Cancer and Metastasis Reviews 7: 263-284 (1988)
`© Kluwer Academic Publishers - Printed in the Netherlands
`
`Human tumor xenografts as model for drug testing
`
`Jurgen Mattern, Mihaly Bak, Eric W. Hahn and Manfred Volm
`Department of Experimental Pathology, German Cancer Research Center, Im Neuenheimer Feld 280,
`6900 Heidelberg, FRG
`
`Key words: human tumor xenografts, immune-deficient animals, drug testing, chemotherapy
`
`Abstract
`
`This paper reviews the history of xenografts, the endpoints commonly used to evaluate response and
`chemotherapeutic results obtained with serially maintained human tumor xenografts from different lab(cid:173)
`oratories, and discusses the potential clinical relevance of the heterotransplant model for cancer chemother(cid:173)
`apy. Specifically, an attempt is made to correlate the published xenograft data with the clinical data. Drug
`testing with different types of xenotransplanted tumors has shown that the response of xenografts obtained in
`immune-deficient animals is comparable to that in clinical practice. In addition, xenografts of a particular
`tumor type are able to identify agents of known clinical activity against that disease.
`
`Introduction
`
`There has been a progressive increase in the use of
`heterotransplants of human tumors in immunolog(cid:173)
`ically incompetent laboratory animals as an experi(cid:173)
`mental model for cancer research. The interest in
`this tumor model derives in part from dissatisfac(cid:173)
`tion with transplantable rat and mouse tumors be(cid:173)
`cause they display a spectrum of drug sensitivities
`that are often not related to the drug sensitivity of
`human tumors. Certainly, empirical screening
`against transplantable animal tumors has produced
`a range of potentially active drugs, but the clinician
`is still faced with the problem of selecting one drug
`or a combination of drugs that might be effective in
`an individual patient.
`Efforts have been made to develop methods for
`predicting the drug sensitivity of tumors removed
`from the patient. The approaches include either in
`vitro culture techniques, or xenografts in immun(cid:173)
`ologically incompetent laboratory animals. All the
`in vitro systems have the obvious disadvantage that
`the tumor cells are cultured in an artificial envi-
`
`ronment, and the influence of the metabolism by
`the host as well as pharmacokinetic properties of
`the drug are lost. For these reasons it is thought
`that the experimental results achieved with human
`tumors growing in xenogenic hosts may perhaps be
`more clinically relevant. Recently , Steel er al.
`[123], reporting on their experience at the Institute
`of Cancer Research in Londen, concluded that the
`human tumor xenografts responded to drug treat(cid:173)
`ment in a way that would be expected on the basis
`of clinical experience.
`In the past, not all human tumors grew in im(cid:173)
`munologically incompetent animals. However,
`manipulation of the grafting techniques and the
`introduction of the athymic nude mouse has led to
`the successful propagation of almost all known va(cid:173)
`rieties of human malignancy , and most of them are
`presently being grown serially. The literature deal(cid:173)
`ing with the use of human tumor transplants for
`drug response evaluation is enormous. However,
`most studies have dealt with only a few tumors
`under specific circumstances. and comparative
`studies, e.g., with a panel of tumors of a certain
`type, are uncommon.
`
`1 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`264
`
`The present paper reviews the history of xe(cid:173)
`nografts, the endpoints commonly used to evaluate
`response, and chemotherapeutic results obtained
`with serially maintained human tumor xenografts
`from different laboratories, and discusses the po(cid:173)
`tential clinical relevance of the heterotransplant
`model for cancer chemotherapy. Specifically, this
`paper attempts to correlate the published xeno(cid:173)
`graft data with the clinical data reported by Was(cid:173)
`serman et al. (130] , who has summarized the single
`agent therapy results in the clinic.
`
`Background
`
`Model systems
`
`The concept of testing the sensitivity of human
`malignancies to cytostatic agents after transplanta(cid:173)
`tion to a foreign host animal is not new. There have
`been many attempts to establish a useful hetero(cid:173)
`transplantation system for this scope, but the re(cid:173)
`sults were mostly unsatisfactory. Human tumors
`have been succesfully transplanted to immunolog(cid:173)
`ically privileged sites in the organism, where rejec(cid:173)
`tion phenomena are not so pronounced, such as the
`anterior chamber of the eye [9, 52], the brain [39,
`106], and the hamster cheekpouch [51] (Table 1).
`Some of these systems were used for cancer chemo(cid:173)
`therapy evaluations, but none fulfilled early expec(cid:173)
`tations. The technical difficulties involved with
`each of the model systems were considerable. The
`
`Table 1. Heterotransplantation systems.
`
`Immunologically privileged sites
`Anterior chamber of the eye
`Brain
`Cheekpouch of the hamster
`Immunologically incompetent organisms
`Embryos
`Newborn animals
`Irradiated animals
`Cortisone-treated animals
`Thymectomized animals
`Antilymphocyte serum-treated animals
`Thymusaplastic nude mice and rats
`Thymusaplastic and asplenic nude mice
`
`anterior chamber of the eye allowed the growth
`and testing of only small amounts of tumor. The
`growth of tumors in the brain is not visible, and is
`therefore impossible to measure. In the hamster
`cheeckpouch model , animals had to be anaesthe(cid:173)
`tized, and tumor growth was predictable only over
`a short period of time, since tumors tended to
`ulcerate or to regress spontaneously. However, the
`major limitation of all these models for cancer che(cid:173)
`motherapy studies derives from the low tumor take
`rate and the difficulty of obtaining serially estab(cid:173)
`lished lines.
`Growth of human tumor xenografts can also be
`achieved in embryos and newborn animals [1 , 19}
`where the immune system is not completely devel(cid:173)
`oped, or by non-specific suppression of the host's
`immune system by radiation [3, 77], antilympho(cid:173)
`cyte serum [ 41], or corticoids [92]. The best results
`were achieved with animals that were either thym(cid:173)
`ectomized, lethally irradiated and reconstituted
`with bone marrow (17, 122, 124], or with animals
`that were thymectomized and given a dose of cyto(cid:173)
`sine arabinoside before a lethal dose of whole-body
`radiation [124]. However, these manipulations had
`their drawbacks. They were time-consuming and
`required skilled personnel, and the animals treated
`in this manner were fragile and susceptible to in(cid:173)
`fections. Perhaps most important was that the im(cid:173)
`munosuppression usually lasted only about 5-{)
`weeks, and longer-term studies were not possible.
`The most important advance in human xenograft
`models has been the use of the athymic nude
`mouse. These animals, which result from the inher(cid:173)
`itance of a recessive mutation, are virtually hair(cid:173)
`less, and exhibit thymus aplasia [26, 91]. As a con(cid:173)
`sequence, the formation ofT-lymphocytes is inhib(cid:173)
`ited [46]. This impairment of cellular immunity and
`the resulting reduced capacity to reject 'foreign'
`tissue permits the successful transplantation of hu(cid:173)
`man tumors without additional immunosuppres(cid:173)
`sion. With the increased availability of the nude
`mouse and the 'lasat' mouse (both athymic and
`asplenic), many types of human tumors have been
`successfully xenografted and serially transplanted.
`Aside from being relatively expensive, the disad(cid:173)
`vantages are that these mice are subject to infection
`and require sterilized food and water and special
`
`2 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`handling and housing. Nevertheless, under these
`conditions 'long-term' studies are possible, where(cid:173)
`as with earlier model systems they were not.
`
`Take rate
`
`Today, almost all of the known human malignan(cid:173)
`cies have been successfully transplanted and serial(cid:173)
`ly maintained in immune-supressed mice or nude
`mice. However, tumors of different anatomical ori(cid:173)
`gin grow with varying degrees of success when
`implanted subcutaneously as fresh surgical tissue.
`The reported take rates of five main tumor types in
`nude mice or in immune-suppressed mice are listed
`in Table 2.
`Colorectal and ovarian tumors exhibit the high(cid:173)
`est take rates ( 60%- 70%) of the most frequently
`studied tumor types, followed by melanomas and
`lung tumors (50%-60% ). Breast tumors seem to be
`the most difficult to grow as xenotransplants, with
`only about 27% takes. The reasons for these re(cid:173)
`ported differences in takes among the various tu(cid:173)
`mor types are not known. The differences in tumor
`take between laboratories may be partially due to
`the definition of 'tumor take'. The term 'take' is not
`precisely defined, and is often used in an arbitrary
`fashion. The vast majority of investigators define a
`succesful take as the progressive growth of the
`tumor. but some groups considered cell viability in
`a static nodule as a successful take (25].
`Since it is generally not feasible to carry out
`chemotherapy experiments on primary tumor
`transplants, it is essential to establish tumors that
`can be serially transplanted. However. the succes(cid:173)
`ful development of transplantable heterotrans(cid:173)
`plants remains low. In general, only about one-half
`of all transplants growing in the first passage be(cid:173)
`come established lines. However, tumors carried
`beyond passage 3 have a higher chance of becom(cid:173)
`ing established (31).
`Comparative investigations of tumor take rates
`in different hosts from one single laboratory are not
`available. However, a review of the literature re(cid:173)
`veals that the differences in take rates among the
`specific tumor groups is not just confined to nude
`or immune-suppressed mice, but is also seen in
`
`265
`
`other heterotransplantation model systems. This
`suggests that success in establishing growth is a
`property of the individual tumor type itself, rather
`than a property of the host system used.
`Many variables can affect the frequency of take
`of tumor transplants. The degree of success de(cid:173)
`pends on the properties of both tumor and host
`(Table 3). In addition to the species of animal, age
`is an important factor. Three-week old nude mice
`have a depressed natural killer-cell (NK) activity as
`compared to older nude mice [ 54 J, and the NK-cell
`activity, by its ability to kill tumor cells, can influ(cid:173)
`ence the metastatic spread [55]. Curiously, hor(cid:173)
`mone-dependent human tumors appear more
`difficult to establish in nude mice. There are re(cid:173)
`ported differences in takes between male and fe(cid:173)
`male nude mice for mammary carcinoma (112),
`prostatic carcinoma (57] , small cell carcinoma of
`the lung (93], and melanoma (128]. The genetic
`background of the nude mice has also been suggest(cid:173)
`ed as a variable for tumor takes (76], but this could
`not be confirmed [109]. The nude mouse and the
`immune-suppressed mouse are very susceptible to
`infections that can considerably shorten the life
`span of these animals. The shorter life span thus
`leads to lower observed take rates of the usually
`slower growing human tumors. Sickness from in(cid:173)
`fection can alter the animal's caloric intake and
`cause severe weight loss which in turn can reduce
`the growth of the tumor [44]. Another factor is that
`an animal with an active infection has a stimulated
`B-celJ activity which favours the rejection of a
`transplanted tumor [74].
`The site of transplantation has been shown to be
`a factor that causes the rates of primary xenografts
`to vary. The anterior lateral thoracic wall appears
`to be a better transplant site for subcutaneous tu(cid:173)
`mor growth than the posterior part of the trunk
`[73]. In other reports intracranial (39] and kidney
`capsule (19] transplantation were found to be supe(cid:173)
`rior to the subcutaneous route. Other workers who
`compared various sites of implantation in the mice,
`including subcutaneous, kidney capsule, intraplen(cid:173)
`ic, foodpad, and intracranial, reported that for the
`majority of tumors the subcutaneous site or the
`kidney capsule worked equally well [100). Certain
`endocrinologically active tumors, however, were
`better accepted intrasplenically (99].
`
`3 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`266
`
`Table 2. Growth of various human tumor types transplanted subcutaneously into immune-deficient animals.
`
`Tumor type
`
`Total number
`
`Number growing
`
`Number of lines
`
`References
`
`Breast
`
`Total
`(%)
`
`Ovarian
`
`Total
`(%)
`
`Lung
`
`Total
`(%)
`
`Melanoma
`
`Total
`(%)
`
`Colorrectal
`
`Total
`(%)
`
`Total
`(%)
`
`200
`188
`93
`87
`32
`28
`28
`19
`18
`
`693
`
`215
`51
`23
`16
`12
`11
`10
`
`338
`
`213
`51
`50
`47
`37
`29
`18
`
`11
`10
`
`466
`
`46
`32
`28
`19
`12
`9
`7
`
`153
`
`83
`24
`18
`14
`9
`9
`7
`4
`
`168
`
`1818
`
`49
`77
`8
`14
`11
`15
`6
`4
`2
`
`186
`27
`
`65
`21
`16
`10
`5
`2
`2
`
`221
`65
`
`96
`24
`41
`24
`12
`13
`12
`
`6
`5
`
`233
`50
`
`31
`14
`14
`17
`3
`5
`6
`
`90
`59
`
`65
`13
`13
`8
`6
`5
`2
`3
`
`115
`68
`
`845
`46
`
`18
`11
`7
`6
`4
`4
`0
`3
`1
`
`54
`8
`
`34
`9
`14
`3
`3
`]
`I
`
`65
`19
`
`59
`15
`30
`14
`5
`10
`4
`
`5
`5
`
`147
`31
`
`22
`7
`8
`8
`2
`5
`4
`
`56
`37
`
`46
`9
`7
`6
`5
`4
`1
`2
`
`80
`48
`
`402
`22
`
`Bastert et al. {8]
`Giovanella et al. [43)
`Fogh et al. [31 J
`Sharkey et al. [111 J
`Rae-Venter et al. [98)
`Kleine et al. [66]
`Mattern et al. [81]
`Sebesteny et al. [105]
`Shimosato et al. [88]
`
`Kleine et al. (66]
`Davy et al. [18]
`Friedlander et al. (34]
`Kullander et al. [72}
`Mattern et al. [81]
`Sharkey et al. [111)
`Ueyama [88)
`
`Mattern et al. [82]
`Fogh et al. [31]
`Fiebig et al. [24]
`Sharkey et al. [111)
`Shimosato [88)
`Gazdar et al. (39)
`Wynn-Williams &
`McCulloch [132)
`Pratesi et al. [97]
`Kistler et al. (88]
`
`Giovanella et al. f 43]
`Povlsen [88}
`Sordat & Merenda [88]
`Fiebig et al. [24]
`Shimosato f 88]
`Fogh et al. [30)
`Sharkey et al. [111)
`
`Fiebig et al. [24]
`Fogh et al. [29)
`Sharkey et al. (111}
`Sordat & Merenda [88]
`Povlsen & Rygaard [96]
`Pratesi et al. [97]
`Kawamura et al. [65}
`Shimosato (88]
`
`4 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`The frequency of take is also correlated with
`properties of the human donor tumor. Apart from
`the type of tumor, tissues resected from metastatic
`or recurrent sites have a greater chance of growing
`as heterotransplants than tissue from primary sites
`(111). Take rates are highest with tumors of a low
`grade of differentiation [111). Transplantation of
`cell cultures of human tumors have a take rate of
`about twice that of fresh surgical tissue [28). The
`proliferative activity of the donor tumor also seems
`to be an important factor for the take rate. Tumors
`that demonstrate serial growth in the nude mouse
`have more S-phase cells than tumors that do not
`grow [81 ]. Disaggregation of the tissue is usually
`carried out, but enzymes should be avoided to min(cid:173)
`imize damage to the tissue. Mechanical dissocia(cid:173)
`tion can be accomplished by mincing the tissue and
`injecting the minced tissue via a needle into the
`flanks of the recipients [80, 100]. An equally suc(cid:173)
`cessful technique involves implanting the animals
`with 2x 2x 2mm fragments or 0.5-1 x 5 x 5mm
`flat pieces into the flanks [24, 112]. Comparative
`investigations with minced tissue and fragments
`showed no differences (99, 100].
`Other factors that may also affect the success
`rate and that should be taken into account are as
`follows: the interval between removal of surgical
`biopsy material and transplantation in the host; the
`correct handling of this tissue during the transfer
`from surgery to the laboratory, and the length of
`time one waits for the development of a progressive
`growing tumor.
`The control of these variables has led to the
`successful growth of tumor types that had previous(cid:173)
`ly resisted xenografting, e.g. , prostatic carcinoma
`
`Table 3. Variables that affect transplantation.
`
`Host
`
`Species
`Age
`
`Sex
`Health of the animal
`Transplant site
`Life span of the animal
`
`Tumor
`
`Tumor type
`Origin (primary, metastatic,
`recurrent)
`Preparation of tumor tissue
`Histology
`Differentiation
`Proliferation
`
`267
`
`[99] and retinoblastomas [38]. In addition. many
`attempts have been made to further improve the
`existing success rate with human tumors in animals
`either by refining the transplantation techniques or
`by altering the host. For instance, an increased
`success rate of ovarian tumor Jines in nude mice
`was effected by injecting cyclophosphamide prior
`to subcutaneous implantation [12, 87). Cyclophos(cid:173)
`phamide was apparently capable of suppressing
`further immune response , which was present in the
`nude mice that caused tissue rejection. Similar sup(cid:173)
`pression was achieved by sublethal whole body
`irradi?-tion of nude mice; a human acute T lympho(cid:173)
`blastic leukemia line was thus able to grow (75}.
`Intraperitoneal pretreatment with India ink, which
`apparently inactivated the macrophages, also in(cid:173)
`creased takes [125]. The observation that young or
`newborn nude mice are more susceptible to tumor
`growth and metastasis suggests that defense mech(cid:173)
`anisms of the recipient mice may be at least partial(cid:173)
`ly responsible for the low incidence of tumor me(cid:173)
`tastasis in adult nude mice [23J.
`
`Growth characteristics
`
`The majority of subcutaneously xenografted tu(cid:173)
`mors grow as well-circumscribed nodules at the site
`of inoculation, without infiltration in the surround(cid:173)
`ing connective tissue. The vascular system and the
`supporting stromal elements originate from the
`host, whereas the tumor parenchyma is of human
`origin. Tumor nutrition is also of host origin. Each
`xenografted tumor line exhibits a characteristic
`growth pattern; the growth curve can be described
`by a Gompertz equation. In most cases the tumors
`initially exhibit an exponential growth , followed by
`a slowing of the growth rate (Fig. 1). Sometimes,
`however, tumor growth is irregular. For instance,
`abrupt changes to a faster or slower growth rate
`have been observed, and in some cases the tumors
`regress completely [100]. The latency period from
`inoculation until initial palpable sustained tumor
`growth varies considerably and has been reported
`to range from 10 to 190 days (for instance, for lung
`tumors) [116]. This varied latency period for lung
`tumors seems to be related to tumor doubling time
`
`5 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`•
`
`268
`
`Ix 10 2 J
`40
`
`30
`
`20
`
`"' ~ 10
`"'
`E
`:J
`0
`>
`0
`E
`:J
`I-
`
`2
`
`•
`
`•
`
`~I-..-~~~~~~~~~~~.-'
`20
`24
`28
`32
`36
`40
`16
`Oays alter transplantation
`
`Fig. I. Growth of human lung tumor xenograft HXLSS in nude
`mice.
`
`[80], which , during the first passage, can range
`from a few days to more than 2 months [82]. Similar
`volume doubling times have also been reported for
`other histological types of xenografts [34 , 104]. Re(cid:173)
`gardless of tumor histology in subsequent passages
`the latency period and the tumor doubling time
`usually become shorter [62, 83, 112, 123]. After the
`third passage the growth rate frequently stabilizes,
`and each tumor line exhibits an individual and
`characteristic volume doubling time, commonly
`ranging from 1 to 2 weeks. Thus, it is common to
`conduct therapeutic studies after passage 3. How(cid:173)
`ever, these tumor doubling times are considerably
`faster than in human tumors. Steel summarized
`growth data from patients with different types of
`tumors [121] and found that volume doubling times
`ranged from about 3 weeks to 3 months, with an
`overall mean of about 2 months.
`These differences in growth rate have been attri(cid:173)
`buted to a variety of factors. For instance, it has
`been suggested that tumors growing in mice are
`either selected for rapid growth or that they in(cid:173)
`crease their growth rate in the new environment.
`Another factor contributing to this difference be(cid:173)
`tween tumor doubling times in situ versus xeno(cid:173)
`grafts may lie in a difference in cell kinetics. The
`
`cell cycle times of the xenografts were within the
`range of those reported in humans [94]; however,
`xenografts in their first passage in mice had a higher
`proportion of S-phase cells [83] and a higher mitot(cid:173)
`ic index [16, 107] than the surgical specimen from
`the patient. A further explanation could be that the
`tumors studied in patients were larger than those
`measured in mice. Since the volume doubling time
`of tumors often increases with size due to an in(cid:173)
`crease in cell loss rather than to a decrease in the
`growth fraction (121), the observed differences in
`volume doubling times may be at least partly a
`consequence of difference in tumor volume [101].
`Nevertheless, it is generally accepted that a high
`growth fraction and low cell loss largely explains
`the differences in volume doubling time in xeno(cid:173)
`grafts and in humans [70]. It should be emphasized,
`therefore, that these changes in cell kinetics could
`lead to an increased responsiveness to prolifer(cid:173)
`ation-dependent chemotherapy and that thera(cid:173)
`peutic results achieved with xenografts are prob(cid:173)
`ably overestimated.
`
`Maintenance of characteristics
`
`The extent to which xenografts maintain the hist(cid:173)
`ologic characteristics of the human donor tumor is
`important in fully evaluating their usefulness as
`models of human cancer. The microscopic appear(cid:173)
`ance of the xenotransplants has generally shown a
`close similarity to the human donor tumor (45, 111,
`112]. This appearance remains constant even after
`long-term passage in the animals, with the excep(cid:173)
`tion that there has often been a pronounced loss of
`tumor stroma after multiple transplantations (112].
`Even the degree of histological differentiation is
`usually well maintained, although there is general(cid:173)
`ly a tendency towards progressive dedifferentiation
`with further passages (80, 110), and in several in(cid:173)
`stances there was increased differentiation [110].
`On the ultrastructural level, human xenotrans(cid:173)
`plants maintain or have maintained a high degree
`of similarity to the donor tumors [ 127).
`Human karyotypes [126], cellular DNA content
`[21, 34], and human isoenzymes [61) were also
`retained during serial passage of xenografts. Occa-
`
`6 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`sionally, the induction of murine tumors following
`human tumor transplantation has been observed
`(61, 120]. These s.c. tumors were invasive, pene(cid:173)
`trating both the muscularis and the epidermis (120].
`It has been proposed that a transformation results
`from the hybridization of the human tumor and
`mouse cells [50], although this still needs to be
`verified.
`Other biochemical characteristics have also been
`found to be maintained during xenografting, in(cid:173)
`cluding the presence of alpha-fetoprotein (14, 127),
`carcinoembryonic antigen (34, 118] and epithelial
`membrane antigen [6], as well as the production of
`hormones [14, 113]. In most instances, when a
`marker was present in a primary tumor it was also
`detected in thexenograft. On the other hand, it was
`also observed that markers were detected in xe(cid:173)
`nografts established from marker negative primary
`tumors (14]. The extent to which chemotherapeutic
`sensitivity is maintained in the xenografts is dis(cid:173)
`cussed later.
`There are also a number of differences that exist
`between tumors in humans and tumors that are
`subsequently xenografted (Table 4). The changes
`in tumor cell kinetics and growth rate were dis(cid:173)
`cussed earlier. In addition, human xenografts
`maintained subcutaneously in laboratory animals
`grow as well-circumscribed nodules at the site of
`inoculation without local invasion, and rarely me(cid:173)
`tastasize [108]. In contrast, established tumor cell
`lines transplanted into the kidney capsule are in(cid:173)
`vasive, non-encapsulated, and have minimal ne(cid:173)
`crosis (85]. Further, human tumors growing in im(cid:173)
`mune-suppressed or nude mice have a mouse stro-
`
`Table 4. Features in the donor tumor and in the xenografted
`tumor.
`
`Similarities
`
`Differences
`
`Histology, ultrastructure
`Hormone production
`Production of tumor marker
`Chromosomes
`DNA content
`
`Cell cycle parameters
`Growth rate
`Metastatic spread
`Invasive propenies
`Stroma and vascularization of
`mouse origin
`Chemotherapeutic sensitivity Metabolism, phannacokinetics
`
`269
`
`ma [129] and are under the constant influence of
`the metabolism and pharmacokinetics of the host
`animal. Nevertheless, even with such limitations,
`the general consensus of the workers in the field is
`that the advantages of studying the drug response
`of the human tumor growing in a living organism,
`rather than in vitro cultured human tumor cells,
`outweigh the disadvantages.
`
`Testing procedures
`
`Drug dosage and toxicity in host
`
`Many drugs have been tested against a vast variety
`of human tumor xenografts. The drugs were gener(cid:173)
`ally given as single injection and the preferred
`route of administration has been intraperitoneal,
`followed by intravenous and subcutaneous routes.
`Very few laboratories have treated the xenografts
`in mice with therapy schedules equivalent to those
`received by patients (24, 37]. The doses usually
`applied to human tumors growing in nude mice are
`
`Table 5. Range of maximum-tolerated doses (MID) used in
`xenograft studies in mice for 19 anticancer agents.
`
`Drugs
`
`5-Azacytidine
`ACNU
`Actinomycin D
`Adriamycin
`BCNU
`Bleomycin
`CCNU
`Cisplatin
`Cyclophosphamide
`DTIC
`5-Fluorouracil
`MCCNU
`Melphalan
`Methotrexate
`Mitomycin C
`Procarbazine
`Vinblastine
`Vincristine
`Vindesine
`
`Approximate LDio dose level
`(mg/kg; single dose i.p.)
`
`100
`4(}..48
`0.5
`8-12
`21- 30
`300
`40-45
`7.5-10
`200-280
`200-300
`80-200
`30-35
`12-15
`100-250
`5-<>.7
`325-1300
`2- 2.5
`1.6-2.0
`3
`
`7 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`270
`
`maximum-tolerated doses (MTD). The dose levels
`have been selected on the basis of toxicity studies
`(LD 10 level). Freireich et al. [32] have shown that
`the LDIO doses of various drugs in mice correlated
`well with the MTD in humans on a dose per surface
`area basis.
`Drug toxicity in animals is known to vary with a
`number of factors, including age, sex, genetic back(cid:173)
`ground, husbandry , and status of health. Even
`within the same strain, a considerable variability in
`toxicity can exist from one laboratory to another.
`Nude mice tolerate higher doses of cytotoxic agents
`than conventional mice [59]. This is probably due
`to the fact that the nude mouse has a higher ~ctivity
`of the hepatic microsomal drug metabolizing en(cid:173)
`zyme system [33].
`The range of maximum-tolerated doses for a
`number of anticancer agents used in xenograft
`studies in different laboratories is given in Table 5.
`The actual doses for some drugs varied consid(cid:173)
`erably from one study to another. Many chemo(cid:173)
`therapeutic studies use a treatment schedule in
`which a single MTD-dose is given [U3), whereas
`others administer lower doses repeatedly [11, 34).
`
`Endpoints for the therapeutic response
`
`The evaluation of treatment effects in xenografts
`has relied largely on changes in tumor volume [113-
`116), and more rarely on changes in life span [7, 60,
`106). Some of the criteria used in clinical assess(cid:173)
`ment have also been applied to xenograft testing
`[25). All of these parameters provide a relative
`measure of drug effectiveness, but the importance
`of having essentially the same tumor burden in
`both controls and in the treatment groups cannot
`be overemphasized. The criteria for evaluating
`treatment effects in xenograft testing have not yet
`been standardized. For the purpose of this report
`we have adopted the same basic rules and end(cid:173)
`points as those established for the assessment of
`drug effects on syngenic transplantable tumors
`[102a).
`
`Percent change in tumor size (TIC x 100). When
`the tumor is implanted subcutaneously, tumor vol-
`
`ume can be estimated from tumor diameter mea(cid:173)
`surements. The tumor volume (V) is calculated for
`an ellipsoid by the formula V = (a2 x b)l2, where a
`is the width and b is the length in mm. The tumor
`sizes are standardized in the different groups by
`obtaining relative tumor volume (RV) calculated
`by the formula RV= V/V0, where v. is the mean
`tumor volume at day x and V0 is the mean initial
`tumor volume at the start of treatment (day 0). The
`TIC % ratio (mean tumor volume of the treated
`tumors/mean volume of control group x 100) is
`calculated each time that the tumors are measured,
`e.g., normally daily or two or three times weekly.
`The lowest value is expressed as the optimal TIC %
`for each group. Many laboratories define a positive
`tumor response to therapy, as suggested by Geran
`et al. [ 40], as any treatment group in which the
`tumor volume is reduced to s42% relative to the
`untreated control group (TIC); i.e. , 58% inhibition
`of growth on any day after treatment (see also Fig.
`2). Other laboratories classify the treatment effect
`as remission (product of the two diameters less
`than 50% of initial value), minimal regression
`(51%-75%), no change (76%-U4%), or progres(cid:173)
`sion (~125% of initial value after 3-4 weeks) [25].
`
`Percent change in life span (TIC x 100). This end(cid:173)
`point is used in the screening of new chemother(cid:173)
`apeutic agents, particularly when tumor measure(cid:173)
`ments are not feasible; for example, intraperito(cid:173)
`neally or intracerebrally. The treatment can be
`given either once or over several days, and the time
`of death of animals in each treatment group is
`recorded daily. The mean time of death of treated
`animals is compared with that of untreated ani(cid:173)
`mals, and the percent increase in lifespan (ILS) is
`calculated from the day of tumor inoculation by the
`formula TIC x 100. An ILS of ;:;:25% is considered
`indicative of activity (40]. However, ethical prob(cid:173)
`lems arise when using this endpoint in the treat(cid:173)
`ment of subcutaneously transplanted solid tumors,
`because mice with large tumors can suffer need(cid:173)
`lessly before dying.
`
`Tumor growth delay (T- C). The tumor growth
`delay is the displacement in time (days) between
`the growth curve of the control group and the
`
`8 of 22
`
`Celltrion, Inc., Exhibit 1028
`
`

`

`271
`
`•
`
`•
`
`•
`•
`• •
`
`•
`• •
`~
`- --- -------- ------- - ---, .
`~ •
`I • • •
`•
`:
`,
`•
`
`..
`
`:
`
`.
`.. ~
`•
`•
`• • • •
`• • • •
`•
`•
`
`10---
`
`- -
`
`-
`
`>
`... 2 -
`~
`'O
`~
`~
`0>1-
`
`01~~~~,~~ ........ ,~~_._,,~~~,~~~,~
`0
`20
`40
`60
`80
`100
`
`lnh1b1t1on of tumor growth ( % )
`
`Fig. 2. Correlation between inhibition of tumor growth and
`specific growth delay.
`
`to chemotherapy and compare these data with the
`clinical response data.
`
`Direct comparison studies between xenograft and
`donor patient
`
`The direct comparison of drug response between a
`xenograft and a donor patient is difficult because of
`the relatively low take rates, the Jong delay be(cid:173)
`tween establishing and subsequently testing xeno(cid:173)
`grafts, and the small proportion of donor patients
`who receive only chemotherapy [5). Another fac(cid:173)
`tor that makes such a comparison difficult is that
`usually in drug testing only one drug is evaluated at
`a time, whereas these same drugs are rarely used as
`single agents in the cJinic. Jn spite of these limita(cid:173)
`tions, comparisons have been possible in several
`instances, and the agreement between patient and
`xenograft responses is good. A survey of the pub(cid:173)
`lished comparisons of xenograft with donor patient
`response is given in Table 6. There appears to be a
`good correlation between the xenograft and the
`
`growth curve of the tumors recurring after treat(cid:173)
`ment. The volume is determined from caliper mea(cid:173)
`surements, and the time taken to double its pre(cid:173)
`treatment volume (Td) is determined. 'Actual' tu(cid:173)
`mor growth delay is calculated as Td 1rcarcd-Td conrro1
`and 'specific' growth delay as Td rr<>•ed-Td conrroffd
`control· The 'specific' growth delay may be regarded
`as an estimate of the number of volume doubling
`times saved by the treatment . It provides a basis for
`comparison of the therapeutic response between
`tumors of different growth rates [89]. In our in(cid:173)
`vestigations. we consider a 'specific' growth delay
`~2 as indicative of activity. Fi