colon cancer,
`including radiation therapy, chemotherapy, and immunother—
`apy, have limited efficacy (3, 4). To date, 5-fluorouracil (S-FU)
`has served as the standard cytostatic drug for adjuvant therapy
`after surgery. However, the overall response rate to 5-FU is
`less than 25%, and the treatment has not significantly im-
`proved patient survival (1—3). Although the improved regimen
`of 5-FU plus levamisolc in the adjuvant setting has proven to
`be more effective in patients with stage 11 and Ill colorcctal
`cancers, the estimated reduction in the mortality rate is still less
`than 30% (2, 5). Thus, there is an urgent clinical need for new
`agents with greater efficacy.
`Conventional chemotherapeutic agents are limited in their
`therapeutic effectivenessby'severe side effects due to their
`poor selectivity for tumors. The development of monoclonal
`antibodies against specific tumor antigens made it possible to
`think of enhancing the selectivity of anticancer drugs by a
`targeted delivery approach. However, several such reported
`____’_______________.————’——
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked “advertisement" in
`accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`'
`
`;‘
`
`MATERTALS AND METHODS -
`Preparation of C242-DM1 Conjugate. Ansamitocin, _3
`(compound 1), provided by Takeda (Osaka) was converter
`the disulfide—Containing maytansinoid DM1 (compound
`(Fig. 1) as described (15). The C242 antibody, a murine lg
`(16), was provided by Pharmacia: C242—DM1 (compounc
`was prepared as described (12). The conjugate was purifier
`gel filtration throngh a column of Scphacryl S300 and thep
`corresponding to monomeric conjugate (>80% overall 'in
`was collected. The final conjugate contained on the aver
`four DMl molecules linked per antibody molecule.
`Specific Affinity of C242-0M1. The specific binding affi
`of C242—DM1 conjugate and C242 antibody to CanAg-pos
`COLO 205 cell membranes was determined by a binding
`as described (17). Samples of C242-DM1 or C242 at vat
`concentrations (10": to 10‘9 M) were incubated for 18
`________________—————-—’
`Abbreviations: S-FU, S-fluorouracil; DMl. ntaylansinoid drug; 1\
`maximum tolerated dose.
`*To whom reprint requests should be addressed.
`
`8618
`
`
`
`PHIGENIX ._
`
`Exhibit 1023-01
`
`

`

`Medical Sciences: Liu et al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`8619
`
`(developed at lmmunoGen) to capture the C242-DM1. The
`amount of bound conjugate was then quantified by detection
`of the C242 antibody using lgGl-specific goat anti-mouse
`IgG-alkaline phosphatase/p-nitrophenyl phosphate as de-
`scribed (17).
`Immunostaining of Cells. Cells grown on coverslips were
`fixed with 2% paraformaldehyde, permeabilized in methanol
`at -20°C, and stained with C242 antibody for fluorescence
`microscopy as described (19). A similar protocol was used for
`flow cytometry (Becton-Dickinson FACScan), except
`that
`cells were trypsinized, stained live without fixation, and then fixed
`with 1% paraformaldehyde in phosphate-buffered saline (PBS).
`Magnetic Bead Depletion. Cells were harvested with trypsin,
`counted, incubated with C242 antibody, and washed. Cells
`were then mixed with magnetic beads (Dynabeads M450, goat
`anti-mouse IgG coated, Dynal, Oslo) at a beads/cells ratio of
`5:1 and incubated for 30 min with rocking at 4°C. Beads plus
`adhering cells were magnetically removed, and an equal num-
`ber of fresh beads were added for a second cycle. The
`remaining cells were analyzed by flow cytometry.
`
`RESULTS
`
`Evaluation of C242-DM1 for Specificity, Cytotoxicity, and
`Selectivity. The delivery agent of C242-DM1, the C242 anti-
`body, recognizes a sialidase-sensitive carbohydrate epitope on
`the CanAg antigen, a mucin-type glycoprotein expressed to
`. various degrees by all human colorectal cancers (20—22). C242
`has only minimal cross-reactivity with normal tissues (21, 22).
`‘C242-DM1 was prepared in a manner similar to that described
`for other maytansinoid conjugates (12) (Fig. I), The conjugate
`contains, on the average, four covalently linked DMl mole-
`cules per antibody molecule. In a binding assay. C242-DM1
`binds as well as unconjugated C242 to the CanAg antigen
`expressed on COLO 205 cell membranes (Fig. 24). indicating
`
`c
`a,
`..
`u
`:3
`3
`3
`g
`u
`g
`3
`E
`
`.2
`
`2.0
`
`1.5
`
`1.0
`
`0.5
`
`0.0
`10'" 10'“
`10°
`
`104°
`
`1
`
`
`
`
`
`
`10°
`
`U
`g
`U 10"
`v.5
`E 10d
`..
`g
`104
`H.
`u 10".
`~55”;
`'E 10‘
`=
`' w 10‘
`"
`
`10°
`
`.2
`
`“N
`
`B
`
`-
`10-10 10‘
`
`161
`
`164‘
`
`D
`
`4
`
`a
`'3
`E 10
`‘5
`
`.
`
`E
`.E
`an 104
`>‘
`.—
`E
`u”:
`
`a
`
`10
`
`'—-'1o.w '1a."""'b1.
`Concentration. M
`
`a
`u
`5° 10‘
`E
`3
`
`E
`_
`go 10‘
`.2
`g
`t.
`'3 104
`
`16‘
`15-0
`1040
`Concentration. M
`
`13-7
`
`01.0
`
`0
`en,
`it
`c A
`it
`0
`cu,
`
`enema—n
`
`'
`
`
`1 ANSAMITOCIN P-3
`
`2 DMl: R 1: Me
`3 C241-DM1: R = C242
`
`FIG. 1. Structural representation of Ansamitocin P-3 (compound
`1), DMl (compound 2), and C242-DM1 (compound 3).
`
`ambient temperature with COLO 205 cell membranes immo—
`bilized in 96-well plates. The membranes were then washed and
`the amount of bound conjugate or antibody was determined using
`a Eu3+—labeled anti-mouse IgG. Results are plotted as concen—
`tration of conjugate or antibody versus relative fluorescence.
`In Vitro Cytotoxicity of C242-DM1 Conjugate.‘The cytotox—
`icity of C242-DM1 was measured on antigen-positive human
`colon carcinoma cell lines COLO 205 [American Type Culture
`Collection (ATCC) CCL 222], LoVo (ATCC CCL 229), and
`HT-29 (ATCC HTB 38) and on the antigen-negative human
`melanoma cell line A-375 (ATCC CRL 1619) in a clonogenic
`assay. Cells were plated in 96-well tissue culture plates with
`each plate containing a fixed number of cells (ranging from 3
`to 10,000 cells per well) in 0.2 ml of DMEM containing 20%
`fetal calf serum. Immunoconjugate at varying concentrations
`(4 X 10‘” to 4 X 10‘“ Mi) was added and the cells were
`maintained in a humidified atmosphere at 37°C and 6% CO:
`for 18—21 days. In some experiments, the cells were incubated
`with C242-DM1 for 24 h and then washed, and the medium was
`replaced with fresh medium without drug. Colonies were then
`counted and the plating efficiency was determined. Surviving
`fractions of cells were then calculated as the ratio of the plating
`efficiency of the treated sample and the plating efficiency of
`the control.
`lmmunohistochemical Studies. Tumor tissues excised from
`either humans ‘or mice were frozen in O.C.T. embedding
`medium (Miles), sectioned, and treated with biotinylated-C242
`antibody. The bound antibody was detected using the avidin-
`biotin immunoperoxidase technique as described (18).
`In Vivo Tumor Growth Assays. Female CB—17 SCID mice,
`6—7 weeks of age, were obtained from Massachusetts General
`Hospital. The human colon cancer cell lines COLO 205, LoVo,
`and HT-29 were maintained as adherent cultures in DMEM
`containing 10% fetal bovine serum at 37°C in a humidified
`atmosphere of 6% C02/94% air. Each mouse was inoculated
`subcutaneously at the right flank with tumor cells (2 X 106 to
`1 x 107 cells in different experiments) in 0.1 ml of medium.
`Treatments were started on days 7—9 after tumor inoculation,
`when the tumor sizes reached from 65 to 500 mm3, depending
`on the experiment. The therapeutic agents were administered
`intravenously to groups of 7—10 mice. Tumor size was measured
`weekly in two dimensions using a caliper, and the volume was
`expressed in mm3 using the formula: V = 0.5a X b2, where a and
`b are the long and short diameters of the tumor, respectively.
`Measurement of Concentration of C242-DIV“ in Serum. A
`group of eight CD1 mice obtained from Charles River Breed-
`ing Laboratories were injected with C242-DM1 at a dose of 300
`pg per kg per dayi for five consecutive days. Blood (0.1 ml) was
`withdrawn from the retroorbital sinus once per day from each
`mouse, either at 1 h (four mice) or at 24 h (four mice) after
`injection of the conjugate. C242-DM1 was determined by an
`ELISA using a murine monoclonal IgGZa anti-DMl antibody
`
`iAll concentrations used in vitro refer to conjugated DM].
`tAll doses used in viva refer to conjugated DM1. A DMl dose of 1 [Lg
`corresponds to 54 pg of C242-0M1 conjugate.
`
`FIG. 2. Evaluation of binding and cytotoxicity of C242-UM]. (A)
`Binding affinity of C242-0M1. The specific affinity of conjugated
`C242 (0) for the CanAg antigen expressed on COLO 205 cell
`membranes is compared with that of C242 (0). (B) In vitro cytotoxicity
`and selectivity of C242-DM1. The in vitro cytotoxicity of C242-DM1
`for antigen-positive human colon carcinoma COLO 205 cells (0) and
`antigen-negative human melanoma A-375 cells (0) was measured in a
`clonogenic assay. (C) In vilro cytotoxicity and specificity of C242-DMl
`for antigen-positive colon carcinoma COLO 205 cells in the absence
`(0) or presence (0) of l X 10‘“ M C242 antibody. Cells were incubated
`with immunoconjugate for 24 h in this experiment. (D) In vitro
`cytotoxicity of C242—DM1 for the human colon cancer cell line LoVo,
`which expresses the CanAg antigen heterogeneously.
`
`
`
`PHIGENIX
`
`Exhibit 1023-02
`
`

`

`8620
`
`Medical Sciences: Liu at al.
`
`Proc. Natl. Acad. Sci. USA 93 (I996)
`
`that the conjugation of DM1 does not diminish the binding
`avidity of C242. The cytotoxic potency and selectivity of
`C242-DM1 was assayed with the antigen—positive COLO 205
`cell line and the antigen-negative A-375 melanoma cell line
`(Fig. 28); both cell lines were equally sensitive to free DM]
`(lCm = 4 X 10"l M). C242-BM) was found to kill COLO 205
`cells with an 1C50 value of 3.2 X 10‘” M (23.5 pg/ml), and
`treatment of cells with a concentration of 4.5 X 10'9 M (3.3
`ng/ml) left a surviving fraction of less than 1 X 10’5
`(>99.999% of cells killed, detection limit of the assay). In
`contrast, C242-DM1 was 1100-fold less cytotoxic for the
`antigen-negative A—375 cells (ICso = 3.6 X 10‘8 M; 26.5
`ng/ml), demonstrating that cell killing was selective for the
`antigen-positive colon cell line (Fig. 28). COLO'205 cells were
`killed even after a 24—h exposure to C242-DM1, with an IC50
`value of 6 X 10—10 M (Fig. 2C). Furthermore, a large excess of
`free C242 antibody greatly diminished the cytotoxicity of the
`conjugate toward the target cells (Fig. 2C), further demon-
`strating that the cytotoxic effect was dependent on specific
`binding through the antibody component of the conjugate.
`The COLO 205 cell line cultured in vitro expresses the target
`antigen homogeneously on all cells (22). We also evaluated the
`cytotoxic potency of C242-DM1.against two colon tumor cell
`lines, LoVo and FIT-29, which express the CanAg antigen
`heterogeneously on only 20—30% of their cells when grown in
`vitro, as judged by indirect immunofluorescence analysis of
`C242 binding using flow cytometry (data not shown). In spite
`of this low expression, treatment of these cells with C242-BM]
`could eliminate 99% of the cells at a concentration of 4 X 10—9
`M (shown in Fig. 2D for the LoVo cell line).
`
`Immunohistochemical Analysis of Tumor Xenografts and
`Human Colon Tumor Samples. The three human colon tun. :r
`cell lines, COLO 205, LoVo and HT-29, were grown subur-
`taneously in SCID mice to test the in viva therapeutic efficacy
`of C242-BM]. The particular cell lines were chosen because
`their antigen expression, when grown in viva, was in the range
`of that seen by immunohistochemical examination of human
`colon tumor specimens from 20 patients. COLO 205 tumor
`xenografts excised from mice on day 7 after tumor inoculation
`exhibited, on immunohistochemical analysis, uniform staining
`of the CanAg antigen (Fig. 3A) in a manner similar to that of
`the section of a human colon tumor biopsy representative of
`6/20 specimens shown in Fig. 38. Tumor xenografts estab-
`lished with LoVo cells expressed the antigen heterogeneously
`at all time points. The staining pattern of a section taken on
`day 7 after tumor inoculation was classified as moderately
`heterogeneous (Fig. 3C) and resembled the staining pattern of
`the typical (10/20 specimens) human colon tumor biopsy
`shown in Fig. 3D. The third human colon tumor xenograft
`model established with HT-29 cells showed very heteroge-
`neous staining for antigen, with many cells being antigen-
`negative (Fig. 3E), again in a fashion similar to that seen in
`some biopsies (4/20) of human colon tumors (Fig. 3F).
`Antitumor Efficacy of C242-DM1.
`ln the first
`therapy
`experiment (Fig. 4A), animals bearing COLO 205 tumors were
`treated with five daily injections of C242-DM1 at a dose of 300
`[.Lg per kg per day, with an equivalent dose of the isotype-
`matched conjugate N901—DM] that does not bind to COLO
`205 cells,-or with a mixture of corresponding amounts of C242
`antibody (16 mg per kg per day) and unconjuguted DMl (300
`ug per kg per day). Treatment with C242—BM] completely
`
`' 4.40.11. 3'94.
`
`
`
`Immunohistochemical analysis of tumor xenografts excised on day 7 after tumor inoculation and comparison with human colon tumor
`FIG. 3.
`biopsies. (A) A COLO 205 xenograft. (B) A human colon tumor biopsy with homogeneous expression of antigen. (C) A LoVo xenograft. (D) A 7
`human colon tumor biopsy with moderately heterogeneous expression of antigen. (E) An HT-29 xenograft. (F) A human colon tumor biopsy with
`very heterogeneous expression of antigen. (G) A relapsed LoVo xenograft removed on day 91 from a mouse that received one course of C242-BM]
`treatment.
`
`M
`
`PHIGENIX
`
`Exhibit 1023-03
`
`

`

`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`8621
`
`given for 5 consecutive days, which is 59% of the MTD. Even
`at the lowest dose tested (150 pg per kg per day for 5 days), a
`' significant delay in tumor growth was observed.
`These results encouraged us to evaluate the therapeutic
`efficacy of C242-DM1 in mice bearing larger (average size, 260
`mm-‘) subcutaneous COLO 205 xenografts (Fig. 5A). Animals
`received two courses of 5-day treatment with C242-DM1 or,
`for comparison, treatment with 5-FU, the standard chemo-
`therapeutic drug used for the treatment of colorectal cancer.
`C242-DM1 again cured all animals rendering them tumor-free
`for greater than 200 days (duration of the experiment). This
`therapeutic effect on large tumors is especially remarkable in
`view of the finding that S-FU at its MTD (15 mg per kg per day
`for 5 days) only slightly (by about 5 days) delayed the tumor
`growth. We extended this study to even larger tumors. A group
`of animals bearing the largest COLO 205 tumor xenografts
`tested (average size 500 mm3) was treated with one course of
`C242-DM‘1 at a dose of 300 pg per kg per day for 5 days (Fig.
`SB). Complete tumor regressions were achieved in all animals.
`In six out of eight animals, the complete response lasted 7
`weeks. In the remaining two animals, no signs of tumor could
`be detected when the experiment was terminated on day 120
`after tumor inoculation (representing more than 17 tumor size
`doubling times in viva).
`-
`The COLO 205 cell line, both cultured in vitro and grown as-
`tumor xenografts, expresses the target antigen homogeneously
`on all cells (Fig. 3/1). We then evaluated the antitumor activity
`of C242-DM1 against established colon tumor xenografts from
`the LoVo and HT-29 cell lines that express the CanAg antigen
`heterogeneously on only 20—30% of their cells when grown in
`
`1500 A
`
`900
`
`600
`
`C
`
`O
`
`
`
`
`
`TumorSize(rnm3)
`
`>
`
`20
`
`.
`40
`
`.
`
`_'
`80 1100 120
`
`6O
`
`VA
`160
`
`A
`, A
`ISO 200
`
`140
`
`1800
`
`" 1500
`
`900
`
`1200
`
`j§§
`
`
`
`
`
`
`_TumorSize(mm3)
`
`0
`
`160
`120
`80
`4o
`Days After 'l'umor Inoculation
`
`In
`
`2
`
`
`
`FIG. 4. Antitumor activity of C242-DM1 conjugate in SCID mice
`bearing COLO 205 human colon tumor xenografts. Each mouse was
`inoculated with 2 X 10" COLO 205 cells. The treatments were given
`from day 7 to day 11 after tumor inoculation (average tumor size =
`65—100 mm-‘). (A) Antigen-specific antitumor activity of C242-DMI.
`The antitumor activity of C242-DM1 (300 pg per kg per day for 5 days)
`(A) was compared with that of PBS (0.2 ml per mouse per day for 5
`days) (0). a mixture of C242 (16 mg per kg per day for 5 days) plus free
`DMl (300 pg per kg per day for 5 days) (D) or a nonbinding conjugate,
`N901-DM1 (300 pg per kg per day for 5 days) (0).
`([3) Dose
`dependence of antitumor activity of C242—DMl. Tumor-bearing ani-
`mals were treated with PBS (0.2 ml per mouse per day for 5 days) (0).
`C242-DM1 (150 pg per kg per day for 5 days) (0). C242-DM1 (225 pg
`per kg per day for 5 days) (A). or C242-DM1 (300 pg per kg per day
`for 5 days) (CI).
`
`eliminated any measurable tumors within 2 weeks of the
`initiation of therapy, and all eight animals were tumor—free for
`200 days (duration of the experiment). Furthermore, toxic side
`effects were minimal at this dose as judged by the absence of
`body weight loss. The dose of C242—DM1 used in this exper-
`iment was below its maximum tolerated dose (MTD), which
`was defined for these experiments ,as the highest dose that
`could beadrninistered to tumor—bearing mice without causing
`Arm-related deaths-(MTD = 380 pg per kg per day for five
`consecutive days). In contrast, very little antitumor activity was
`observed in animals treated with nontargeted conjugate or
`with the mixture of antibody and free DMl (Fig. 4A). Thus, the
`DMl moiety is a potent
`therapeutic agent against colon
`cancers in vivo when targeted to the tumors as a conjugate with
`the C242 antibody and shows high antitumor efficacy at doses
`that cause little toxicity.
`The circulating serum concentrations of C242-DM1 were
`determined in CD1 mice by ELISA. Que
`after-teach
`
`injectiouifive-dailyinjeotions‘of 300i?
`"‘day), the
`sconoentmtio'nt‘of (QAZSDMI-,wasmbout:1a8mM;~equivalent to
`"DMl' at 1.3 pg/mL After.24.h,~~.theserum-conoentration was
`about 0.76 'pM,v-whichlis;still 58-‘f01d'high’er than the concen-
`tration required to kill =>'99.999%-‘oeus ill-vim).
`Next, the dose—response effect of C242—DM1 in the COLO
`205 xenograft model was evaluated. Animals were treated with
`C242—DM1 at doses ranging from 150 to 300 pg per kg per day
`for 5 days (Fig. 4B). C242-DM1 eliminated tumors in all
`animals at a daily dose as low as 225 pg per kg per day when
`
`1200
`
`
`
`
`0
`20
`40
`60
`80
`100
`120
`Days After Tumor Inoculation
`
`FIG. 5. Antitumor activity of C242-DM1 against large COLO 205
`tumors. Each mouse was inoculated with 5 X 10" COLO 205 cells and
`treatments were started on day 7 after tumor inoculation. (A) Efficacy
`in treatment of large COLO 205 xenografts (mean tumor size = 260
`mm3). Tumor-bearing animals were treated with PBS (0.2 ml per
`mouse per day for 5 days) (0), 5-FU (15 mg per kg per day for 5 days)
`(1:1),or two courses of C242-DMl (300 pg per kg per day for 10 days:
`days 7—11 and days 14—18) (A). (8) Efficacy in treatment ofvery large
`COLO 205 xenografts (mean tumor size = 500 mm“). Tumor-bearing
`animals were treated with PBS (0.2 ml per mouse per day for 5 days)
`(0). 5-FU (15 mg per kg per day for 5 days) (El), or one course of
`C242-DM1 (300 pg per kg‘per day for 5 days. days 7—11) (A).
`
`
`
`PHIGENIX
`
`Exhibit 1023-04
`
`

`

`
`
`
`
`__.,-..._-,_.‘;—.____._‘.__‘———L—
`
`8622
`
`Medical Sciences: Liu el al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`vitro or, as shown in Fig. 3 (C and E), in viva. Animals bearing
`LoVo tumor xenografts were treated with either one or two
`courses of C242-DM1 (300 pg per kg per day for 5 days). Two
`additional groups of tumor-bearing animals were treated with
`either a mixture of C242 antibody (16 mg per kg per day for
`5 days) and unconjugated DMl (300 pg per kg per day for 5
`days) or with S-FU at its MTD (15 mg per kg per day for
`5 days). Remarkably, complete tumor regressions lasting 5
`weeks were observed in all animals treated with one course of
`C242-DM1 (Fig. 6A). The LoVo tumors from mice that were
`treated with C242-DM1 and relapsed after the period of
`complete regressions were evaluated for antigen expression. A
`section of the tumor excised on day 91 exhibited similar
`heterogeneous staining with C242 (Fig. 30) as was seen prior
`to therapy (day 7, Fig. 3C). The period of complete regression
`could be prolonged to 9 weeks by a second course of treatment
`with C242-BM] initiated 21 days after the start of the first
`c0urse (Fig. 6/1). No indication of toxic side effects as assessed
`by body weight loss was observed for either treatment protocol.
`In contrast, tumors in animals that were treated with the
`mixture of antibody and DM1, or with S-FU, grew rapidly to
`large sizes (Fig. 6A). Similar effects were obtained in the tumor
`model with HT-29, classified as expressing the CanAg antigen
`very heterogeneously (see Fig. 3E). C242-DM1 (375 pg per kg
`per day for 5 days) induced complete tumor regressions lasting
`4 weeks (Fig. 6B). These results demonstrate that C242-DM1
`is an effective therapeutic agent against colorectal cancer xe-
`
`
`
`
`
`TumorSize(mm3)
`
`
`
`Days After Tumor Inoculation
`
`in SCID mice bearing
`FIG. 6. Antitumor activity of C242-DMl
`human colon tumor xenografts that express the antigen heteroge-
`neously. (A) Antitumor activity of C242-0M1 against LoVo tumor
`xenografts (mean tumor size = 103 mm-‘). Each mouse was inoculated
`with 5 X 10" LoVo cells, and treatments were started on day 9 after
`tumor inoculation. Tumor-bearing mice were treated with PBS (0.2 ml
`per mouse per day for 5 days) (0), S-FU (15 mg per kg per day for 5
`days) (i), a mixture of C242 (16 mg per kg per day for 5 days) plus
`DMl (300 pg per kg per day for 5 days) (A), one course of C242-BM]
`(300 pg per kg per day for 5 days) (A), or two courses of C242-DMl
`(300 pg per kg per day for 10 days, days 9-13 and‘days 30-34) (D). (B)
`Antitumor activity of C242-BM] against HT-29 tumor xenografts
`(mean tumor size = 130 mm3). Each mouse was inoculated with 1 X
`107 HT—29 cells and treatments were given from days 7 to 11 after
`tumor inoculation. The tumor-bearing mice were treated with PBS (0.2
`ml per mouse per day for 5 days) (0) or C242-DM1 (375 pg per kg per
`day for 5 days) (A).
`
`-" 2
`
`.
`
`nografts including those that express the antigen heterogenc: #181»
`and causes little toxicity even after two courses of treatmcn; '
`Relationship Between CanAg Expression and Colony Fur-
`mation in Colon Cancer Cell Lines. The rather unexpected in
`viva results that C242-BM] induced complete regressions of
`tumors in which 70—80% of the cells do not express detectable
`amounts of the CanAg antigen led us to evaluate whether it was
`possible in vitro to generate homogeneous antigen-positive and
`antigen-negative sublines from cell
`lines that express the
`antigen heterogeneously. Repeated attempts to subcl nesthe
`LoVo and PIT-29 cell lines to select sublines that were com-
`pletely. antigen-negative (by staining with (3242) were uri§tic~
`cecsfulgtno CanAg-negative subclones grew in a total of 210
`subclones. In fact, most subclones expressed more antigen
`(about 60% positive cells) than the parental cell lines (20—30%
`positive cells). In another attempt, an HT—29 cell culture was
`first depleted of antigen—positive cells using magnetic beads
`coated with the C242 antibody and the remaining cells, which
`had a very low level of CanAg expression (<1%, by flow
`cytometry analysis of C242 binding), were subcloned. Initial
`screening of subconfluent cultures by immunofluorescence
`showed that many clones expressed very low levels of CanAg
`(about 70% of 79 clones had <10% antigen-positive ccits).
`However, as the colonies were expanded, CanAg expression on
`some sublines .again increased to levels greater than in the
`parental line or else the colonies died out (only 34 sublines
`could be expanded sufficiently for flow cytometry analysis).
`These results suggest that there is a direct relationship in «my
`between the ability to form colonies and CanAg'expressi n
`and may’ also provide an explanation for the experiniefif‘gl
`observations in the LoVo and HT-29 tumor xenograft mode s_
`whereby the, elimination of antigen—positive cells in viv "J”
`C242-BM] may cause the collapse of the entire tumor in an..
`
`DISCUSSIOI\~
`
`C242—DMI represents a new generation of immunoconjugates
`that may yet realize the potential of effective cancer therapy
`through antibody targeting ol'cytotoxic agents. The conjugate
`is highly cytotoxic in t'ilro in an antigen-dependent and tumor-
`cell-selective manner and produced long-term cures of mice
`bearing human colon tumor COLO 205 xenografts at doses
`that caused little toxicity.,Cures were men obtained at t.‘.
`-:s
`that were well below the MTD of the conjugate. lmportaa..tly,
`the circulating serum concentrations of C242-DM1 (1.8 ”Wigg-
`I h after injection of 300 pg/kg) were about 380-fold highEr
`than that required to kill greater than 99.999% of target cells
`in vitro. C242-DM1 was capable of curing mice bearing very
`large COLO 205 tumors, even those that were 500 mm3 at the
`start of treatment. In contrast, 5—FU, the standard chemother—
`apeutic drug used for the treatment of colorectal cancer,
`showed very little therapeutic benefit against the same tumors.
`In many human colorectal tumor biopsies, it was demon—
`strated that CanAg, the target antigen for C242, is expressed
`in a heterogeneous manner. C242-DM1 showed remarkable
`antitumor activity in tumor models derived from‘ the LoVo and
`FIT-29 colon cancer cell lines that express the antigen on only
`20—30% of the cells. The immunoconjugate induced complete
`regressions of such tumors lasting 4—5 weeks. Tumors har-
`vested from regrowing LoVo xenografts after this period ol
`complete regression exhibited heterogeneous staining similar
`to that seen prior to treatment (see Fig. 3 C and G), suggesting
`that the phenotype of the regrowing tumor xenografts was
`unchanged. Indeed, the period of complete regression could be
`prolonged by a second course of treatment that was m'll
`tolerated by the mice, suggesting that using multiple cycles 01
`this immunoconjugate for treatment of colorectal cancer ma)
`be a feasible clinical regimen, with the potential for totall}
`eradicating even those tumors that show heterogeneous ex
`pression for the CanAg antigen. The similarity of antigcr
`
`.—
`
`PHIGENIX
`
`Exhibit 1023-05
`
`

`

`a-
`
`Medical Sciences: Liu el al.
`
`expression and tumor morphology between the xenografts and
`the human tumor biopsies suggests to us that antitumor activity
`of C242-BM] exhibited in these models may well predict
`pOtent antitumor activity in the clinical situation.
`.
`The in vitro experiments suggest
`that CanAg antigen-
`negative cells do not grow (for .long) in the absence of
`antigen-positive cells, which could provide a mechanism by
`which C242-DM1 could induce complete regressions of tumors
`that express the CanAg antigen heterogeneously. Secretion of
`cytokines essential for the survival of CanAg-negative cells by
`the CanAg-positive cells could explain these results, although
`initial experiments indicated that conditioned medium did not
`support growth of CanAg—negative colonies derived from the
`HT-29 or LoVo cell lines. Other possible mechanisms may be
`(i) all cell lineages go through a period of transient CanAg
`expression during which they are susceptible to the action of
`C242-DM1; (ii) a bystander effect, that is, antigen-positive
`cells may concentrate sufficient C242-DM1 at the tumor site
`to cause killing of neighboring antigen-negative cells either
`directly or by release of DM1 from the conjugate concentrated
`inside antigen-positive tumor cells, followed by diffusion of
`DM1 into neighboring cells via gap junctions; and (iii) recruit-
`ment of host immune mechanisms upon killing of CanAg-
`positive cells, including activated macrophages and natural
`killer cells (23). Also, it cannot be ruled out that CanAg-
`negative cells actually express antigen at low density, below the
`sensitivity of detection by flow cytometry or immunofluores-
`cence microscopy but sufficient to mediate cell death using this
`potent immunoconjugate.
`C242-DM1 exhibits a degree of potency and selectivity in
`vitro and in vivo that is superior to that of other immunocon-
`jugates against colorectal cancer described thus far (6-8,
`24—26). C242 has been linked to other cytotoxic agents, such
`as ricin A chain (24) or Pseudomonus exotoxin (25, 26). The
`ricin-A-chain conjugates showed very limited efficacy in the
`COLO 205 model (inducing-a tumor growth delay of 16 days)
`and no significant efficacy in tumor models that express the
`CanAg antigen heterogeneously (24). The Pseudomonas exo-
`toxin conjugates also showed ,littlc efficacy in the mouse
`models (25, 26). In one case where a “cure” was reported, the
`follow-up period was short (only 4 weeks), and the dose used
`was toxic and left only one survivor in the group (24). Of the
`other antibody-drug conjugates developed against colorectal
`cancer (6