Rationale and Application of the Protocol S

AntieVascular Endothelial Growth Factor

Algorithm for Proliferative Diabetic
Retinopathy
Jennifer K. Sun, MD, MPH,1 Adam R. Glassman, MS,2 Wesley T. Beaulieu, PhD,2 Cynthia R. Stockdale, MSPH,2
Neil M. Bressler, MD,3 Christina Flaxel, MD,4 Jeffrey G. Gross, MD,5 Michel Shami, MD,6 Lee M. Jampol, MD,7
for the Diabetic Retinopathy Clinical Research Network

Purpose: To present the rationale, guidelines, and results of ranibizumab treatment for proliferative diabetic
retinopathy (PDR) in Diabetic Retinopathy Clinical Research Network (DRCR.net) Protocol S.
Design: Post hoc analyses from a randomized clinical trial.
Participants: Three hundred five participants (394 study eyes) having PDR without prior panretinal photo-
coagulation (PRP).
Methods: Intravitreous ranibizumab (0.5 mg) versus PRP for PDR. Ranbizumab-assigned eyes (n ¼ 191)
received monthly injections for 6 months unless resolution was achieved after 4 injections. After 6 months, in-
jections could be deferred if neovascularization was stable over 3 consecutive visits (sustained stability). If
neovascularization worsened, monthly treatment resumed. Panretinal photocoagulation could be initiated for
failure or futility criteria.
Main Outcome Measures: Neovascularization status through 2 years.
Results: At 1 month, 19% (35 of 188) of ranibizumab-assigned eyes showed complete neovascularization
resolution and an additional 60% (113) showed improvement. At 6 months, 52% (80 of 153) showed neo-
vascularization resolution, 3% (4) were improved, 37% (56) were stable, and 8% (13) had worsened since the last
visit. Among eyes with versus without resolved neovascularization at 6 months, the median (interquartile range)
number of injections between 6 months and 2 years was 4 (1e7; n ¼ 73) versus 7 (4e11; n ¼ 67; P < 0.001).
Injections were deferred in 68 of 73 eyes (93%) meeting sustained stability at least once during the study; 62% (42
of 68) resumed injections within 16 weeks after deferral. At 2 years, 43% (66 of 154) showed neovascularization
resolution, 5% (7) showed improvement, 23% (36) were stable, and 27% (42) had worsened since the last visit.
Only 3 eyes met criteria for failure or futility through 2 years.
Conclusions: The DRCR.net treatment algorithm for PDR can provide excellent clinical outcomes
through 2 years for patients initiating antievascular endothelial growth factor (VEGF) therapy for PDR. When
choosing between anti-VEGF and PRP as first-line therapy for PDR, treatment decisions should be guided
by consideration of the relative advantages of each therapeutic method and anticipated patient compliance
with follow-up and treatment recommendations. Ophthalmology 2019;126:87-95 ª 2018 by the American
Academy of Ophthalmology

Supplemental material available at www.aaojournal.org.

Treatment options for proliferative diabetic retinopathy therapy for PDR is noninferior to PRP at 2 years for change
(PDR) include panretinal photocoagulation (PRP), in visual acuity from baseline (5-letter noninferiority
antievascular endothelial growth factor (VEGF) therapy, and margin).4 Ranibizumab also provided several benefits over
vitrectomy.1e4 Vitrectomy typically is reserved for cases of PRP through 2 years. The ranibizumab group showed
nonclearing vitreous hemorrhage or traction retinal detach- superior gain in visual acuity over the course of 2 years,
ment threatening or involving the macula. Although PRP has decreased peripheral visual field loss, and fewer
been standard care since the 1970s,5 recent clinical trial vitrectomies. In addition, ranibizumab-treated eyes were
results suggest anti-VEGF therapy is a reasonable alterna- less likely to develop central-involved diabetic macular
tive to PRP for treatment of PDR through at least 2 years, edema (CI-DME) causing vision impairment of 20/32
contingent on patient compliance and access to treatment.4,6 or worse. Similar 1-year outcomes were reported using
The Diabetic Retinopathy Clinical Research Network aflibercept in the Clinical efficacy and mechanistic evalua-
(DRCR.net) Protocol S trial demonstrated that ranibizumab tion of aflibercept for proliferative diabetic retinopathy

ª 2018 by the American Academy of Ophthalmology https://doi.org/10.1016/j.ophtha.2018.08.001 87

Published by Elsevier Inc. ISSN 0161-6420/18
 Ophthalmology Volume 126, Number 1, January 2019

(CLARITY) randomized clinical trial.6 Based on these

findings, the DRCR.net recommends considering anti-
VEGF therapy for eyes with PDR, especially for eyes
with coexisting CI-DME for which anti-VEGF is indicated
already.7
The decision to use anti-VEGF versus PRP varies based
on the individual circumstances of each patient, including
anticipated compliance and access to treatment. For physi-
cians considering the use of anti-VEGF to treat PDR, this
article clarifies the rationale and guidelines used in Protocol
S. We include data from Protocol S outlining the advantages
and disadvantages of anti-VEGF versus PRP in specific
clinical situations.

Methods
Diabetic Retinopathy Clinical Research Network
AntieVascular Endothelial Growth Factor
Treatment Algorithm for Proliferative Diabetic
Retinopathy
The DRCR.net anti-VEGF treatment algorithm for PDR was based
on the clinical experiences of the network investigator group and
retinopathy outcomes achieved with an analogous defer-and-
extend DRCR.net anti-VEGF regimen for CI-DME.8 Protocol S
evaluated whether PDR could be controlled with an initial series
of consecutive 0.5-mg monthly ranibizumab injections with sub- Figure 1. Principles of the Diabetic Retinopathy Clinical Research
sequent therapy determined based on retinal neovascularization Network (DRCR.net) antievascular endothelial growth factor treatment
status at each visit. The study adhered to the tenets of the Decla- algorithm for proliferative diabetic retinopathy. *Four-week, not 1-month,
ration of Helsinki9 and was approved by multiple institutional intervals were used. yFour injections were required every 4 weeks initially; it
review boards. Study participants provided written informed is not known whether a different number of injections initially would have
consent. worked as well. The DRCR.net also required 2 additional injections at
The treatment algorithm can be summarized in 3 steps. First, months 5 and 6 unless there was complete absence of neovascularization
start with 6 monthly injections with 1 exception: if all neo- (NV; resolved disease). zRelevant details: (1) deferral of injections owing to
vascularization is resolved after 4 or 5 injections, injections can be stability occurred only when sustained stability criteria were met, defined as
deferred. Second, after 6 months, continue injections if the neo- NV clinically unchanged at the current visit and the last 2 injection visits;
vascularization continues to worsen or improve, but defer (2) resolved was defined as absence of NV of the iris, NV of the disc, NV
injections if the neovascularization is stable at the current visit and elsewhere, and NV of the angle (if the angle was assessed). xInjection was at
the last 2 visits (sustained stability). Third, resume monthly anti- investigator discretion if NV status was sustained stability or resolved and
VEGF injections if neovascularization worsens after withholding was performed 15% and 23% of the time in these cases, respectively.
k
injections. If sustained stability is achieved again, injections may Panretinal photocoagulation was permitted if failure criteria were met,
be deferred once again, but this requires at least 3 consecutive namely, if NV worsened substantially despite at least 4 monthly injections
injections (1 given for the initial status of worsened or iris NV involving the angle developed. {Follow-up continued every 4
neovascularization and 2 more while the neovascularization is weeks through the 52-week visit and did not permit extension of follow-up
stable). Panretinal photocoagulation is administered only if neo- until after the 52-week visit. If injection was withheld because of no res-
vascularization worsens substantially despite anti-VEGF. Onset or olution or sustained stability at 3 consecutive visits after the week 52 visit,
worsening of preretinal or vitreous hemorrhage is not necessarily the follow-up interval was doubled to 8 weeks and then again to 16 weeks if
graded as neovascularization worsening unless the hemorrhage still no change.
precludes evaluation of neovascularization. Further explanations
are provided in Figure 1 and Table 1, which summarize the key
terms used in the DRCR.net anti-VEGF treatment algorithm for each visit, investigators used the following methods, individually
PDR. or in combination, to assess neovascularization status in 3410 ex-
Determining the Status of Neovascularization. At each pro- aminations in Protocol S: extended ophthalmoscopy in 2595
tocol visit, DRCR.net investigators determined the status of retinal (76%), undilated examination of the iris in 904 (27%), fundus
neovascularization by categorizing neovascularization of the iris, photography in 471 (14%), fluorescein angiography in 80 (2%), or
neovascularization of the disc, and neovascularization elsewhere as ultrasound in 3 (<1%).
resolved (absent), present but improved, stable, or worsened Extension of Follow-up Visits. To decrease the burden of visits
compared with the last visit (Fig 2). Assessment of the angle was while minimizing risk of vision loss, follow-up intervals could be
required only if neovascularization of the iris was present or if the extended gradually: first to every 8 weeks, if injections were deferred for
eye met criteria for increased IOP. 2 consecutive months, and then to every 16 weeks if there was still no
The methods for evaluation of neovascularization status at each recurrence or worsening of neovascularization. If neovascularization
visit were left to investigator discretion based on usual clinical recurred or worsened at any point after deferral, monthly anti-VEGF
practice. In addition to slit-lamp and dilated fundus examinations at was resumed until resolution or sustained stability over 3 consecutive

88
 Sun et al 
Anti-VEGF Treatment Algorithm for PDR

Table 1. Neovascularization Assessment Definitions

Term Definition
Resolution (or success) Apparent absence of neovascularization (including fibrous proliferans) of the retina, disc, iris, and angle (if the angle was
assessed)
Stable Clinically unchanged neovascularization (no improvement or worsening) from the prior visit
Improved Decrease in the size of neovascularization or diminished density of neovascularization in any area (retina, disc, iris, or
angle) compared with the prior visit
Sustained stability Stable neovascularization at the current visit compared with the last 2 injection visits
Worsening Recurrent or worsening neovascularization since the last visit or presence of vitreous or preretinal hemorrhage that
prevented evaluation of neovascularization status
Failure Growth of neovascularization after at least 4 consecutive injections such that the neovascularization is greater in extent
than when treatment was initiated or definite neovascularization worsening, after an injection, that the investigator
believed was likely to lead to vision loss in the absence of panretinal photocoagulation
Futility Neovascularization persisted or recurred such that it was equal to or greater than when treatment was initiated

injections (Fig 1). In Protocol S, extension of follow-up was not allowed neovascularization was resolved completely at 1513 of 3410 visits
before 1 year to characterize the monthly response of neo- (44%) when neovascularization was assessed (Fig 3). At these
vascularization, even when injections were being withheld. visits, injections could be performed at investigator discretion
Treatment of Diabetic Macular Edema in Eyes with Prolif- and were administered 351 of 1513 times (23%). Among eyes
erative Diabetic Retinopathy. Treatment for diabetic macular with resolved neovascularization at 6 months, the median
edema (DME) in Protocol S was at investigator discretion; how- number of subsequent injections between 6 months and 2 years
ever, guidelines for DME treatment were provided and were was 4 (IQR, 1e7; n ¼ 73), which was fewer than eyes that were
consistent with the approach used in DRCR.net Protocol T.10 not resolved at 6 months, which received 7 (IQR, 4e11; n ¼ 67;
Therefore, for eyes with DME and vision loss in addition to P < 0.001). Among 149 eyes with apparent resolution of
PDR, the Protocol T DME treatment algorithm would be neovascularization at any time, 139 (93%) received an additional
appropriate to follow. In such circumstances, an injection could anti-VEGF injection for either PDR or PDR with concurrent DME.
be administered if indicated for either PDR or DME, and the
follow-up visit interval was the shorter of the 2 intervals indicated. Improvement of Neovascularization
The initial 4-month period of consecutive monthly ranibizumab
Results dosing resulted in neovascularization improvement or resolution
for most eyes. At the 4-, 8-, 12-, and 16-week visits, 60% (113 of
Number of Injections 188), 32% (58 of 181), 26% (44 of 171), and 16% (28 of 171) of
eyes showed improved neovascularization status compared with
As previously reported, the median (interquartile range) number of the prior visit. In addition, 79% (148), 67% (121), 66% (113), and
ranibizumab injections administered for PDR or DME in the rani- 65% (112) of eyes showed either resolution or improvement from
bizumab group was 5 (IQR, 4e6) over the first 6 months, 2 (1e4) in the prior visit (Fig 3). Starting at 24 weeks through 2 years, during
the second 6 months, and 3 (1e6) in the second year, totaling which time monthly injections were not mandated and when many
10 (7e14) over 2 years.4 Compliance with treatment guidelines for eyes already showed resolution, rates of improvement at each
PDR was 97%. For eyes in the ranibizumab group that did not have protocol visit were lower and ranged from 3% to 7%. Over the
vision-impairing central-involved DME at baseline, the median same period, the rate of neovascularization resolution or
number of injections was 5 (4e6) over the first 6 months, 2 (1e4) in improvement at each visit ranged from 47% to 59%.
the second 6 months, and 3 (1e5) in the second year, totaling
10 (6e13) over 2 years. For eyes in the ranibizumab group with Stability of Neovascularization
baseline vision-impairing central-involved DME, the median
number of injections for PDR or DME was 5 (4e6) over the first After the initial 6 months, deferral of injections was permitted at
6 months, 4 (2e6) in the second 6 months, and 5 (2e7) in the investigator discretion if the neovascularization status achieved
second year, totaling 14 (10e17) over 2 years. was sustained stability (i.e., became stable and remained stable
after 2 additional, consecutive injections). Among 153 eyes
Resolution of Neovascularization (Treatment completing the 6-month visit (the first time sustained stability
Success) criteria could be met), 56 (37%) met stability criteria compared
with the last visit, including 34 (22%) achieving sustained stability
At 1 month (the first follow-up visit), 35 of 188 eyes evaluated compared with the last 3 consecutive injections. The distribution of
(19%) showed apparent complete resolution of neovascularization stability and sustained stability at protocol visits through 2 years is
(including the absence of fibrous proliferans) according to the presented in Figure S1 (available at www.aaojournal.org). When
treating ophthalmologist. An increasing proportion of eyes showed sustained stability was achieved, injections were at investigator
resolution from the 2-month visit through the 6-month visit (Fig 3). discretion. Investigators chose to administer treatment in 66 of
At 4 months, the first time point at which injections could be 439 visits (15%) at which sustained stability was noted.
deferred for success, neovascularization had resolved in 84 of Injections were deferred in 68 of the 73 eyes (93%) meeting
171 eyes (49%). Among the 84 eyes with neovascularization of sustained stability at least once during the study. Among these 68
the disc based on investigator judgment at baseline, 38 (45%) eyes, 11 (16%) did not receive another injection through 2 years
showed no neovascularization at the 4-month visit, compared and 66 (97%) deferred for at least 2 consecutive injections. Of the
with 45 of 85 (53%) with only neovascularization elsewhere 68 eyes deferring at least 1 injection for sustained stability, 42
at baseline. Across all follow-up visits through 2 years, (62%) resumed injections by 16 weeks after deferral (Fig 4).

89
 Ophthalmology Volume 126, Number 1, January 2019

had worsened by 2 levels or more of diabetic retinopathy

severity from baseline (99% of 191 eyes had gradable
photographs at baseline and 89% of 160 eyes had gradable
photographs at 2 years). When eyes achieved sustained stability
for which an injection was deferred, worsened neovascularization
was present at the next visit 24% of the time (83 of 353 visits).
Three of 191 eyes met failure criteria for substantial PDR
worsening. Of these 3 cases, 1 eye demonstrated worsening neo-
vascularization of the angle and 2 received PRP (1 for vitreous
hemorrhage with vision loss and 1 for neovascularization growth).
Among eyes that received PRP at any point over 2 years, most (8
of 12) received PRP during vitrectomy; only 4 eyes received PRP
in the clinic strictly for failure criteria or with protocol chair
approval. Only a single eye met futility criteria for worsening PDR
despite monthly anti-VEGF injections.

Discussion

Although Protocol S used ranibizumab, results from other

studies of anti-VEGF agents seem to show similar outcomes
with excellent regression of neovascularization. This includes
a recent randomized trial of aflibercept for PDR treatment
over 1 year in the absence of CI-DME, but not necessarily
without prior PRP.6 In addition, similar findings were seen in
phase 3 trials of aflibercept and ranibizumab for DME.11e14
Thus, recommendations from the DRCR.net for the clinical
scenarios below are given for anti-VEGF therapy in general,
with the caveat that results obtained with other agents or
treatment regimens may not be identical to those obtained in
Protocol S. Based on the findings from Protocol S, specific
scenarios are discussed below.

Proliferative Diabetic Retinopathy in the

Absence of Central-Involved Diabetic Macular
Edema
Eyes with PDR and no CI-DME received monotherapy with
either ranibizumab, according to the PDR treatment algorithm
detailed above, or PRP. Based on Protocol S results, both PRP
and anti-VEGF seem to be viable therapeutic options in eyes
with PDR and no DME. Although anti-VEGF therapy is
effective at regressing retinal neovascularization and
Figure 2. Fundus photographs showing examples of neovascularization
reducing the risk of CI-DME developing, it is not cost
change over time. Paired (A, C, E, G) baseline and (B, D, F, H) follow-up
fundus photographs of retinal neovascularization showing 4 types of neo-
effective in this cohort compared with PRP within the $50 000
vascularization change over time. A, B, Resolved: neovascularization that to $150 000 per quality-adjusted life-year range frequently
resolved fully without any residual vessels or fibrosis 1 week after cited in the United States.15 Distinct advantages of each
antievascular endothelial growth factor (VEGF) treatment. C, D, treatment method may guide the clinical decision for
Improved: neovascularization decreased in extent and severity 1 week after individual eyes (summary in Table 2).
anti-VEGF therapy, but with some residual neovascularization and fibrosis.
E, F, Stable: neovascularization that remained stable in extent and severity Proliferative Diabetic Retinopathy in the
over a period of 1 year without treatment. G, H, Worsened: neo- Presence of Central-Involved Diabetic Macular
vascularization of the disc that worsened in extent and severity over 32
Edema
weeks.
Anti-VEGF is considered first-line therapy in most eyes
with CI-DME. Ranibizumab was highly effective at treating
Worsening Neovascularization PDR in Protocol S; aflibercept was highly effective at
Rates of worsening were low (2%e7%) during the first 4 months treating PDR in CLARITY.4,6 Thus, the use of PRP may be
of the study when eyes were required to receive monthly ranibi- deferred in eyes with PDR that otherwise are receiving anti-
zumab (Fig 3). At 2 years, 42 of 154 eyes (27%) showed VEGF treatment for DME because the same anti-VEGF
worsening from the prior visit. However, among these 42 eyes, agent will treat both the DME and the PDR. As soon as
only 5 of 38 eyes (13%) with gradable photographs at 2 years anti-VEGF therapy is no longer required because of

90
 Sun et al 
Anti-VEGF Treatment Algorithm for PDR

Figure 3. Graph showing status of retinal neovascularization at each visit as determined by investigators (R ¼ randomization visit at which injection was
required).

resolution or stability of DME, these eyes can be re- Therefore, anti-VEGF treatment can function as mono-
evaluated based on neovascularization status to see if therapy for PDR, at least for the first 2 years if patient
treatment for PDR is still necessary and whether anti-VEGF compliance with follow-up is good. However, neo-
will be continued or if PRP will be used. In such situations, vascularization worsening was observed at approximately
the use of anti-VEGF agents through 2 years can be 13% of visits, with more worsening later in the 2-year
considered cost effective within the $50 000 to $150 000 per follow-up period. Therefore, higher rates of PRP for wors-
quality-adjusted life-year range cited in the United States ening PDR may result in less frequent dosing or longer
when treating the worse-seeing eye.15 follow-up intervals than those used in Protocol S.

High-Risk Proliferative Diabetic Retinopathy Vitrectomy for Proliferative Diabetic

with or without Vision-Impairing Diabetic Retinopathy
Macular Edema Vitrectomy was permitted in Protocol S for traction retinal
For eyes with high-risk PDR (i.e., Early Treatment Diabetic detachment causing visual impairment or nonclearing vitre-
Retinopathy Study level 71 or more) that did not require ous hemorrhage. However, investigators were asked to defer
vitrectomy at baseline, rates of PDR-worsening events (e.g., surgery for at least 8 weeks after the onset of vitreous hem-
vitreous hemorrhage, retinal detachment, neovascularization orrhage. Among ranibizumab-treated versus PRP-treated
of the angle, neovascular glaucoma) were greater compared eyes, there were fewer vitreous hemorrhages (27% vs. 34%;
with eyes having less than high-risk PDR, regardless of P ¼ 0.09), retinal detachments (6% vs. 10%; P ¼ 0.08), and
treatment.16 However, the average improvement in visual vitrectomy procedures (4% vs. 15%; P < 0.001).4 Eight eyes
acuity over 2 years was greater with ranibizumab than in the ranibizumab group underwent vitrectomy for the
PRP, and the difference seemed to be greater among eyes
with high-risk PDR.17 Thus, it seems appropriate to
consider ranibizumab and possibly other anti-VEGF
agents as monotherapy for eyes with PDR, even when the
PDR is considered high-risk PDR and prompt vitrectomy is
not being planned.

Worsening Proliferative Diabetic Retinopathy in

Eyes Receiving Ranibizumab for Proliferative
Diabetic Retinopathy
Very few eyes randomized to ranibizumab met failure
criteria for substantial PDR worsening (3 of 191 eyes) and
only 12 eyes received PRP, 8 of which were during
vitrectomy. These results suggest that eyes being treated Figure 4. Graph showing the cumulative probability of resuming in-
with ranibizumab for PDR according to the Protocol S al- jections after deferral because of sustained stability of neovascularization
gorithm are unlikely to need PRP for treatment of PDR. (stability at 3 consecutive visits).

91
 Ophthalmology Volume 126, Number 1, January 2019

Table 2. Comparison of the Relative Advantages and Disadvantages of AntieVascular Endothelial Growth Factor versus Panretinal
Photocoagulation for Treatment of Eyes with Proliferative Diabetic Retinopathy

AntieVascular Endothelial Growth Factor Panretinal Photocoagulation

No. of procedures for PDR Multiple injections needed for optimal results (e.g., Can often be performed in 1e2 sessions, but may require
treatment median number of injections over 2 yrs was 14 in eyes additional treatments (e.g., 45% of eyes treated with
with baseline central-involved DME and 10 in eyes PRP needed additional PRP within 2 yrs of initial
without baseline central-involved DME) treatment)
Frequency of follow-up visits Monthly or near monthly visits may be needed through First follow-up after PRP usually scheduled in 3e4 mos
the first year with subsequent follow-up periods with subsequent follow-up periods extended if
extended if neovascularization remains quiescent neovascularization remains quiescent
Visual acuity after 2 yrs Noninferior visual acuity at 2 yrs to PRP (5-letter Visual acuity at 2 years with anti-VEGF was non-inferior
noninferiority margin) to PRP (5-letter non-inferiority margin)
Visual acuity over 2 yrs Average visual acuity over 2 yrs superior to PRP Average visual acuity over 2 yrs inferior to anti-VEGF
Effect on peripheral visual Nondestructive therapy with better preservation of Destructive therapy that leads to defects in peripheral
field peripheral visual field as compared with PRP visual field
Effect on DME status Improves DME status and reduces rate of central- Three-fold higher rate of central-involved DME onset
involved DME onset compared with PRP compared with anti-VEGF over 2 yrs; in addition,
DME develops in some eyes after PRP, which can lead
to decreased vision
Need for additional Few eyes receiving anti-VEGF by the DRCR.net Few eyes that receive PRP will need anti-VEGF for
procedures for PDR or algorithm will need PRP for persistent or worsening persistent or worsening neovascularization; need for
DME treatment neovascularization; need for vitrectomy over 2 yrs is vitrectomy over 2 yrs is higher in PRP-treated eyes;
lower in anti-VEGFetreated eyes; few eyes will need eyes may need anti-VEGF for DME onset or
additional DME treatment worsening
Risk of adverse events Invasive procedure with associated risks of Noninvasive with no risk of endophthalmitis and
associated with invasive endophthalmitis, traumatic retinal tear, or cataract minimal risk of retinal tear or cataract
procedures
Risk of vitreous hemorrhage Possibly lower rate of vitreous hemorrhage through 2 yrs Possibly higher rate of vitreous hemorrhage through
2 yrs
Risk of traction retinal Possibly lower rate of retinal detachment through 2 yrs; Possibly higher rate of retinal detachment through 2 yrs;
detachment increased retinal traction is possible, but low rates of increased retinal traction is possible, but low rates of
traction retinal detachment for eyes that do not have traction retinal detachment for eyes that do not have
macular-threatening traction at baseline macular-threatening traction at baseline
Risk of glaucoma Low rates of substantial increases in intraocular pressure, Low rates of substantial increases in intraocular pressure,
need for ocular anti-hypertensives, or glaucoma need for ocular antihypertensives, or glaucoma
surgery surgery; ciliochoroidal effusion and angle-closure
glaucoma are very rare, but possible associated adverse
events
Risk of systemic adverse Theoretical higher risk of thromboembolic events and No increased risk of thromboembolic or other systemic
events higher rates of these events in the Protocol S events
ranibizumab group have been reported, although
increased risk has not been demonstrated conclusively
in studies of intravitreal anti-VEGF for ocular
conditions
Cost effectiveness* Ranibizumab is cost effective versus PRP only if central- PRP is more cost effective than ranibizumab in eyes with
involved DME also is present PDR without central-involved DME
Additional considerations Not a good option for patients with major systemic Long durability of effect in regressing neovascularization
comorbidities in whom PDR may recur or worsen that can last decades; excellent option for treatment
substantially if they cannot comply with monthly of PDR for patients who have difficulty with frequent
follow-up visits. follow-up visits

DME ¼ diabetic macular edema; DRCR.net ¼ Diabetic Retinopathy Clinical Research Network; PDR ¼ proliferative diabetic retinopathy; PRP ¼
panretinal photocoagulation; VEGF ¼ vascular endothelial growth factor.
*At this time, based on current drug costs and available estimates of treatment effectiveness.

following reasons: nonclearing vitreous hemorrhage (n ¼ 6), an eye with PDR, with or without high-risk characteristics
retinal detachment (n ¼ 1), and vitreous hemorrhage with (but without macular-threatening traction at baseline), causes
retinal detachment (n ¼ 1).16 Note that eyes with traction traction detachments more frequently than PRP.18e20
detachment threatening the macula or for which intraocular Protocol S did not compare ranibizumab versus PRP in
surgery was anticipated were excluded from enrollment in eyes for which surgery for PDR already was planned. The
Protocol S as well as the DRCR.net Protocols I, J, and N. safety of anti-VEGF compared with no anti-VEGF for
Thus, investigators were asked not to enroll eyes judged to nonclearing vitreous hemorrhage for which vitrectomy is
be at significant risk for traction retinal detachment after planned is being evaluated in Protocol AB (available at
anti-VEGF therapy. However, the results from these studies https://clinicaltrials.gov/ct2/show/NCT02858076; accessed
do not support the hypothesis that anti-VEGF administered to June 4, 2018).

92
 Sun et al 
Anti-VEGF Treatment Algorithm for PDR

Table 3. Frequently Asked Questions in Management of Eyes with PDR, Management in Protocol S, and DRCR.Net Suggestions for
Management in a Clinical Setting

What are treatment considerations for  Central-involved DME: initiate treatment with anti-VEGF monotherapy and re-evaluate need for
determining first-line therapy for eyes with PRP once eye no longer requires anti-VEGF for DME.
PDR and  Nonecentral-involved DME: consider using anti-VEGF as first-line therapy since anti-VEGF
treatment reduces the rate of progression to central-involved DME compared with PRP.
 Central-involved DME?  No DME: anti-VEGF and PRP are both reasonable alternatives. Anti-VEGF results in greater
 Nonecentral-involved DME? average visual acuity over time, less visual field reduction, less need for vitrectomy and reduced rate of
 No DME? central-involved DME onset but is invasive and requires frequent, often monthly visits. PRP is a
noninvasive therapy without associated risk of endophthalmitis that can usually be delivered in 1-2
sessions with excellent durability of NV regression over subsequent decades although additional
treatment is often necessary.
How to manage eyes with PDR in patients Increased risk of thromboembolic events is present in patients treated with systemic anti-VEGF at doses
with major systemic comorbidities such as that are approximately 400 times those given via intraocular injection. Large or even moderate
myocardial infarction or stroke? increases in risk for these events have not been demonstrated in studies of eyes receiving intravitreal
anti-VEGF for ocular conditions such as AMD and DME. Nonetheless, a small increase in risk cannot
be ruled out due to the small numbers of thromboembolic events that occur in these studies. Consider
additional oral and/or written consent processes for patients with these conditions, especially when
they have experienced recent medical events, to review the potential increased risks vs. benefits.
Patients who have a more remote history of cardiac events or stroke and who are otherwise systemically stable
should understand the potential risks and benefits as well, although whether there is an increased risk for a
subsequent cardiac event or stroke with versus without subsequent anti-VEGF treatment remains unknown.
How to manage eyes being treated with Eyes receiving anti-VEGF for PDR that develop vitreous hemorrhage precluding evaluation of the retina
anti-VEGF for PDR that develop vitreous should continue receiving anti-VEGF on a monthly basis until the hemorrhage clears and the retinal
hemorrhage? neovascular status can be adequately assessed. While anti-VEGF treatment is ongoing, regular
evaluation with fundus examination or ultrasonography should be performed to ensure that traction
retinal detachment, especially threatening the macula, is not developing or worsening.
How should NV status in eyes with PDR be Fluorescein angiography (FA) provides the most sensitive method for detecting retinal NV and assessing
evaluated? neovascular status. However, FA, including ultra-widefield imaging, was not routinely implemented or
required in Protocol S. Dilated fundus examination was performed at all follow-up visits for the
evaluation of NV. Fundus photographs were acquired at annual visits for study data purposes.
Additional imaging procedures at non-annual visits were left to investigator discretion. Wide-field
angiography ultimately may provide the most sensitive information regarding NV status, but its
relationship to the benefits of retreatment in the absence of any other evidence of NV worsening
remains unknown. The excellent results obtained in Protocol S in the absence of such imaging should
be recognized. The potential utility of ultra-widefield fundus or FA imaging is being investigated in the
ongoing DRCR.net Protocol AA.
Does anti-VEGF lead to more traction DRCR.net studies, including Protocols I, J, N, and S have not demonstrated any increased risk of traction
retinal detachments in eyes with PDR? retinal detachments in eyes with PDR that were treated with anti-VEGF. However, it is important to
note that all these studies excluded eyes with pre-existing traction retinal detachment involving the
macula, and that investigators may have elected not to enroll other eyes judged to be at high risk for
increasing retinal traction with or without anti-VEGF therapy.
Are there differences in outcomes between Since Protocol S was implemented before the FDA approval of 0.3-mg ranibizumab, and since the
the 0.3- and 0.5-mg doses of ranibizumab potential benefits of ranibizumab to treat PDR when used in a PRN regimen after 6 months in Protocol
for PDR? I21 utilized the 0.5-mg dose, the same dosage was chosen for Protocol S. The study does not address
whether there would be a difference in outcomes between the 0.3- and 0.5-mg doses when utilized for
PDR treatment. Nonetheless, we expect that similar results will be obtained with these 2 doses owing
to the near total blockade of VEGF with the 0.3-mg dose, the similar effects seen on diabetic
retinopathy in RIDE/RISE22 with monthly 0.3-mg compared with 0.5-mg ranibizumab, and the rapid
regression of retinal NV that has been documented in response to a much smaller molar dose
equivalent of bevacizumab.23
Are there differences in outcomes between Only 0.5-mg ranibizumab was evaluated in Protocol S and only 2.0-mg aflibercept was evaluated in
different anti-VEGF agents for treatment CLARITY.6 Therefore, it is unknown exactly how these 2 anti-VEGF agents would compare in terms
of PDR? of efficacy and safety for the treatment of PDR, nor how 1.25-mg repackaged (compounded)
bevacizumab would do. Data from the DRCR.net Protocol T demonstrated that differences in efficacy
can exist between anti-VEGF agents when used as treatment for central-involved DME. However,
differences, if they do exist, in outcomes between anti-VEGF agents in the treatment of PDR may not
be substantial given the high potency of all currently available anti-VEGF agents and documented
regression of retinal NV with a bevacizumab dose as low as 1.25 mg.23

AMD ¼ age-related macular degeneration; DME ¼ diabetic macular edema; DRCR.net ¼ Diabetic Retinopathy Clinical Research Network; FA ¼
fluorescein angiography; FDA ¼ Food and Drug Administration; PDR ¼ proliferative diabetic retinopathy; NV ¼ neovascularization; PRP ¼ panretinal
photocoagulation; VEGF ¼ vascular endothelial growth factor.

93
 Ophthalmology Volume 126, Number 1, January 2019

Additional Considerations diabetes over 6 years from diagnosis. Diabetologia.

2001;44(2):156e163.
Why 6 monthly injections initially? The DRCR.net recom- 3. The Diabetic Retinopathy Study Research Group. Photocoag-
mends 6 initial injections to maximize the number of eyes ulation treatment of proliferative diabetic retinopathy. Clinical
that attain early resolution of neovascularization. When application of Diabetic Retinopathy Study (DRS) findings, DRS
using the Protocol S injection algorithm, more than 50% of report number 8. Ophthalmology. 1981;88(7):583e600.
eyes showed resolution of neovascularization by 6 months. 4. Writing Committee for the Diabetic Retinopathy Clinical
Furthermore, an increasing proportion of eyes showed res- Research Network, Gross JG, Glassman AR, et al. Panretinal
olution of neovascularization after the fourth, fifth, and sixth photocoagulation vs intravitreous ranibizumab for proliferative
injections (Fig 3). It is unknown how results would differ if diabetic retinopathy: a randomized clinical trial. JAMA.
2015;314(20):2137e2146.
injections were stopped before 6 months. It is also unknown 5. The Diabetic Retinopathy Study Research Group. Preliminary
whether withholding injections on first neovascularization report on effects of photocoagulation therapy. Am J Oph-
resolution would be an effective strategy with fewer thalmol. 1976;81(4):383e396.
injections. 6. Sivaprasad S, Prevost AT, Vasconcelos JC, et al. Clinical ef-
Is it safe to withhold anti-VEGF therapy after 6 months if ficacy of intravitreal aflibercept versus panretinal photocoag-
neovascularization is stable but present? Protocol S data ulation for best corrected visual acuity in patients with
suggest that it is safe to withhold anti-VEGF therapy after 6 proliferative diabetic retinopathy at 52 weeks (CLARITY): a
months as soon as neovascularization is stable, provided at multicentre, single-blinded, randomised, controlled, phase 2b,
least 3 monthly injections are administered if neo- non-inferiority trial. Lancet. 2017;389(10085):2193e2203.
vascularization worsens after withholding anti-VEGF ther- 7. Gross JG, Glassman AR. A novel treatment for proliferative
diabetic retinopathy: anti-vascular endothelial growth factor
apy and that patients comply with the follow-up schedule to
therapy. JAMA Ophthalmol. 2016;134(1):13e14.
monitor neovascularization. When withholding anti-VEGF 8. Diabetic Retinopathy Clinical Research Network, Writing C,
therapy after sustained stability, there is only a small Aiello LP, et al. Rationale for the diabetic retinopathy clinical
chance of visual acuity loss, need for PRP, or need for research network treatment protocol for center-involved dia-
vitrectomy. In addition, this chance likely is smaller than if betic macular edema. Ophthalmology. 2011;118(12):e5ee14.
PRP were used to treat the PDR. Among eyes in which 9. World Medical Association. World Medical Association
injections were withheld at least once, only 4 (7%) lost 15 Declaration of Helsinki: ethical principles for medical research
letters or more from baseline at 2 years. involving human subjects. JAMA. 2013;310(20):2191e2194.
10. Diabetic Retinopathy Clinical Research Network, Wells JA,
Additional Clinical Scenarios Glassman AR, et al. Aflibercept, bevacizumab, or ranibizumab
for diabetic macular edema. N Engl J Med. 2015;372(13):
In addition to the situations discussed above, Table 3 1193e1203.
presents several frequently asked questions in the 11. Bressler SB, Qin H, Melia M, et al. Exploratory analysis of the
management of eyes with PDR. Each question is effect of intravitreal ranibizumab or triamcinolone on wors-
accompanied by a discussion of how this issue was ening of diabetic retinopathy in a randomized clinical trial.
managed in Protocol S or how the question might be JAMA Ophthalmol. 2013;131(8):1033e1040.
managed in clinical practice based on the experience of 12. Brown DM, Schmidt-Erfurth U, Do DV, et al. Intravitreal
aflibercept for diabetic macular edema: 100-week results from
DRCR.net investigators with similar cases in this trial.
the VISTA and VIVID studies. Ophthalmology. 2015;122(10):
In conclusion, the results of Protocol S demonstrate that 2044e2052.
ranibizumab therapy over 2 years is an effective alternative 13. Nguyen QD, Brown DM, Marcus DM, et al. Ranibizumab for
to PRP for PDR treatment, warranting a detailed under- diabetic macular edema: results from 2 phase III randomized
standing of the DRCR.net anti-VEGF treatment algorithm trials: RISE and RIDE. Ophthalmology. 2012;119(4):789e801.
for PDR. When choosing between anti-VEGF and PRP as 14. Bressler SB, Liu D, Glassman AR, et al. Change in diabetic
first-line therapy, treatment decisions for individual eyes retinopathy through 2 years: secondary analysis of a random-
with PDR should be guided by careful consideration of the ized clinical trial comparing aflibercept, bevacizumab, and
relative advantages of each treatment and anticipated patient ranibizumab. JAMA Ophthalmol. 2017;135(6):558e568.
compliance with follow-up and treatment recommendations. 15. Ross EL, Hutton DW, Stein JD, et al. Cost-effectiveness of
Application of the DRCR.net anti-VEGF treatment algo- aflibercept, bevacizumab, and ranibizumab for diabetic mac-
ular edema treatment: analysis from the diabetic retinopathy
rithm for PDR can provide excellent clinical outcomes
clinical research network comparative effectiveness trial.
through 2 years if compliance is good. Data through 5 years JAMA Ophthalmol. 2016;134(8):888e896.
of follow-up are now available.24 16. Bressler SB, Beaulieu WT, Glassman AR, et al. Factors
associated with worsening proliferative diabetic retinopathy in
References eyes treated with panretinal photocoagulation or ranibizumab.
Ophthalmology. 2017;124(4):431e439.
17. Bressler S, Beaulieu WT, Glassman A, et al. Photocoagulation
1. Klein R, Klein BE, Moss SE, et al. The Wisconsin epidemi- versus ranibizumab for proliferative diabetic retinopathy: Should
ologic study of diabetic retinopathy. III. Prevalence and risk of baseline characteristics affect choice of treatment? Submitted 2018.
diabetic retinopathy when age at diagnosis is 30 or more years. 18. Diabetic Retinopathy Clinical Research N, Elman MJ, Aiello LP,
Arch Ophthalmol. 1984;102:527e532. et al. Randomized trial evaluating ranibizumab plus prompt or
2. Stratton IM, Kohner EM, Aldington SJ, et al. UKPDS 50: risk deferred laser or triamcinolone plus prompt laser for diabetic
factors for incidence and progression of retinopathy in type II macular edema. Ophthalmology. 2010;117(6):1064e1077.e35.

94
 Sun et al 
Anti-VEGF Treatment Algorithm for PDR

19. Diabetic Retinopathy Clinical Research Network, Googe J, focal/grid photocoagulation for diabetic macular edema.
Brucker AJ, et al. Randomized trial evaluating short-term Ophthalmology. 2008;115(9):1447e1449, 9 e1-10.
effects of intravitreal ranibizumab or triamcinolone acetonide 22. Brown DM, Nguyen QD, Marcus DM, et al. Long-term out-
on macular edema after focal/grid laser for diabetic macular comes of ranibizumab therapy for diabetic macular edema: the
edema in eyes also receiving panretinal photocoagulation. 36-month results from two phase III trials: RISE and RIDE.
Retina. 2011;31(6):1009e1027. Ophthalmology. 2013;120(10):2013e2022.
20. Diabetic Retinopathy Clinical Research Network. Ran- 23. Avery RL. Regression of retinal and iris neovascularization
domized clinical trial evaluating intravitreal ranibizumab after intravitreal bevacizumab (avastin) treatment. Retina.
or saline for vitreous hemorrhage from proliferative 2006;26(3):352e357.
diabetic retinopathy. JAMA Ophthalmol. 2013;131(3): 24. Gross JG, Glassman AR, Liu D, et al. Five-year outcomes
283e293. of panretinal photocoagulation vs intravitreous ranibizumab
21. Diabetic Retinopathy Clinical Research Network. A random- for proliferative diabetic retinopathy: a randomized clinical
ized trial comparing intravitreal triamcinolone acetonide and trial. JAMA Ophthalmol. 2018;136(10):1138e1148.

Footnotes and Financial Disclosures

Originally received: June 5, 2018. study and review of the manuscript, but not directly in the design or conduct
Final revision: July 30, 2018. of the study, nor in the collection, management, analysis, or interpretation
Accepted: August 1, 2018. of the data; preparation, review, or approval of the manuscript; nor decision
Available online: August 7, 2018. Manuscript no. 2018-1291. to submit the manuscript for publication. Per the DRCR.net Industry
1
Joslin Diabetes Center, Beetham Eye Institute, Department of Ophthal- Collaboration Guidelines (available at http://www.drcr.net), the DRCR.net
mology, Harvard University, Boston, Massachusetts. had complete control over the design of the protocol, ownership of the data,
2
Jaeb Center for Health Research, Tampa, Florida. and all editorial content of presentations and publications related to the
3 protocol.
Wilmer Eye Institute, Johns Hopkins University School of Medicine,
Baltimore, Maryland. A complete list of the Diabetic Retinopathy Clinical Research Network
4 investigators who participated in this trial is available in JAMA.
Casey Eye Institute, Portland, Oregon. 2015;314(20):2137e2146.
5
Carolina Retina Center, PA, Columbia, South Carolina. HUMAN SUBJECTS: Human subjects were included in this study. The list
6
Texas Retina Associates, Lubbock, Texas. of IRBs was provided at submission. All research adhered to the tenets of
7
Feinberg School of Medicine, Northwestern University, Chicago, Illinois. the Declaration of Helsinki. All participants provided informed consent.
Presented at: 36th American Society of Retina Specialsts Annual Meeting, No animal subjects were included in this study.
July 20-25, 2018 in Vancouver, BC, Canada; Association for Research in Author Contributions:
Vision and Ophthalmology Annual Meeting, 2018; and the Macula Society Conception and design: Sun, Glassman, Beaulieu, Stockdale, Bressler,
Annual Meeting, Beverly Hills, California, February 21-24, 2018.
Flaxel, Gross, Shami, Jampol
Financial Disclosure(s):
Analysis and interpretation: Sun, Glassman, Beaulieu, Stockdale, Bressler,
The author(s) have made the following disclosure(s): J.K.S.: Financial Flaxel, Gross, Shami, Jampol
support e Optovue, Boston Micromachines, Genentech, Eleven Bio-
therapeutics, Vindico Medical Education, Current Diabetes Reports, Kal- Data collection: Sun, Glassman, Beaulieu, Stockdale, Bressler, Flaxel,
vista, Merck, Allergan, Kowa, Novartis, Regeneron, Bayer, Novo/Nordisk, Gross, Shami, Jampol
Adaptive Sensory, Technologies, Boehringer Ingelheim. Obtained funding: Glassman, Bressler, Jampol, Sun
A.R.G.: Financial support (to institution) e Genentech, Regeneron. Overall responsibility: Sun, Glassman, Beaulieu, Stockdale, Bressler,
Flaxel, Gross, Shami, Jampol
W.T.B.: Financial support (to institution) e Genentech, Regeneron.
C.R.S.: Financial support (to institution) e Genentech, Regeneron. Abbreviations and Acronyms:
N.M.B.: Financial support (to institution) e Bayer, Novartis, Roche CI-DME ¼ central-involved diabetic macular edema;
(Genentech), Samsung Bioepis. CLARITY ¼ Clinical efficacy and mechanistic evaluation of aflibercept
for proliferative diabetic retinopathy; DME ¼ diabetic macular edema;
J.G.G.: Financial support e Genentech, Regeneron, Acucela, Ohr. DRCR.net ¼ Diabetic Retinopathy Clinical Research Network;
Supported through a cooperative agreement from the National Eye Institute IQR ¼ interquartile range; PDR ¼ proliferative diabetic retinopathy;
and the National Institute of Diabetes and Digestive and Kidney Diseases, PRP ¼ panretinal photocoagulation; VEGF ¼ vascular endothelial growth
National Institutes of Health, Bethesda, Maryland (grant nos.: EY14231 factor.
[A.R.G.], EY23207 [L.M.J.], and EY18817 [N.M.B.]); and the Juvenile
Diabetes Research Foundation. Genentech provided ranibizumab for the Correspondence:
Wesley T. Beaulieu, PhD, Jaeb Center for Health Research, 15310 Amberly
study and funds to DRCR.net to defray the study’s clinical site costs. The
National Institutes of Health participated in oversight of the conduct of the Drive, Suite 350, Tampa, FL 33647. E-mail: [email protected].

95