`Age-related Macular Degeneration: Applying Clinical
`Trial Results to the Treatment of Everyday Patients
`
`DAVID M. BROWN AND CARL D. REGILLO
`
`● PURPOSE: The vision gains reported with monthly intra-
`vitreal ranibizumab in the MARINA and ANCHOR trials
`led to an immediate paradigm shift in the treatment of
`neovascular AMD with retina physicians universally
`switching to the pan-VEGF blocking agents ranibizumab
`and bevacizumab, and patients expecting visual improve-
`ment. As these agents are primarily used on a pro re nata
`(PRN) dosing schedule (because neither patients nor
`physicians want monthly injections), the factors involved
`in making the treatment and retreatment decisions are
`very important in any attempt to maximize vision gain.
`● DESIGN: Analysis of literature, ongoing clinical trials,
`and the clinical assessments that can aid clinicians in
`treatment and retreatment decisions.
`● METHODS: Literature review and perspective.
`● RESULTS: If a monthly injection protocol is not used,
`clinicians should use both functional and anatomic criteria
`to attempt to guide treatment and retreatment decisions.
`Qualitative optical coherence tomography (OCT) appears
`to be the most sensitive and practical assessment tool to
`determine anatomic response to treatment but should be
`used in conjunction with clinical examination.
`● CONCLUSIONS: If monthly intravitreal injections are
`not performed, a combination of clinical examination
`(looking for new hemorrhage) and qualitative OCT (to
`assess response to treatment and early signs of recurrent
`leakage) can be used to guide anti-vascular endothelial
`growth factor (anti-VEGF), treatments with the goal of
`maintaining a “normal” retinal anatomy in an attempt to
`maximize the benefit (visual acuity gains) to risk (num-
`ber of injections required) ratio.
`(Am J Ophthalmol
`2007;144:627– 637. © 2007 by Elsevier Inc. All rights
`reserved.)
`
`T REATMENTS FOR NEOVASCULAR AGE-RELATED MACU-
`
`lar degeneration (AMD) have improved dramatically
`since 1999 when the only treatment clinically avail-
`able was Macular Photocoagulation Study (MPS)-style laser
`photocoagulation.1 For subfoveal choroidal neovasculariza-
`tion (CNV), MPS guided laser photocoagulation leads to an
`
`Accepted for publication Jun 20, 2007.
`From the Vitreoretinal Consultants, Houston, Texas (D.M.B.); and the
`Wills Eye Institute, Philadelphia, Pennsylvania (C.D.R.).
`Inquiries to David M. Brown, Vitreoretinal Consultants, Texas Med-
`ical Center Office, Scurlock Tower, 6560 Fannin, Suite 750, Houston,
`TX 77030; e-mail: dmbmd@houstonretina.com.
`
`immediate, permanent loss of central vision with its aim to
`limit the spread of the uncontrollable disease analogous to
`surgical amputation in general surgery. Photodynamic
`therapy with verteporfin (vPDT) became available in
`2000. Unlike laser photocoagulation, vPDT involves rel-
`atively selective photochemical damage to CNV, with less
`damage to the associated choroid and retina. vPDT de-
`creased the rate of moderate and severe vision loss,
`demonstrated improved vision (a gain of ⱖ15 letters on
`the Early Treatment of Diabetic Retinopathy Study
`(ETDRS) chart in up to 6% of patients with subfoveal
`CNV secondary to AMD) (Table).2,3 However, 30% to
`40% of vPDT-treated patients still lost at least 15 letters
`from the pretreatment baseline visual acuity (VA). Fur-
`thermore, although PDT clinical trials showed consistent
`benefit when treating patients with subfoveal predomi-
`nantly classic CNV, other CNV subtypes appeared to
`benefit only if the lesion was relatively small in size.2,3,7
`Combining vPDT treatment with intravitreal (ITV) injec-
`tion of a corticosteroid was popularized by Spaide and
`associates.8 This combination approach appeared to de-
`crease the number of required vPDT treatments to achieve
`a nonleaking CNV, but has yet to demonstrate definitive,
`clinically significant, added visual benefit compared with
`vPDT alone.
`
`THE SEARCH FOR A MORE
`TARGETED TREATMENT
`
`NEOVASCULAR AMD IS ASSOCIATED WITH INCREASED VAS-
`cular permeability as well as CNV infiltration. This in-
`crease in vascular permeability leads to abnormal fluid
`collection within or below the retina that causes visual
`dysfunction when it involves the center of the macula. For
`the past 50 years, researchers have attempted to determine
`the underlying mechanism of neovascularization and vas-
`cular leakage in the eye. In 1983, Senger and associates
`identified a “tumor vascular permeability factor” (VPF)
`that could induce vascular leakage.9 In 1989, Ferrara and
`Henzel reported the isolation and sequencing of an endo-
`thelial cell mitogen dubbed vascular endothelial growth
`factor (VEGF),10 which was later determined to be the
`same molecule as VPF. VEGF-A (also referred to as simply
`VEGF) is a vascular endothelial cell-specific growth factor
`
`0002-9394/07/$32.00
`doi:10.1016/j.ajo.2007.06.039
`
`© 2007 BY ELSEVIER INC. ALL RIGHTS RESERVED.
`
`627
`
`0001
`
`CELLTRION - EXHIBIT 1043
`
`
`
`TABLE. Summary of Published Results for Several Neovascular Age-Related Macular Degeneration Studies Using 2-m Early
`Treatment of Diabetic Retinopathy Study Refracted Visual Acuities
`
`Mean Change in
`letters of VA
`from Baseline*
`
`Efficacy Endpoints
`
`ⱖ15 Letters Lost
`(% Patients)*
`
`ⱖ15 Letters
`Gained
`(% Patients)*
`
`6 2
`
`6 2
`
`26*
`33*
`4
`36*
`40*
`6
`12
`13
`10
`
`⫺2.2
`⫺3.5
`N/A
`
`5.4*
`6.6*
`⫺14.9
`8.5*
`11.3*
`⫺9.5
`⫺1.6*
`⫺0.2*
`⫺16.3
`
`39*
`54
`30*
`45
`
`8*
`10*
`47
`6*
`4*
`36
`17*
`10*
`51
`
`Study
`
`Study Design
`
`TAP (12-month time point)2
`
`VISION (12-month time point)4
`
`MARINA (24-month time point)5
`
`ANCHOR (12-month time point)6
`
`PIER (12-month time point)
`(Regillo et al, unpublished
`data)
`
`PDT every 3 months for
`24 months (n ⫽ 609)
`ITV pegaptanib every 6
`weeks for 24 months
`(n ⫽ 1,186)
`ITV ranibizumab monthly
`for 24 months
`(n ⫽ 716)
`ITV ranibizumab monthly
`for 24 months
`(n ⫽ 423)
`ITV ranibizumab monthly
`for 3 months, then
`every 3 months for a
`total of 24 months
`(n ⫽ 184)
`
`Study
`Arms
`
`PDT
`Placebo
`0.3 mg
`Sham
`
`0.3 mg
`0.5 mg
`Sham
`0.3 mg
`0.5 mg
`PDT
`0.3 mg
`0.5 mg
`Sham
`
`ITV ⫽ intravitreal; N/A ⫽ not applicable (not directly reported in the article); PDT ⫽ photodynamic therapy; VA ⫽ visual acuity.
`*P ⱕ .001 compared with the control arm.
`
`and, within the VEGF family, is the most potent promoter
`of angiogenesis and inducer of vascular permeability.11 In
`addition, VEGF-A is 50,000 times more potent as an
`enhancer of vascular permeability than is histamine.12
`VEGF-A has been shown to be involved in the develop-
`ment of ocular diseases such as neovascular AMD.13
`The first ocular anti-VEGF agent was an aptomer-based
`therapeutic designed to target VEGF165, the most abun-
`dant isoform of VEGF-A in ocular disease. The VISION
`trials using this aptomer (Pegaptanib sodium, Macugen;
`OSI/Eyetech Pharmaceuticals, New York, New York, USA)
`demonstrated efficacy in all CNV subtypes secondary to
`AMD, and the number of pegaptanib-treated patients with all
`CNV subtypes who had 15-letter improvements was compa-
`rable to PDT-treated patients with predominantly classic
`CNV secondary to AMD in other studies (4% to 6%;
`Table).4 The proportion of patients losing ⱖ15 letters from
`the pretreatment baseline VA, however, was still about
`40% despite ITV pegaptanib injections every six weeks
`during the two-year trial. Nonetheless, because of the
`broad treatment effect with some visual benefit in all CNV
`subtypes secondary to AMD, along with the good safety
`profile, pegaptanib sodium was approved by the Food and
`Drug Administration in 2004 for the treatment of neovas-
`cular AMD.
`Ranibizumab (Lucentis; Genentech, Inc, South San
`Francisco, California, USA) is an anti-VEGF agent that is
`Food and Drug Administration approved for the treatment
`of all subfoveal CNV subtypes secondary to AMD. Ranibi-
`
`zumab is a humanized antigen-binding fragment that
`targets all VEGF-A isoforms and their biologically active
`degradation products, and it has been shown to bind and
`inhibit VEGF165, VEGF121, and VEGF110. Ranibizumab
`has been investigated in clinical trials in patients with
`neovascular AMD since 2000 (Table; Figure 1). MARINA
`was a randomized Phase III clinical trial of minimally
`classic or occult CNV secondary to AMD treated with
`monthly ITV ranibizumab (0.3 mg or 0.5 mg) or sham
`injections for 24 months. Anatomically, ranibizumab treat-
`ment was associated with arrested growth of and cessation
`of leakage from the CNV (including intense, progressive
`staining of the retinal pigment epithelium [RPE]). The
`mean change from baseline in each of the ranibizumab-
`treated groups differed significantly from the sham-injected
`group at 12 and 24 months (P ⬍ .001). Mean retinal
`thickness on ocular coherence tomography (OCT) was
`reduced dramatically in the first week after initial treat-
`ment, and this reduction was associated with mean im-
`provement in VA. With monthly injections, both the
`reduction in retinal thickness and the VA gains were
`maintained. At the end of the 24-month study, signifi-
`cantly more ranibizumab-treated patients had maintained
`or improved vision (lost ⬍15 letters of VA) than sham-
`injected patients, with 92% and 90% of ranibizumab-
`treated patients (0.3 mg and 0.5 mg, respectively) losing
`fewer than 15 letters of VA, compared with 53% of
`sham-injected patients (P ⬍ .001).5 In addition, the mean
`change in VA from baseline was a gain of 5.4 and 6.6
`
`628
`
`AMERICAN JOURNAL OF OPHTHALMOLOGY
`
`OCTOBER 2007
`
`0002
`
`
`
`>So
`
`= Ranibizumab 0.5 mg
`Ranibizumab 0.3 mg
`
`x
`-
`4
`
`i
`
`i
`
`2 10.
`3
`=
`Ss.
`2e
`22
`£2
`
`§
`
`mw
`
`\
`
`ij_—_i_]
`
`i
`
`|
`
`+
`
`B Ranibizumab 0.5 mg
`Ranibizumab0.3mg
`
`8°ge -
`
`£O
`
`o
`§
`=
`
`Sham
`MeanChangeinVisualAcuity(no.ofletters)
`(no.ofletters)
`
`ANCHORwasa randomized PhaseIII clinical trial of
`patients with predominantly classic CNV secondary to
`AMDtreated with monthly ITV ranibizumab (0.3 mg or
`0.5 mg) and sham PDT orsham injection and PDT for 24
`months. Anatomically, fluorescein angiography (FA) at 12
`monthsrevealed a dramatic reduction of leakage from the
`CNV(and areaofintense, progressive staining of the RPE)
`compared with baseline among ranibizumab-treated pa-
`tients compared with an overall increase in leakage among
`PDT-treated patients (P < .01). Analogous to the ana-
`tomic changes seen on OCT in the MARINAtrial, retinal
`thickness was markedly reduced within a week ofinitial
`treatment, and this reduction was associated with com-
`mensurate gains in VA. Repeated injections maintained
`this anatomical improvement and allowed for unprece-
`dented gains in VA at the primary endpoint at 12 months.
`At the 12-month time point, 94% and 96% of ranibi-
`zumab-treated patients (0.3 mg and 0.5 mg, respectively)
`had lost fewer than 15 letters of VA, compared with 64%
`of PDT-treated patients (P < .0001).° In addition, the
`mean change in VA from baseline was a gain of 8.5 and
`11.3 letters for ranibizumab-treated patients (0.3 mg and
`0.5 mg, respectively), compared withalossof 9.5 letters in
`PDT-treated patients (P < .001).° After 12 months, 36%
`and 40% of ranibizumab-treated patients gained 15 or
`more letters of VA (0.3 and 0.5 mg, respectively), com-
`pared with 6% of PDT-treated patients (P < .001).°
`While ranibizumab is an anti-VEGFantibody fragment,
`bevacizumab is a related humanized,
`full-length anti-
`VEGFantibody that was also developed by Genentech. At
`the July 2005 American Society of Retina Specialists
`(ASRS) meeting in Montreal, Canada, Phillip Rosenfeld,
`MD,presented a small series of patients with neovascular
`AMDtreated with off-label ITV bevacizumab (Food and
`Drug Administration approved for intravenous treatment
`of metastatic colon cancer in combination with chemo-
`therapy). The preliminary results from this series demon-
`strated good short-term safety andefficacy similar to that seen
`in patients treated with ranibizumab.!+ Other investigators
`have since demonstrated comparable clinical results with
`larger but still short-term and uncontrolled series of pa-
`tients with neovascular AMD treated with off-label ITV
`bevacizumab.'>-!7 These reports resulted in a relatively
`quick and widespread adoption of off-label ITV bevaci-
`zumab by retina specialists before the clinical availability
`of ranibizumab. Bevacizumabuse has continued around the
`world, and increasing pharmacokinetic, histopathology,
`and toxicity studies have subsequently been performed in
`RPEcell lines and bovine, rabbit, pig, and monkey models
`that demonstrate no evidence of ocular toxicity at the dose
`commonly administered.!8-° Durability of effect of bev-
`acizumab appears comparable to ranibizumab with appar-
`ent intravitreal concentrations up to one month after
`injection.2! There are no data available to assess the
`relative efficacy and safety of off-label ITV bevacizumab
`but with support from the National Eye Institute, The
`
`.
`@ POT
`
`* Ranibizumab 0.3 mg
`
`
`
`MeanChangeinVisualAcuity
`
`..
`
`Month
`
`FIGURE1. Visual acuity over time for the (Upper panel)
`MARINA,
`(Middle panel) ANCHOR, and (Lower panel)
`PIER PhaseIII trials. All three panels show the mean change
`from baseline in visual acuity [VA] (numberof letters) over
`time (months). The black boxes represent the mean change in
`VA for the 0.5 mg ranibizumab group,
`the gray triangles
`represent the mean change in VAfor the 0.3 mg group, and the
`gray circles represent the mean change in VA for the control
`group in each of the studies. Pairwise analysis of variance
`(ANOVA) adjusting for the VA score on day zero was used
`to analyze the mean change in VA from baseline at each
`monthly assessment. At each monthly assessment, there was
`a significant difference between the ranibizumab and control
`groups. (MARINA and ANCHORpanels are reproduced
`with permission from Rosenfeld and associates, N Engl
`J Med 2006;355:1419-1431 and Brown and associates,
`N Engl J Med 2006;355:1432-1444).
`
`letters for ranibizumab-treated patients (0.3 mg and 0.5
`mg, respectively), compared with a loss of 14.9 letters in
`sham-treated patients (P < .001). After 24 months, 26%
`and 33% of ranibizumab-treated patients gained 15 or
`more letters of VA (0.3 and 0.5 mg, respectively), com-
`pared with 4% of sham-injected patients (P < .001).°
`
`VOL. 144, NO. 4
`
`ANTI-VEGF AGENTS IN THE TREATMENT OF NEOVASCULAR AMD
`
`629
`
`0003
`
`0003
`
`
`
`Comparison of Age-Related Macular Degeneration Treat-
`ments Trial (CATT) is being organized to determine the
`safety and efficacy of intravitreal bevacizumab compared
`with ranibizumab.22 The trial will be a head-to-head,
`prospective, randomized, masked clinical trial of ranibi-
`zumab and off-label ITV bevacizumab in 1,200 patients.
`Monthly ranibizumab and bevacizumab injection regimens
`will be included as well as pro re nata (PRN) treatment
`regimens with both agents and is expected to begin this
`year. Direct comparisons of the efficacy and potential
`adverse events of off-label ITV bevacizumab and ranibi-
`zumab can only be made at the conclusion of such a
`head-to-head study.
`
`OCULAR RISKS AND POSSIBLE
`SYSTEMIC RISKS OF
`ANTI-VEGF THERAPY
`
`THE OCULAR RISKS OF ANTI-VEGF INJECTIONS APPEAR TO
`be primarily related to the injection process itself. In the
`MARINA, ANCHOR, and PIER trials, ranibizumab treat-
`ment was associated with a 1.7% or lower rate of key
`serious ocular adverse events such as endophthalmitis,
`retinal detachment, and uveitis.5,6 The theoretical risk
`exists that intraocular pan-VEGF suppression may cause
`choriocapillaris damage. However, these findings have not
`been collaborated by wide-field angiography looking for
`choriocapillaris damage in our patients who had more than
`24 injections of ranibizumab (unpublished data). Because
`intravenous anti-VEGF blockade is known to be associated
`with increased rates of thromboembolic events, systemic
`absorption of these agents from intraocular injections may
`increase these rates in treated patients. In animal studies,
`ITV ranibizumab was cleared from the vitreous with a
`half-life of approximately three days.23 After reaching a
`maximum at approximately one day, serum concentrations
`decline in parallel to vitreous concentrations with a
`systemic level ⬍1000 times the vitreous concentrations. In
`human trials, serum ranibizumab levels ranged from 0.3
`ng/ml to 2.36 ng/ml following monthly injections.24 These
`levels were below the concentration of ranibizumab (11
`ng/ml to 27 ng/ml) thought to be necessary to inhibit 50%
`of the biologic activity of VEGF-A. Systemic absorption of
`ITV bevacizumab has also been demonstrated in approxi-
`mately 11.8% of patients (Csaky K, personal communica-
`tion). This may be significant given bevacizumab’s
`terminal half-life in the systemic circulation is approxi-
`mately 20 days compared with 12 hours for ranibizumab. In
`the initial Phase III trials with ranibizumab, the actual key
`serious nonocular adverse events were not found to be
`significantly different
`than the corresponding control
`groups.5,6 However, in January 2007, Genentech released
`data that showed in preliminary analysis of the SAILOR
`cohort (a Phase IIIb study with much larger sample size
`than the Phase III trials), that the 0.5 mg dose had a
`
`significantly higher risk of stroke compared with the 0.3
`mg dose (1.2% vs 0.3%; P ⫽ .02).25 Ongoing safety
`analysis is being performed in patients in the SAILOR trial
`as well as longer term follow-up in the HORIZON trial
`(the extension study of MARINA and ANCHOR). The
`CATT trial is also designed to provide systemic safety data
`for both ranibizumab and bevacizumab.
`
`ALTERNATIVE DOSING STRATEGIES
`FOR ANTI-VEGF AGENTS
`
`BECAUSE OF BOTH OCULAR AND SYSTEMIC SAFETY CON-
`cerns, as well as cost and physician and patient conve-
`nience issues, ideally an alternative dosing strategy can be
`used to decrease the number of required injections for
`anti-VEGF agents. A recently completed Phase IIIb study,
`PIER, was a randomized, double-masked, sham-controlled
`clinical trial of patients with subfoveal CNV secondary to
`AMD treated with ITV ranibizumab monthly for the first
`three months, followed by mandated quarterly dosing for
`24 months (Regillo et al, 2006, unpublished data). For the
`first three months, both the anatomic improvements on
`OCT and the VA gains mirrored the ANCHOR and
`MARINA trials. After this period, the average VA de-
`creased to baseline levels, because quarterly injections
`were unable to maintain the anatomic improvements seen
`with monthly injections. This is thought to be directly
`related to the mandated quarterly dosing in PIER, with
`some patients requiring more frequent injections to main-
`tain the anatomic and visual improvements achieved in
`the first three months of the study. At the 12-month time
`point, the mean change in VA from baseline was a loss of
`1.6 and 0.2 letters for ranibizumab-treated patients (0.3 mg
`and 0.5 mg, respectively), compared with a loss of 16.3
`letters for sham-treated patients (P ⬍ .0001). In addition,
`83% and 90% of ranibizumab-treated patients (0.3 and 0.5
`mg, respectively) lost fewer than 15 letters of VA, com-
`pared with 49% of sham-injected patients. After 12
`months, 12% and 13% of ranibizumab-treated patients
`gained 15 or more letters of VA (0.3 and 0.5 mg,
`respectively), compared with 10% of
`sham-injected
`patients. Although the PIER trial results are far superior
`to those seen with any previous agent in AMD trials, the
`high expectations of overall improvement in mean visual
`acuity demonstrated in the ANCHOR and MARINA trials
`were not achieved. These initial PIER study data show that
`treating patients with CNV secondary to AMD every
`three months is less effective than monthly dosing with
`ranibizumab.
`Another small, uncontrolled, but very well-designed
`study exploring an alternate dosing strategy for ranibi-
`zumab is PrONTO, a prospective OCT imaging study of
`patients with all subtypes of CNV secondary to AMD
`treated with ITV ranibizumab monthly for the first two
`months (three injections), followed by PRN dosing there-
`
`630
`
`AMERICAN JOURNAL OF OPHTHALMOLOGY
`
`OCTOBER 2007
`
`0004
`
`
`
`after.26 OCT was used to evaluate quantitative retinal
`thickness, and retreatment decisions were based on these
`quantitative OCT measurements as well as on other
`criteria, such as visual loss associated with any edema
`observed on qualitative OCT, the presence of new classic
`CNV on FA, and the presence of new hemorrhage or
`persistent fluid on qualitative OCT. Of note, fluid was
`observed with qualitative OCT 100% of the time at the
`time of ITV injection, whereas retinal thickness increases
`greater than 100 m were observed with quantitative
`OCT and were the reason for injection only 16% of the
`time.26 ITV injections were also administered for decreased
`vision with evidence of fluid on qualitative OCT 41% of
`the time, for persistent fluid 29% of the time, for new
`subretinal hemorrhage with concomitant fluid on qualita-
`tive OCT 7% of the time, and for new classic CNV
`observed on FA as well as fluid on qualitative OCT 6% of
`the time. Preliminary results from the 40 patients enrolled
`in the study show that mean VA from baseline improved
`by 9.3 letters in ranibizumab-treated patients (P ⬍ .001)
`and mean central retinal thickness decreased by 178 m
`(P ⬍ .001). At one month (one injection), 85% of
`ranibizumab-treated patients had resolution of intraretinal,
`subretinal edema and 77.5% had resolution of subretinal
`fluid. At two months (two injections), 92.5% of ranibi-
`zumab-treated patients had resolution of subretinal fluid
`and intraretinal cysts, and at three months (three injec-
`tions), 97.5% had resolution of subretinal fluid, whereas
`92.5% had resolution of intraretinal cysts.26 At 12 months,
`the mean number of injections was 5.6, and 17.5% (7/40)
`of patients did not require retreatment after the initial
`three ITV injections of ranibizumab. This OCT-guided
`dosing strategy achieved results comparable to the MARINA
`and ANCHOR trials with less than half of the required
`monthly injections used in the Phase III trials.
`
`WHY DO PAN-VEGF BLOCKING
`AGENTS IMPROVE VA IN
`NEOVASCULAR AMD?
`
`THERE ARE NUMEROUS EXAMPLES OF RETINAL DISEASES IN
`which the anatomy of the fovea must be restored before
`visual improvements can occur. Two examples are macula-
`off retinal detachments and Irvine-Gass cystoid macular
`edema, where VA improves only when the anatomic
`pathology is corrected. Conversely, irreversible visual loss
`from progressive intraretinal fibrosis or atrophy occurs
`when the retina is subjected to prolonged subretinal and
`intraretinal edema as in diffuse diabetic macular edema
`and chronic central serous retinopathy. The MARINA
`and ANCHOR trials demonstrated a marked reduction in
`leakage from CNV in ranibizumab-treated patients com-
`pared with increased leakage in the controls, with quali-
`tative OCT images of the retina of ranibizumab-treated
`patients appearing completely dry in almost all patients
`
`after 24 ITV injections. These anatomic changes were
`associated with overall visual
`improvement in ranibi-
`zumab-treated patients.
`No previous treatment has been as efficacious as ranibi-
`zumab at eliminating the edema in the subretinal space
`and neurosensory retina in neovascular AMD. Pegaptanib
`sodium, which theoretically only blocks one isoform of
`VEGF-A in the eye, rarely eliminates all macular edema
`even with continued use. In contrast, most ranibizumab-
`treated patients (⬎70%) have resolution of all macular
`edema as seen by OCT within one month of their first
`injection, and more than 90% have resolution of all edema
`after three consecutive monthly injections.26 The visual
`gains seen in the initial three months of therapy in the
`ANCHOR, MARINA, and PIER trials are very similar
`and demonstrate visual gains very quickly after initiation
`of therapy and resolution of retinal edema. The on-average
`return to baseline (loss of
`the improvement
`in VA
`achieved in the first three months) in the PIER trial is
`likely related to the inability of mandated quarterly ITV
`ranibizumab injections to eliminate CNV leakage for all
`patients; some patients likely required more frequent ITV
`injections, whereas others could have used less frequent
`ITV injections, as supported by OCT findings in the trial
`(Regillo et al, 2006, unpublished data). This suggests that
`greater control through a criteria-based retreatment strat-
`egy may be warranted.
`
`● OCT AND THE MANAGEMENT OF OCULAR DISEASES:
`OCT is a noninvasive diagnostic imaging tool that can
`perform high-resolution, micron-scale, cross-sectional, or
`tomographic imaging in tissue by measuring the echo time
`delay and intensity of backscattered light.27 OCT allows
`real-time in situ imaging of tissue with resolution of up to
`15 m and enables examination of the epiretinal, in-
`traretinal, and subretinal morphology.27 OCT uses several
`measures to generate its images,
`including the central
`subfield, center point thickness, automated retinal bound-
`ary lines, inner retinal surface, and the RPE band. In OCT
`images, bright colors such as red and white represent
`regions of high reflectivity such as the nerve fiber layer, the
`plexiform layers, or the RPE, whereas dark colors such as
`blue or black represent regions of low reflectivity, such as
`the nuclear layers and photoreceptor inner and outer
`segments. Only the hyporeflective photoreceptor layer is
`visible at the foveola, where the indentation of the foveal
`pit is at its maximum.
`OCT has been used in the management of many retinal
`diseases, including macular holes,28 macular edema,29,30
`and neovascular AMD.31 In AMD, OCT is useful
`in
`evaluating and documenting retinal angiomatous prolifer-
`ation (RAP) lesions before and after laser photocoagula-
`tion in patients with RAP lesions secondary to AMD.31
`The early stages of RAP are difficult to detect with both
`contact lens biomicroscopy and angiographic examina-
`tions, but typical intraretinal features can be observed with
`
`VOL. 144, NO. 4
`
`ANTI-VEGF AGENTS IN THE TREATMENT OF NEOVASCULAR AMD
`
`631
`
`0005
`
`
`
`FIGURE 2. Examples of optical coherence therapy (OCT) images from different patients being treated with ranibizumab for
`neovascular age-related macular degeneration (AMD) demonstrating various patterns of recurrent choroidal neovascularization
`(CNV) leakage such as (Top row) diffuse edema; (Second row) intraretinal cysts; (Third row) subretinal fluid; or (Bottom row)
`subretinal pigment epithelium (RPE) fluid. Top row (Left): Three months after completing the MARINA protocol, the patient’s
`visual acuity (VA) in the involved eye was 20/25 with no injections for four months. Automated center point foveal thickness was
`263 . (Right) Diffuse edema: The first sign of disease progression was evident on OCT six weeks later with diffuse edema and
`loss of the foveal depression. The VA declined to 20/50. Automated center point thickness increased to 341 . Second row (Left):
`The patient finished the MARINA study with a VA of 20/63 in the treated eye and a normal foveal depression but quantitative
`OCT revealed atrophy with a foveal central thickness of 98 . (Right) Intraretinal cysts: Five months after the patient’s last
`required intravitreal injection in the study, intraretinal cysts developed with no change in her best-corrected VA (20/63). Center
`point foveal thickness increased but was still in “normal” range at 178 . Third row (Left): The patient finished the MARINA
`study with a VA of 20/25 in the treated eye and an OCT scan showing a near normal appearing foveal contour with underlying
`CNV. The central foveal thickness was 138 . (Right) Subretinal fluid: Just six weeks into the extension study (10 weeks after
`her last scheduled injection in the MARINA trial), subretinal fluid became evident on the OCT with a commensurate decrease in
`VA to 20/40. Quantitative center point foveal thickness increased to 204 . Bottom row (Left): The patient completed the
`MARINA study with a VA of 20/50 and no evidence of CNV leakage determined by OCT. Automated center point foveal
`thickness was 160 . (Right) Sub-RPE fluid: Three months after her last ITV ranibizumab injection, OCT revealed fluid in the
`sub-RPE space. Automated center point foveal thickness increased slightly to 190 , but VA improved to 20/25.
`
`632
`
`AMERICAN JOURNAL OF OPHTHALMOLOGY
`
`OCTOBER 2007
`
`0006
`
`
`
`qualitative OCT.31 In the management of diabetic reti-
`nopathy, OCT has been demonstrated to be more sensitive
`than contact lens biomicroscopy for the detection of
`mild foveal thickening in patients with diabetic macular
`edema.29 An OCT study of patients with diabetic macular
`edema found that best-corrected VA was significantly
`correlated with central foveal thickness (r ⫽ ⫺0.528;
`P ⬍ .01).30 There was 77% agreement between OCT
`and clinical examination. In addition, cystoid macular
`edema was detected in 15.4% of eyes with OCT, but
`cystoid macular edema was not detected with contact
`lens biomicroscopy and FA in 40% and 60% of these
`eyes, respectively. In this study, OCT detected serous
`foveal detachments detected that could not be seen with
`either contact lens biomicroscopy or FA.30 This finding
`illustrates that OCT can document subtle disease activ-
`ity in macular disease often undetectable by standard
`clinical examinations.
`
`● QUANTITATIVE VS QUALITATIVE OCT: Quantitative
`OCT uses fast macular thickness-based acquisition proto-
`cols, in which six radial scans are centered on a patient’s
`fixation and then compressed into one scan, which can be
`acquired quickly (1.92 seconds) and contains 128 A-scans
`per line.31 However, fast macular thickness-based acquisi-
`tion protocols provide low-resolution images that have
`some inherent errors particularly in eyes with neovascular
`AMD.32 Errors of retinal boundary detection and thickness
`measurement were observed in 92% of AMD eyes when
`fast macular thickness-based acquisition protocols were
`used.32 Poor VA in patients with neovascular AMD makes
`fixation difficult to maintain, and the center point thick-
`ness measurement is dependent on the fovea being in the
`center of each vector line. Thus any shift of the patient’s
`fixation causes significant error. Even when fixation is
`excellent, the anatomic heterogeneity of CNV lesions
`(eccentric with variable intraretinal cysts and subretinal
`fluid pockets) often misleads the automated detection of
`retinal boundaries necessary for accurate quantitative mea-
`surements. An additional and significant problem of quan-
`titative OCT is that sub-RPE fluid (pigment epithelial
`detachment) is not quantified even if fixation and retinal
`boundary detection is accurate.
`In contrast, qualitative OCT using normal acquisition
`protocols (Macular Thickness Map acquisition on the
`Stratus OCT 3, software version 4.1; Carl Zeiss Meditec,
`Dublin, California, USA) provides a transverse resolution
`of 512 A-scans per B-scan that yields more detailed
`intraretinal information.31 Fortunately, fixation difficulties
`inherent with AMD patients do not affect qualitative
`OCT analysis because it relies much less on each vector
`cut being centered on fixation. Qualitative OCT simply
`looks for any anatomic evidence of CNV leakage in the
`imaged tissues. This more detailed scan is useful
`for
`visualizing neovascular-associated leakage and provides
`higher-resolution images without relying on the machine
`
`to interpret the images. The actual thickness and morphol-
`ogy of the CNV complex are quite difficult to assess on
`OCT images because the posterior CNV boundary is
`difficult to ascertain. Higher resolution OCT machines (in
`development by both Topcon and Carl Zeiss Meditec) may
`allow this component to be more readily assessed. Until
`higher resolution OCT images are available to facilitate
`this assessment, however, one must look for evidence of
`CNV leakage (or VEGF-A-mediated leakage from other
`sources) in the intraretinal, subretinal, and sub-RPE space
`using the available OCT protocols.
`Just as there are recognizable OCT patterns in diabetic
`macular edema,33 specific patterns of structural changes
`resulting from CNV leakage can be seen on qualitative
`OCT in patients with neovascular AMD. To define the
`patterns of leakage after successful anatomic correction of
`CNV leakage with ranibizumab, we monitored qualitative
`OCT at our clinical trial center (which recruited and
`followed the largest number of patients in both the
`ANCHOR and MARINA Phase III clinical trials) with
`monthly ETDRS refractions, clinical examinations, and
`high-resolution qualitative OCT. We looked for the first
`anatomic changes indicating recurrent disease activity on
`the six radial qualitative OCT scans (using Macular
`Thickness Map acquisition protocol). There were four
`distinct patterns: diffuse retinal edema, intraretinal cysts,
`subretinal fluid, and sub-RPE fluid or serous pigment
`epithelial detachment (Figure 2). Diffuse retinal edema is
`a sponge-like thickening that appears as increased retinal
`thickness with areas of reduced intraretinal reflectivity
`compared with retina without thickening (Figure 2, Top
`row). Intraretinal cysts (Figure 2, Second row) and sub-
`re