0021-972X/05/$15.00/0
`Printed in U.S.A.
`
`The Journal of Clinical Endocrinology & Metabolism 90(2):1114 –1122
`Copyright © 2005 by The Endocrine Society
`doi: 10.1210/jc.2004-1572
`
`Single Injections of Vascular Endothelial Growth
`Factor Trap Block Ovulation in the Macaque and
`Produce a Prolonged, Dose-Related Suppression of
`Ovarian Function
`
`Hamish M. Fraser, Helen Wilson, John S. Rudge, and Stanley J. Wiegand
`Medical Research Council Human Reproductive Sciences Unit (H.M.F., H.W.), Centre for Reproductive Biology, Edinburgh
`EH16 4SB, United Kingdom; and Regeneron Pharmaceuticals (J.S.R., S.J.W.), Tarrytown, New York 10591
`
`Follicular development is associated with intense angiogen-
`esis and increased permeability of blood vessels under the
`control of locally produced angiogenic factors such as vascu-
`lar endothelial growth factor (VEGF). The aim of the present
`study was to evaluate the effects of transient inhibition of
`VEGF on pituitary-ovarian function in the macaque. Animals
`were given a single, iv injection of a potent, receptor-based
`VEGF antagonist, the VEGF Trap. VEGF Trap was given at a
`dose of 4, 1, or 0.25 mg/kg in the midfollicular phase or at 1.0
`mg/kg in the late follicular phase. Controls were treated with
`vehicle or a control protein, recombinant human Fc (1 mg/kg).
`Blood samples were collected once daily for 12 d after injec-
`tion, and three times per week thereafter until normal ovu-
`latory cycles had resumed. The VEGF Trap produced a rapid
`
`suppression of estradiol and inhibin B concentrations at all
`doses tested, followed by a marked and sustained increase in
`LH and FSH. Ovulation and formation of a functional corpus
`luteum, as evidenced by increased serum progesterone levels,
`failed to occur at the anticipated time. Normal ovarian activ-
`ity resumed when plasma concentrations of unbound VEGF
`Trap fell below about 1 mg/liter. When treatment was initiated
`in the midfollicular phase, control macaques ovulated 7.2 ⴞ
`0.4 d later, but ovulation was delayed in a dose-dependent
`manner by VEGF Trap, occurring 23 ⴞ 0.7, 30 ⴞ 1.4, and 43 ⴞ
`0.8 d after injection of 0.25, 1, or 4 mg/kg, respectively. Thus,
`the VEGF Trap exerts a potent, dose-dependent, but revers-
`ible inhibitory effect on ovarian function. (J Clin Endocrinol
`Metab 90: 1114 –1122, 2005)
`
`IT IS NOW generally accepted that follicular angiogenesis
`
`and vascular permeability are closely regulated by a
`complex interplay of stimulatory and inhibitory vasoactive
`factors produced by the theca and granulosa (1). Given the
`close association between structural and functional vascular
`remodeling and follicular maturation, it has long been pos-
`tulated that manipulation of follicular angiogenesis could
`affect ovarian function (2– 4). However, initial pharmacolog-
`ical studies that attempted to establish the link between ovar-
`ian angiogenesis and function met with mixed success be-
`cause they of necessity employed relatively nonspecific
`compounds whose mechanism of action, and potency, were
`largely unknown (5–7). Over the last few years, the devel-
`opment of improved reagents targeted to defined angiogenic
`factors or their receptors has enabled studies that have un-
`equivocally confirmed the importance of angiogenesis in
`ovarian function, and particularly implicated vascular en-
`dothelial growth factor (VEGF)-A and its receptors as key
`mediators of this process.
`In these experiments, the VEGF pathway has been inhib-
`ited using antibodies directed against VEGF itself (8, 9),
`antibodies to the VEGF receptor VEGFR-2/Flk (10, 11), a
`VEGF receptor tyrosine kinase inhibitor (12), or by decoy
`
`First Published Online November 23, 2004
`Abbreviations: OHSS, Ovarian hyperstimulation syndrome; PCOS,
`polycystic ovarian syndrome; VEGF, vascular endothelial growth factor.
`JCEM is published monthly by The Endocrine Society (http://www.
`endo-society.org), the foremost professional society serving the en-
`docrine community.
`
`VEGF receptors (13, 14). We employed highly potent receptor-
`based VEGF antagonists. Initial studies in marmosets used a
`prototypical receptor-based antagonist, VEGF TrapA40, which
`comprised the immunoglobulin domains 1–3 of VEGFR-1 fused
`to the Fc portion of human IgG. Surprisingly, acute systemic
`administration during the early luteal phase inhibited not only
`luteal angiogenesis (15) but also follicular angiogenesis (16). In
`subsequent studies a successor molecule, VEGF TrapR1R2, was
`employed to evaluate the effects of inhibition of VEGF through-
`out the follicular phase. These studies confirmed that selective
`inhibition of VEGF markedly attenuated thecal angiogenesis
`and restricted follicular growth in the marmoset (17).
`The small size of the marmoset restricts the number of
`blood samples that may be obtained to monitor the concom-
`itant effects of VEGF inhibition on pituitary and ovarian
`hormones. In contrast, the stump-tailed macaque is an old
`world primate with a body weight of 12–15 kg and menstrual
`cycles similar to the human female, whose hormone profiles
`can be determined by well-established assays from blood
`samples collected at close intervals. We used this species
`previously to evaluate the effects of GnRH analogs (18) be-
`fore initiating clinical investigations in women (e.g. Ref. 19).
`The first objective of the current study was to assess the acute
`and longer-term effects of a single, iv injection of the VEGF
`TrapR1R2 at the midfollicular phase in the macaque. This
`phase was selected for detailed study because the follicle that
`will eventually ovulate is being selected at this time. Our
`earlier studies in the marmoset suggested that VEGF-
`mediated angiogenesis is of crucial importance in the selection
`
`1114
`
`Celltrion Exhibit 1033
`Page 1
`
`

`

`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122 1115
`
`and growth of the dominant follicle and that if VEGF action
`were effectively abrogated, the follicular cycle could not con-
`tinue and a new phase of follicular recruitment would have
`to be initiated. A second objective was to determine the
`minimal dose of VEGF TrapR1R2 that would be required to
`interrupt follicular development and whether the duration of
`the subsequent suppression of ovarian function would also
`be dose related. A third objective was to assess the effects of
`acute VEGF inhibition during the late follicular phase when
`the thecal vasculature of the ovulatory follicle is already fully
`developed.
`
`Materials and Methods
`
`Animals
`
`Thirteen adult female stump-tailed macaques aged 7–22 yr and
`weighing 10 –16 kg used in the study were captive bred in the United
`Kingdom or Holland and housed in a unit opened in 1996 and designed
`with an emphasis on environmental enrichment (20). The animals
`moved freely from their living rooms via a tunnel to cages for their water
`supply and sleeping area and collection of blood samples.
`Vaginal swabs were taken with a cotton-tipped applicator each morn-
`ing and the pattern of menstrual bleeding recorded. The animals were
`trained to enable blood sample collection by femoral venipuncture with-
`out anesthesia and with minimal or no restraint. All the animals in the
`study had regular ovulatory menstrual cycles as determined from men-
`strual pattern and serum concentrations of estradiol-17␤ and proges-
`terone in blood samples obtained three times per week before treatment.
`The study was approved by the local Primate Ethical Committee and
`carried out under a project license granted by the United Kingdom
`Home Office.
`
`Treatments
`
`Endogenous VEGF was inhibited by administration of VEGF
`TrapR1R2, a recombinant, chimeric protein comprising Ig domain 2 of
`human VEGF-R1 and Ig domain 3 of human VEGF-R2, expressed in
`sequence with the human Fc. Compared with earlier versions of recep-
`tor-based fusion proteins, the VEGF TrapR1R2 exhibits greater affinity for
`VEGF-A (affinity constant ⬃1 pm) as well as improved bioavailability
`and pharmacokinetic properties (21). VEGF TrapR1R2 (Regeneron Phar-
`maceuticals, Inc., Tarrytown, NY) was provided at a concentration of
`24.3 mg/ml in 2-ml aliquots in buffer composed of 5 mm phosphate, 5
`mm citrate, 100 mm NaCl (pH 6.0), and 0.1% wt/vol Tween 20, with
`either 20% glycerol or 20% sucrose. Human Fc, for control treatments,
`was provided at a concentration of 19.7 mg/ml in buffer composed of
`40 mm phosphate and 20 mm NaCl (pH 7.4). VEGF Trap vehicle alone
`also was employed during some control cycles. The compounds were
`stored at ⫺20 C until required, at which time they were thawed. Any
`compound remaining was stored at 4 C and used within 2 wk.
`In a pilot study in two macaques, a single iv injection of 12.5 mg/kg,
`VEGF TrapR1R2 was found to effectively inhibit ovarian function for
`more than 40 d. Therefore, in the main study, we elected to investigate
`the response to 4.0, 1.0, and 0.25 mg/kg (n ⫽ 4 per group) administered
`as a single dose (iv) during the midfollicular phase, d 6 – 8 of the cycle.
`Late-follicular-phase administration was studied at the intermediate
`dose of 1 mg/kg (n ⫽ 4). In comparable control cycles, macaques were
`treated with either 1 mg/kg human Fc (iv) during the mid- (n ⫽ 3) or
`late follicular phase (n ⫽ 2) or vehicle administered during the mid- (n ⫽
`3) or late follicular phase (n ⫽ 3).
`After treatment with VEGF TrapR1R2, vehicle, or Fc, (d 0), blood
`samples were collected at 0 and 15 min and 4, 6, and 8 h and then daily
`for the next 12–14 d. Thereafter, blood samples were collected three times
`per week until normal ovulatory cycles were reestablished, as evidenced
`by elevation of progesterone levels consistent with luteal values mea-
`sured in pretreatment cycles for that macaque. Because of limits in
`numbers of animals available, five animals received two treatments with
`VEGF TrapR1R2. In addition, one animal had been treated in the pilot
`study. In this case, antibodies to the VEGF TrapR1R2 were detected
`
`during the second treatment cycle, and the affected cycle was excluded
`from further analysis (see Results).
`Because no differences were noted in the effects of vehicle and Fc
`administration (neither treatment produced an appreciable affect on
`ovarian or pituitary hormones), control cycles from both control treat-
`ment conditions were combined for statistical analyses.
`
`Assays
`Estradiol-17␤and progesterone were measured by RIAs as described
`previously (22), detection limits being 30 pm and 0.7 nm, respectively.
`LH and FSH were measured by RIAs based on recombinant cynomolgus
`monkey LH and FSH supplied by the National Hormone and Pituitary
`Program (Dr. A. F. Parlow, National Institute of Diabetes and Digestive
`and Kidney Diseases). FSH was measured using anticynomolgus FSH
`with a detection limit of 2 ␮g/liter (National Institute of Child Health
`and Human Development, rec-mo-FSH-RP-1, AFP-6940A), and inter-
`assay coefficient of variance of 12%. LH was measured using a rabbit
`antiserum to cynomolgus LH (AFP342994) used at a final dilution of
`1:750,000. Rec-mo LH-RP-1 (AFP-6936H) was used for radioiodination
`as instructed and results expressed as micrograms per liter of the same
`preparation. Assay sensitivity was 0.3 ␮g/liter and interassay coefficient
`of variance 11%. Inhibin B was measured throughout the study period
`in animals in the VEGF Trap treatment groups only. The assay was as
`described previously (20) and had a detection limit of 10 ng/liter.
`VEGF TrapR1R2 was measured by an ELISA, using human VEGF 165
`to capture and an antibody to the human Fc region as the reporter (21).
`Capture of the VEGF TrapR1R2 by VEGF coated on the microplate re-
`quires a vacant VEGF binding site. Consequently, this ELISA specifically
`detects only VEGF Trap that is not already bound to endogenous VEGF.
`Serum samples were diluted in assay buffer and run against standards
`also prepared in assay buffer. Each dilution level was assayed, and those
`that read on the linear part of the standard curve, in which the samples
`ran parallel to that of the reference standard, were selected for analysis.
`If in the initial assay, values were below the limit of detection, samples
`were reassayed neat and the standards spiked with an equivalent vol-
`ume of mouse serum. Assay sensitivity was 0.14 ␮g/liter, and interassay
`variation based on low-, medium-, and high-quality controls were less
`than 10%. Concentration vs. time curves were constructed from ELISA-
`generated VEGF Trap values obtained from individual animals. The
`pharmacokinetic parameter estimates were determined by fitting the
`serum concentration vs. time profile to a noncompartmental model
`(WinNonLin, version 2.0, Pharsight Corp., Mountain View, CA).
`
`Data analysis and statistics
`
`The day of ovulation was defined as the day of the LH peak. Peak
`levels of estradiol were typically noted on the previous day or in some
`cycles on the same day as the LH peak. In normal cycles, the LH peak
`was followed within 1 d by arise in progesterone levels, which were
`sustained for 2 wk. In pre- and posttreatment cycles in which blood
`samples were obtained three times per week, gonadotropin and ovarian
`steroid levels were evaluated to provide a best estimate of the day of
`ovulation.
`Data for ovarian and pituitary hormones as well as VEGF TrapR1R2
`concentrations for each individual animal were plotted with reference
`to the day of treatment (d 0). Data around d 0 were also plotted as
`mean ⫾ sem values for each treatment group: beyond the point at which
`daily blood samples were available (after 12–14 d), mean values were
`obtained by averaging samples taken from all animals at the nearest
`equivalent times (e.g. in a group of four, collections of n ⫽ 1 at d 20, n⫽
`2 at d 21, and n ⫽ 1 at d 22 were averaged and plotted as n ⫽ 4 at d 21).
`Data for time to ovulation after treatment and effects of treatment on
`hormone concentrations were subjected to ANOVA using the Prism
`program 4 for Macintosh (GraphPad Prism, San Diego, CA) followed by
`Bonferroni’s multiple comparison tests. To determine effects of treat-
`ment within an animal, the mean of the three pretreatment values for
`each hormone was used. The posttreatment period subjected to statis-
`tical analysis for each group was based on the observed duration of
`response to treatment defined as the average period of suppression of
`estradiol below pretreatment value. The final day of response was the
`one that preceded three consecutive values above pretreatment value.
`
`Celltrion Exhibit 1033
`Page 2
`
`

`

`1116 J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122
`
`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`According to this definition, average duration of response was 30, 17,
`and 10 d for the 4, 1, and 0.25 mg/kg doses, respectively. Similar
`estimates of the duration of response were obtained using suppression
`of inhibin B as an end point. Area under the curve for progesterone and
`estradiol peaks was compared for pre- and posttreatment cycles. Dif-
`ferences were considered significant at a level of P ⬍ 0.05.
`
`Results
`Midfollicular phase treatment
`A representative example of the hormone profile for ve-
`hicle-treated or Fc-treated control cycles is shown in Fig. 1.
`After control injections, plasma estradiol levels continued to
`rise, increasing sharply 4 – 6 d later, before ovulation. The
`ovulatory surge in LH/FSH took place 7.2 ⫾ 0.4 d (mean ⫾
`
`sem) after treatment and was followed by a sustained ele-
`vation in plasma progesterone, which reached a plateau 8 –12
`d post ovulation before falling to follicular phase values
`around luteal d 14 –16. Of the six midfollicular control cycles
`studied, ovulation was not observed at the anticipated time
`in one vehicle-treated animal, and this cycle was excluded
`from further evaluation. This animal had an extended fol-
`licular phase of 20 d, but this was distinguished from the
`response to VEGF Trap treatment in that the delay in ovu-
`lation was not accompanied by a suppression of estradiol
`levels or a prolonged rise in LH and FSH (see below).
`The VEGF TrapR1R2 was well tolerated at all doses tested.
`Typical hormonal responses observed after administration of
`
`FIG. 1. Serum concentrations of estradiol (open circles), progesterone (P4, closed circles) (left panel), FSH (closed squares), and LH (open squares)
`(right panel) in individual macaques injected in the midfollicular phase with either Fc (1 mg/kg, iv) (control) or 0.25, 1, or 4 mg/kg VEGF TrapR1R2
`iv (arrow). Note the suppression of estradiol, failure of ovulatory progesterone rises, and increases in LH and FSH after treatment, followed
`by dose-related recovery of normal pituitary-ovarian function.
`
`Celltrion Exhibit 1033
`Page 3
`
`

`

`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122 1117
`
`0.25, 1.0, and 4.0 mg/kg of the VEGF TrapR1R2 are shown in
`Fig. 1. Pituitary-ovarian function was clearly altered in every
`case at each dose tested, and the observed changes in hor-
`monal profiles followed a consistent and distinctive pattern.
`Treatment with VEGF TrapR1R2 not only inhibited the late
`follicular phase rise in plasma estradiol but also resulted in
`a rapid decline in estradiol concentrations to early follicular
`phase levels. Within 2–3 d of injection, a progressive increase
`in plasma LH and FSH was observed, and the FSH to LH
`ratio was increased in all animals relative to those normally
`observed during the preovulatory surge of LH and FSH
`(Fig. 1).
`Comparison of mean data for control and treated cycles is
`shown in Fig. 2. In the posttreatment period, serum estradiol
`levels in VEGF Trap-treated cycles were significantly lower
`than in control cycles for all doses tested (P ⬍ 0.0001). A
`sustained reduction of estradiol levels below pretreatment
`follicular values was observed in the 4 mg/kg group (P ⬍
`0.0001), in which estradiol levels remained significantly be-
`low normal follicular levels between d 6 and 28 post treat-
`ment. Conversely, VEGF Trap treatment resulted in a sig-
`
`nificant stimulatory effect (P ⬍ 0.01) on LH concentrations at
`all three dose levels. Serum FSH levels also were significantly
`higher (P ⬍ 0.0001) than control cycle values in all three
`treatment groups. FSH appeared to rise and reach a plateau
`slightly more rapidly than LH. The rate of the rise in serum
`gonadotropins appeared unaffected by dose of VEGF Trap,
`although peak values seen between 15–20 in the higher dose
`groups may not have been obtained in the 0.25 mg group as
`a result of the more rapid recovery.
`Evaluation of percentage change from pretreatment values
`during the first 48 h after injection showed that inhibin B
`concentrations exhibited a more rapid and marked decline
`than estradiol, being significantly suppressed as early as 8 h
`(0.25 and 1 mg/kg doses), whereas estradiol was not signif-
`icantly different from pretreatment value until 24 or 48 h
`(Fig. 3).
`
`Pharmacokinetics and pharmacodynamics
`Mean peak concentrations of free VEGF TrapR1R2 mea-
`sured in the first postinjection blood sample at 15 min were
`
`FIG. 2. Serum concentrations of estradiol,
`progesterone, LH, and FSH after treat-
`ment with vehicle/Fc or 0.25, 1, or 4 mg/kg
`VEGF TrapR1R2 in the midfollicular phase.
`Values are means ⫾ SEM plotted with
`reference to the day of injection (d 0). Note
`the timely rise in estradiol, followed by the
`preovulatory LH/FSH surge and sus-
`tained rise in luteal progesterone levels in
`controls. VEGF TrapR1R2 treatment re-
`sults in the failure of the characteristic
`increases in estradiol and progesterone
`and the induction of a persistent increase
`in LH and FSH. Note the dose-related du-
`ration of response followed by recovery ini-
`tiated by the rise in estradiol.
`
`Celltrion Exhibit 1033
`Page 4
`
`

`

`1118 J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122
`
`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`this atypical case was assayed for antibodies to the VEGF
`TrapR1R2 and found to be positive, 6478 mIU/ml. This ma-
`caque had been injected with 12.5 mg/kg VEGF Trap R1R2 in
`the pilot study 16 months previously.
`
`Characteristics of posttreatment recovery of ovarian cycles
`Evaluation of the hormonal profiles on recovery of ovarian
`function suggested that posttreatment menstrual cycles were
`characteristically normal in terms of length as well as the
`pattern and magnitude of pituitary gonadotrophin and ovar-
`ian steroid levels. This impression was confirmed by quan-
`titative analyses, which showed that peak preovulatory es-
`tradiol levels and area under the curve for progesterone
`during subsequent luteal phase were not statistically differ-
`ent in the immediate pre- and posttreatment cycles (Table 1).
`
`Vaginal bleeding
`In control cycles, bleeding was detected during the first
`week post treatment in one of six cases. Nine of the 12
`macaques treated with VEGF TrapR1R2 during the midfol-
`licular phase exhibited bleeding during this period, likely
`reflecting the abrupt and sustained reduction in estradiol
`levels.
`
`Late follicular phase treatment
`Where control injections were given during the late fol-
`licular phase, serum estradiol levels continued to rise and a
`distinct LH/FSH surge occurred 1–5 d later, followed by a
`sustained rise in serum progesterone. In marked contrast, the
`anticipated preovulatory rise in estradiol and ovulatory pro-
`gesterone were blocked in all macaques treated with VEGF
`Trap (1 mg/kg); rather there was a rapid and sustained
`decrease in plasma estradiol levels (Figs. 6 and 7), which
`persisted for an average of 19 d in three of the four treated
`macaques. Treatment was followed within 1 d by amarked
`increase in LH and FSH secretion (Figs. 6 and 7). In contrast
`to the midcycle gonadotrophin surge seen in normal and
`control cycles, the LH/FSH increase produced by adminis-
`tration of VEGF TrapR1R2 was characterized by a marked
`increase in the FSH to LH ratio. In the remaining animal, the
`suppression in estradiol was maintained for only 8 d, and the
`duration of the LH/FSH rise was similarly abbreviated.
`The peak of unbound VEGF TrapR1R2 (39 mg/liter) in the
`serum, as well as other pharmacokinetic parameters (not
`shown), closely resembled those observed after midfollicular
`phase injection of 1 mg/kg VEGF Trap. As was the case for
`midfollicular phase injections, recovery of ovarian function,
`as evidenced by an increase in estradiol levels followed by
`normalization of gonadotrophin levels, occurred only after
`circulating levels of unbound VEGF TrapR1R2 fell below
`about 1 mg/liter. The more rapid recovery of estradiol se-
`cretion observed in one animal was not accompanied by
`accelerated clearance of free VEGF TrapR1R2.
`In all four cases, the luteal phase progesterone rise failed
`to occur at the anticipated time, and progesterone remained
`at follicular phase levels until after the first posttreatment
`ovulation, which occurred 32 ⫾ 0.9 d after treatment, a sig-
`nificant delay, compared with control cycles (P ⬍ 0.001) in
`
`FIG. 3. Percentage change from pretreatment values in inhibin B and
`estradiol after treatment with vehicle/Fc or 0.25, 1, or 4 mg/kg VEGF
`TrapR1R2 in the midfollicular phase. Note the more rapid decline in
`inhibin B, which was significantly lower by 8 h, whereas estradiol was
`not significantly suppressed until 24 or 48 h.
`
`10.8, 34, and 140 mg/liter for the 0.25, 1, and 4 mg/kg groups,
`respectively. The clearance at the different doses ranged from
`5.8 to 8.7 ml/d/kg, and the steady-state volume of distri-
`bution was approximately 34.8 ml/kg.
`The differential duration of suppression in ovarian func-
`tion observed at each dose of VEGF TrapR1R2 was temporally
`correlated with clearance of unbound VEGF Trap from the
`circulation. The mean estradiol and inhibin B values for each
`of the three dose groups are plotted in relation to levels of
`VEGF Trap in Fig. 4. At each dose, inhibin B was suppressed
`to the detection limit of the assay, whereas estradiol was main-
`tained around early follicular levels until plasma concentrations
`of VEGF TrapR1R2 fell below approximately 1 mg/liter.
`The duration of suppression in ovarian function produced
`by administration of the VEGF Trap was clearly dose related.
`The time to the first posttreatment ovulation was signifi-
`cantly longer (P ⬍ 0.001) in the 4 mg/kg group than either
`the 1 or 0.25 mg/kg groups, and the 1 mg/kg dose also
`resulted in a significantly longer (P ⬍ 0.002) period of sup-
`pression than 0.25 mg/kg (Fig. 5).
`A single exception to the above pattern was observed in
`one macaque that received a 4 mg/kg injection. The initial
`10-d posttreatment period was associated with hormonal
`profiles indistinguishable from that of others in this dose
`group. However, estradiol levels began to increase rapidly
`around d 11. This uncharacteristically rapid escape from the
`effect of VEGF inhibition was associated with an early and
`abrupt fall in serum-free Trap levels to 1.5 mg/liter at d 10,
`compared with around 20 mg/liter at this time point in the
`other animals in this group. The sample taken at d 10 from
`
`Celltrion Exhibit 1033
`Page 5
`
`

`

`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122 1119
`
`VEGF
`of
`FIG. 4. Serum profile
`TrapR1R2 in relation to inhibin B and
`estradiol after treatment with 0.25, 1,
`or 4 mg/kg given as a single, iv injection
`during the midfollicular phase. Note
`that in each group inhibin B and estra-
`diol concentrations begin to recover af-
`ter VEGF Trap values fall less than 1
`mg/liter. Values are means ⫾ SEM.
`
`which ovulation occurred 3.4 ⫾ 0.7 d after injection of vehicle
`or hFc (Fig. 7).
`Comparison of mean hormonal data for late follicular
`phase control and treated cycles is shown in Fig. 7. Serum
`estradiol levels were significantly reduced, compared with
`pretreatment values (P ⬍ 0.0001) after administration of
`VEGF TrapR1R2. Progesterone concentrations also were sig-
`nificantly suppressed (P ⬍ 0.0001) in the treated cycles rel-
`ative to controls beyond d 6 (i.e. coincident with the luteal
`phase rise in progesterone in vehicle and Fc control cycles).
`Similar to the observations after treatment in the midfol-
`licular phase, three of the four treated animals exhibited
`vaginal bleeding during the first week, compared with one
`of five controls.
`
`Discussion
`VEGF TrapR1R2, administered at the mid- or late follicular
`phase, caused a rapid inhibition of estradiol and inhibin B
`
`FIG. 5. Time to posttreatment ovulation in macaques treated with
`0.25, 1, or 4 mg/kg VEGF TrapR1R2 at the midfollicular phase or after
`1 mg/kg at the late follicular phase.
`
`secretion and blockade of ovulation in all 16 menstrual cycles
`evaluated, irrespective of dose. This demonstrates that VEGF
`is essential for the maintenance of normal ovarian function
`during the follicular phase in the macaque. The results were
`particularly remarkable with respect to the duration of the
`response observed after single injections of VEGF TrapR1R2.
`The interval between treatment and recovery of ovarian
`function was clearly dose dependent, and the timing of func-
`tional recovery was very consistent within each dose group.
`Zimmermann et al. (9, 10) have shown that administration of
`antibodies to VEGF-R2 or VEGF beginning in the early or late
`follicular phase, respectively, also delay follicular develop-
`ment and ovulation in the rhesus monkey. However, in these
`studies, repeated antibody injections were required to
`achieve this effect, and the duration of suppression in ovar-
`ian function obtained was of shorter duration and more
`variable than observed after a single injection of the lowest
`dose of VEGF TrapR1R2 studied, 0.25 mg/kg.
`The present findings in macaques confirm and extend our
`previous observations on effects of VEGF inhibition on ovar-
`ian structure and function in the marmoset, which focused
`on molecular and cellular changes (16, 17). In marmosets,
`VEGF inhibition throughout the follicular phase severely
`inhibits thecal angiogenesis and inhibits the growth of sec-
`ondary follicles beyond the early antral stage (17). The in-
`hibition of the normal, progressive increase in estradiol levels
`and blockade of ovulatory progesterone rises observed in the
`present study after single injections of the VEGF Trap in
`macaques are consistent with the anatomical changes ob-
`served in the marmoset ovary after Trap treatment. Some-
`what surprisingly, macaques treated during the late follic-
`ular phase also exhibited a marked and persistent decline
`
`Celltrion Exhibit 1033
`Page 6
`
`

`

`1120 J Clin Endocrinol Metab, February 2005, 90(2):1114 –1122
`
`Fraser et al. (cid:127) VEGF Trap and Follicular Function
`
`TABLE 1. Peak preovulatory concentrations of estradiol and area under the curve for luteal progesterone in pre- and post treatment
`recovery cycles associated with treatment starting at the midfollicular (MF) or late follicular (LF) phase
`
`Stage
`Peak estradiol (pmol/liter)
`MF
`
`LF
`Luteal progesterone (nmol/liter)
`(area under the curve)
`MF
`
`LF
`Values are means ⫾ SEM. NS, Not significant.
`
`Dose (mg/kg)
`
`Pre treatment
`
`Post treatment
`
`Significance
`
`0.25
`1
`4
`1
`
`0.25
`1
`4
`1
`
`1034 ⫾ 179
`1545 ⫾ 180
`1441 ⫾ 425
`1264 ⫾ 527
`
`181 ⫾ 47
`235 ⫾ 46
`230 ⫾ 54
`199 ⫾ 12
`
`1629 ⫾ 302
`956 ⫾ 259
`1460 ⫾ 552
`739 ⫾ 201
`
`184 ⫾ 50
`227 ⫾ 50
`191 ⫾ 36
`199 ⫾ 27
`
`NS
`NS
`NS
`NS
`
`NS
`NS
`NS
`NS
`
`in estradiol levels. Because the dominant follicle has been
`selected by this time, this finding indicates that it too is
`susceptible to VEGF inhibition. Because much of the thecal
`vasculature of the preovulatory follicle has already been
`elaborated (23, 24), diminished vascular permeability may
`play a particularly significant role in the apparent ovulatory
`failure observed after VEGF inhibition in the late follicular
`phase. Taken together with the marmoset data, the above
`observations indicate that the inhibition of follicular matu-
`ration produced by administration of VEGF TrapR1R2 is sec-
`ondary to attenuation of the follicular vascular density
`and/or permeability, which in turn reduces the availability
`of growth factors, hormones, lipoproteins for steroid pro-
`duction, nutrients, and oxygen to the growing follicles.
`The rapid and persistent rise in LH and FSH levels con-
`sistently observed after administration of the VEGF TrapR1R2
`is likely a direct consequence of the marked and abrupt
`attenuation of serum estradiol and inhibin B concentrations
`produced by VEGF inhibition. A similar scenario was de-
`scribed after administration of anti-VEGFR-2 in the rhesus
`monkey starting at the early follicular phase (10). This is the
`expected response to withdrawal of the negative feedback
`effect of the ovarian steroids and inhibin B on pituitary LH/
`FSH secretion. In the present study, the rise in gonadotrophin
`levels was most prompt when VEGF inhibition was initiated
`in the late follicular phase, perhaps reflecting increased sen-
`
`sitivity to withdrawal of negative feedback at this time. Ir-
`respective of the timing of VEGF Trap administration, the
`magnitude of the FSH increase was relatively greater than
`that observed for LH and was similar in magnitude to that
`observed in the same species after specific inhibition of es-
`tradiol (20). However, the tonic elevation in pituitary gona-
`dotrophin levels observed after administration of VEGF
`TrapR1R2 was ineffective in promoting follicular develop-
`ment in the face of ongoing VEGF inhibition. Rather, resti-
`tution of follicular maturation, as evidenced by increasing
`serum estradiol and inhibin B levels, was evident only when
`circulating levels of free VEGF TrapR1R2 fell less than 1 mg/
`liter. Similarly, administration of exogenous gonadotropins
`failed to stimulate ovarian follicular angiogenesis and
`growth in mice when endogenous VEGF signaling was in-
`hibited by administration of antibodies against VEGFR-2
`(11). These results, together with similar observations in the
`rhesus macaque, have led to the proposal that the access of
`FSH to the ovary is impeded when VEGF is inhibited (10).
`The rapid nature of the decline in inhibin B after inhibition
`of VEGF observed in the current study and VEGFR-2 in the
`early follicular phase (10) also suggests that the secretion of
`proteins from the developing follicle may also be impaired.
`The more gradual decline in serum estradiol could indicate
`that steroid secretion is less affected by the most immediate
`
`FIG. 6. Serum concentrations of estra-
`diol (open circles), progesterone (P4,
`closed circles) (left panel), FSH (closed
`squares) and LH (open squares) (right
`panel) in macaques injected in the late
`follicular phase with either Fc (1 mg/kg,
`iv) (control) or 1 mg/kg VEGF TrapR1R2
`iv (arrow). In the treated animal, note
`the suppression of estradiol, failure of
`ovulatory progesterone rises, and in-
`creases in LH and FSH, followed by
`recovery of normal pituitary-ovarian
`function.
`
`Celltrion Exhibit 1033
`Page 7
`
`

`

`Fr