Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`DOI 10.1186/s40942-015-0010-y
`
`REVIEW
`Endophthalmitis following intravitreal
`anti-vascular endothelial growth factor (VEGF)
`injection: a comprehensive review
`
`Open Access
`
`Rohan Merani1,2,3,4* and Alex P Hunyor1,2,3,5
`
`Abstract
`The purpose of this review is to report and summarize previously reported studies and assess many of the individual
`steps of the intravitreal injection procedure’s possible effect on the prevention of endophthalmitis. The pooled
`endophthalmitis rate from 20 large retrospective case series of anti-VEGF injections was 144/510,396 (0.028%;
`1/3,544). Injections may be performed in an office-based location or in an operating room (OR) and low rates of
`endophthalmitis can be achieved in either location with careful attention to asepsis. Pre- or post-injection topical
`antibiotics have not been shown to be effective, and could select for more virulent microorganisms. Povidone-iodine
`prior to injection is accepted as the gold-standard antiseptic agent, but aqueous chlorhexidine may be an alterna-
`tive. Antisepsis before and after gel or subconjunctival anesthetic is suggested. The preponderance of Streptococcal
`infections after intravitreal injection is discussed, including the possible role of aerosolization, which can be minimized
`by using face masks or maintaining silence. As with other invasive procedures in medicine, the use of sterile gloves,
`following adequate hand antisepsis, may be considered. Control of the eyelashes and lid margin is required to avoid
`contamination of the needle, but this can be achieved with or without a speculum. Techniques to minimize vitreous
`reflux have not been shown to reduce the risk of endophthalmitis. Same day bilateral injections should be performed
`as two separate procedures, preferably using drug from different lots, especially when using compounded drugs.
`Keywords: Endophthalmitis, Intravitreal injection, Anti-VEGF, Streptococcus, Masks, Antisepsis, Povidone-Iodine,
`Chlorhexidine, Antibiotics, Speculum
`
`Introduction
`Intravitreal injection (IVI) is the most commonly per-
`formed ophthalmic procedure. In the USA, the number
`of injections performed has increased exponentially, from
`4,215 injections in 2001 to 82,994 in 2004, to 812,413 in
`2007, to 1.27 million in 2009 and to 2.5 million injections
`in 2011 [1, 2]. Similar increases have been observed in
`Canada and the United Kingdom [3, 4].
`Infectious endophthalmitis (IE) secondary to IVI is a
`potentially devastating complication. It can be difficult
`to distinguish infectious endophthalmitis from “sterile”
`or non-infectious endophthalmitis. For the purpose of
`this review, IE refers to endophthalmitis that is clinically
`
`*Correspondence: rmerani@med.usyd.edu.au
`1 Retina Associates, Level 4, 8 Thomas St, Chatswood, NSW 2067, Australia
`Full list of author information is available at the end of the article
`
`suspected to be infectious, and treated as such with a vit-
`reous tap and injection of antibiotics and/or vitrectomy
`surgery.
`Bacteria are most likely inoculated into the vitreous
`cavity at the time of injection, or much less likely gain
`access later through the needle tract [5, 6]. The potential
`sources of bacteria include the patient’s ocular or perio-
`cular surfaces, aerosolized bacteria, or contamination of
`the needle, instruments, drug or drug vial [7].
`Two meta-analyses including both retrospective series
`and clinical trials have calculated the pooled rate of
`endophthalmitis after anti-VEGF injections. McCannel
`found a rate of 52/105,536 injections (0.049%; 1 in 2030)
`[8] and more recently, Fileta et al. [9] calculated a rate of
`197/350,535 (0.056%; 1 in 1,779). As patients typically
`receive ongoing intravitreal therapy, the per-patient risk
`
`© 2015 Merani and Hunyor. This article is distributed under the terms of the Creative Commons Attribution 4.0 International
`License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any
`medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons
`license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.
`org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
`
`Novartis Exhibit 2011.001
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 2 of 19
`
`of endophthalmitis is significantly higher than the per-
`injection risk.
`The rate of needle contamination after IVI has been
`reported as between 0.36 and 18%, which is significantly
`higher than the incidence of endophthalmitis after
`IVI [5, 7, 10]. The threshold inoculum size required to
`develop endophthalmitis is related to the type of bac-
`teria and their virulence, intraocular immune mecha-
`nisms and anatomical characteristics of the vitreous
`[11, 12]. Animal studies have shown that a smaller num-
`ber of bacterial colony-forming units are required to
`induce endophthalmitis when injected into the vitreous
`compared to when they are injected into the anterior
`chamber [13]. Endophthalmitis following intravitreal
`injection often presents earlier than after cataract sur-
`gery [14, 15].
`The purpose of this review is to estimate the rate of
`endophthalmitis after intravitreal injection and to exam-
`ine each step of the injection procedure that may influ-
`ence the risk of endophthalmitis. To be able to prove that
`a particular measure reduces the risk of endophthalmitis
`would need huge numbers of patients in a randomized
`controlled trial, given that endophthalmitis is a relatively
`rare outcome. There is thus no Level 1 evidence for any
`preventative measure to reduce the incidence of endoph-
`thalmitis after intravitreal injection. As a result, this
`review largely summarizes retrospective papers, with
`their inherent biases.
`
`Methods
`A systematic literature search of the Medline database
`from 1996 to December 2014 was performed through
`Ovid, using search terms relevant to each section. Fur-
`ther literature was sourced from the reference lists of
`retrieved publications.
`To estimate the per-injection rate of endophthalmitis
`after anti-VEGF injection, retrospective case series with
`at least 10,000 such injections were included. Studies
`that did not report a breakdown of the drugs used were
`excluded to avoid including triamcinolone and other
`injections in this calculation. Questionnaire-based and
`population-based studies were excluded given the incom-
`plete data. Clinical trials were excluded as they may not
`reflect real-world practice, with more stringent require-
`ments regarding injection technique often included in
`the protocols.
`
`Results
`Twenty retrospective case series meeting the inclusion
`criteria were identified. Details of the injection procedure
`and associated aseptic measures used in each series are
`
`summarized in the Table 1. Where data were missing, the
`corresponding author for each study was contacted by
`email. Only two authors were not contactable.
`We identified 144 cases of endophthalmitis from
`510,396 anti-VEGF injections which equates to a pooled
`endophthalmitis rate of 0.028% or 1 in 3,544 injections
`[16–33].
`
`Review
`Location—office vs operating room (OR)
`In the 2013 American Society of Retinal Specialists
`(ASRS) Preferences and Trends (PAT) Survey, over 98%
`of USA-based specialists reported performing injections
`in an office setting, compared with only 47% of inter-
`national specialists [34]. In Germany and other parts of
`Europe, more injections are performed in the operating
`room (OR) [35, 36].
`It has been [29] suggested that an advantage of the OR
`location is the superior air circulation systems. However,
`the ESCRS endophthalmitis study group was not able to
`find a relationship between the number of air changes per
`hour and the incidence of endophthalmitis after cataract
`surgery when they compared locations with minimal air-
`flow, 20 air changes per hour and ultraclean air systems
`using laminar flow principles [37, 38].
`Pooling the results of three OR-based injection series,
`the endophthalmitis rate was just 6/78,506 (0.0076% or 1
`in 13,084) [19, 23, 25]. Common to these studies was the
`careful attention to asepsis with the use of sterile gloves,
`face masks, and drapes which were not used in most
`other office-based series (see Table 1). A notable excep-
`tion is Shimada et al’s series with no cases of endophthal-
`mitis out of 15,144 injections where similar strict aseptic
`measures were followed in an office setting [27].
`Abell et  al. [29] reported an endophthalmitis rate of
`4/3,376 (0.12%) for office-based injections compared
`with 0/8,873 (0%) for OR-based injections. In this non-
`randomized series, patients with private health insurance
`were treated in the OR while those without insurance
`were treated in the office. The difference in endophthal-
`mitis rates may be a reflection of socioeconomic or other
`factors [39]. Tabendeh et al. [30] reported an endophthal-
`mitis rate of 3/8,210 (0.037%) anti-VEGF injections in
`the office compared with 2/3,047 in the operating room
`(0.066%), in another non-randomised study that was not
`powered to be able to detect a difference. Compared with
`office-based injections, there was no apparent benefit to
`an OR environment in this small study.
`Although there is no doubt that the OR has many advan-
`tages, there are logistical hurdles that make access to OR
`facilities difficult for many patients, and the OR location
`
`Novartis Exhibit 2011.002
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 3 of 19
`
`Office
`
`?
`
`Variable:
`
`?
`
`5–10%
`
`?
`
`?
`
`?
`
`mitis
`endophthal-
`cases with
`of the 15
`used in 14
`
`OR
`
`Sterile
`
`Yes
`
`Drops
`
`5–10%
`
`Yes (adhesive)
`
`Yes
`
`other)
`no in the
`one hospital,
`Variable (yes in
`
`Office
`
`Nil or non-
`
`sterile
`
`Yes
`
`Subconj
`Gel
`Drops
`
`5%
`
`No
`
`No
`
`Yes
`
`Office
`
`Non-sterile
`
`Yes
`
`Drops
`
`5%
`
`No
`
`No*
`
`Variable
`
`No
`
`Office
`
`Nil*
`
`Variable
`
`Subconj (rarely)
`Drops
`
`5%
`
`No
`
`No*
`
`Variable
`
`Variable
`
`Location
`
`Gloves
`
`speculum
`Sterile lid
`
`agents used
`Anaesthetic
`
`tration
`iodine concen-
`povidone-
`Conjunctival
`
`Drape
`
`Mask
`
`antibiotics
`Post injection
`
`antibiotics
`injection
`Pre-
`
`Variable
`
`Bev: 5/8,039 (0.062%;
`
`(0.044%; 1/2,258)
`
`15/30,736 (0.049%;
`Overall
`Peg: 0/6
`Aflib:0/89
`Bev: 0/3,518
`
`Rani: 10/22,579
`By Agent
`1/2,049)
`
`(0.0082%, 1/12,133)
`
`Rani: 3/36,398
`By Agent
`in 13,337)
`
`Peg: 0/128
`1/1,608)
`
`No
`
`3/40,011 (0.0075%;1
`Overall
`
`Yes
`
`17/49,002 (0.035%; 1
`Overall
`Peg: 0/2,015
`
`(0.018%; 1/5,671)
`
`(0.027%; 1/3,721)
`
`Bev: 7/39,700
`
`44/117,171 (0.038%;
`Overall
`
`12/60,322 (0.020%;
`Overall
`Aflib: 0/1,373
`
`(0.045%;1/2,200)
`
`Bev: 20/44,007
`
`(0.033%; 1/2,991)
`
`Rani: 24/71,791
`By Agent
`1/2,663)
`
`Rani: 5/18,607
`By Agent
`1/5,027)
`
`(0.0099%;1/10,149)
`
`(0.052%;1/1,914)
`
`Bev: 15/28,705
`
`Rani: 2/20,297
`By Agent
`in 2,882)
`
`30,736
`
`Ophthal 2009
`Klein et al. [20]
`
`Multi-centre,
`31 Jul 2007
`1 Aug 2006 to
`
`USA
`
`40,011
`
`Casparis et al.
`
`[19]
`
`Retina 2014
`
`Switzerland
`Multicentre,
`2004 to 2012
`
`49,002
`
`Retina 2013
`et al. [18]
`Chaudhary
`
`Single-centre,
`31 Dec 2011
`1 Jan 2007 to
`
`USA
`
`60,322
`
`Retina 2011
`et al. [17]
`Moshfeghi
`
`USA
`(multi-site),
`Single-centre
`31 Dec 2010
`1 Jan 2005 to
`
`117,171
`
`Ophthal 2013
`
`Storey et al.
`
`group)
`study
`[16] (PIE
`
`1 Jan 2009 to 1
`
`Single-centre,
`
`Oct 2012
`
`USA
`
`Table 1 Endophthalmitis following intravitreal anti-VEGF injection—retrospective cases series with at least 10,000 injection
`
`suspected IE
`Rate of clinically
`
`injections
`n = 
`
`Authors
`
`and location
`Period
`
`Novartis Exhibit 2011.003
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 4 of 19
`
`OR
`
`Sterile
`
`Yes
`
`Drops
`
`1%
`
`adhesive)*
`
`Yes (non-
`
`Yes
`
`Yes
`
`Office
`
`Non-sterile
`
`Yes
`
`Drops
`
`5% (before and
`
`No
`
`No
`
`No
`
`tion)
`after injec-
`
`OR
`
`Sterile
`
`Yes
`
`Drops
`
`5%
`
`Yes (adhesive)
`
`Yes
`
`Variable
`
`Office
`
`Nil*
`
`Variable (used
`
`mitis)
`endophthal-
`cases of
`in all the
`
`minority
`eyelids in a
`spread the
`
`Fingers to
`
`Office
`
`Non-sterile*
`
`Yes
`
`Subconj
`Gel
`Drops
`
`Subconj
`Gel
`Drops
`
`Location
`
`Gloves
`
`speculum
`Sterile lid
`
`agents used
`Anaesthetic
`
`5–10%
`
`No*
`
`No*
`
`Variable (used
`
`mitis)
`endophthal-
`cases of
`in all the
`
`5–10%
`
`No*
`
`No*
`
`Yes
`
`Yes
`
`tration
`iodine concen-
`povidone-
`Conjunctival
`
`Drape
`
`Mask
`
`antibiotics
`Post injection
`
`antibiotics
`injection
`Pre-
`
`No
`
`No
`
`No
`
`No*
`
`(0.0065%, 1/15,479)
`
`(0.010%, 1/10,097)
`
`(0.025%;1/3,932)
`
`Bev: 2/7,865
`
`Peg: 0/240
`
`3/18,202 (0.016%;
`Overall
`Peg: 0/370
`Aflib: 0/148
`
`Rani: 1/10,097
`By Agent
`1/6,067)
`
`1/17,666 (0.0057%,
`Overall
`Aflib: 0/269
`Rani: 0/20,024
`By Agent
`0/20,293
`Overall
`
`Bev: 1/15,479
`Rani: 0/1,669
`By Agent
`1/17,666)
`
`Bev: 3/6,675 (0.045%;
`
`(0.036%; 1/2,792)
`
`11/29,995 (0.037%;
`Overall
`
`Rani: 8/22,336
`By Agent
`1/2,727)
`
`(0.021%; 1/4,773)
`
`Bev: 3/12,585
`
`(0.024%; 1/4,195)
`
`6/26,905 (0.022%;
`Overall
`Peg: 0/984
`1/2,225)
`
`Rani: 3/14,320
`By Agent
`1/4,484)
`
`18,202
`
`Retina 2014
`et al. [25]
`Nentwich
`
`Single-centre,
`
`Germany
`
`2012
`end July
`Jan 2005 to
`
`17,666
`
`Sandler [24]
`Bhavsar and
`
`Retina 2015
`
`Single-centre,
`31 Oct 2012
`1 Aug 1997 to
`
`USA
`
`20,293
`
`Brynskov et al.
`
`[23]
`
`Retina 2014
`
`March 2007 to
`
`Single-centre,
`
`May 2013
`
`Denmark
`
`26,905
`
`Fintak et al.
`
`Retina 2008
`
`[22]
`
`Multicentre,
`7 Aug 2007
`1 Jun 2005 to
`
`USA
`
`29,995
`
`Chen et al.
`
`Retina 2011
`
`[21]
`
`USA
`(multi-site),
`Single-centre
`
`July 2010
`July 2000 to
`
`suspected IE
`Rate of clinically
`
`injections
`n = 
`
`Authors
`
`and location
`Period
`
`Table 1 continued
`
`Novartis Exhibit 2011.004
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 5 of 19
`
`OR = 2/3,063
`Office = 3/8,647
`
`Non-sterile in
`Sterile in OR
`
`office
`
`Yes
`
`Subconj
`Drops
`
`5%
`
`No in office
`Yes in OR
`
`No in office
`Yes in OR
`
`Yes
`
`office
`
`OR = 0/8,873
`Office = 4/3,376
`
`Sterile
`
`Yes
`
`Drops + Gel*
`
`10%
`
`adhesive)*
`
`Yes (non-
`
`Yes
`
`until 2011)*
`Variable (used
`
`Office
`
`Nil*
`
`Yes
`
`Gel
`Drops
`
`10%
`
`No
`
`No*
`
`Variable
`
`Office
`
`Yes
`
`Yes
`
`conj
`Drops + Sub-
`
`0.25% (before
`
`Yes (adhesive)
`
`Yes
`
`Yes
`
`injection)
`and after
`
`Office*
`
`Sterile*
`
`Yes*
`
`Drops*
`
`2.5%*
`
`Variable
`
`Yes*
`
`Yes*
`
`No*
`
`Location
`
`Gloves
`
`speculum
`Sterile lid
`
`agents used
`Anaesthetic
`
`tration
`iodine concen-
`povidone-
`Conjunctival
`
`Drape
`
`Mask
`
`antibiotics
`Post injection
`
`antibiotics
`injection
`Pre-
`
`No in
`Yes in OR
`
`No
`
`Rani: 3/2,724 (0.11%;
`By Agent
`5/11,257
`Overall
`
`Bev: 1/1,675 (0.060%;
`
`(0.028%; 1/3,525)*
`
`1/1,675)*
`
`Bev: 2/8,533 (0.023%;
`
`1/4,267)
`
`1/908)
`
`4/12,249 (0.033%;
`Overall
`Peg: 0/121
`1/1,347).
`
`Rani: 3/10,574
`By Agent
`1/3,062)
`
`Rani: 3/9,453 (0.032%;
`By Agent
`1/2,137)
`
`Bev: 4/5,386 (0.074%;
`
`1/3,151)
`
`Variable
`
`Yes
`
`(0.050%;1/1,991)
`
`Bev: 7/15,035
`(0.14%; 1/705)
`Rani: 1/705
`By Agent
`
`(0.044%;1/2,275)
`
`7/14,960 (0.047%;
`Overall
`Peg: 0/548
`Bev: 0/846
`Rani: 0/13,750
`By Agent
`0/15,144
`Overall
`Peg: 0/185
`
`8/15,925
`Overall
`
`11,257
`
`Retina 2014
`et al. [30]
`Tabandeh
`
`12,249
`
`BJO 2012
`Abell et al. [29]
`
`Multi-centre,
`
`Dec 2011
`Jan 2009 to
`
`Italy
`USA and
`
`Multi-centre
`
`Mar 2012
`Mar 2006 to
`
`Australia
`surgeon),
`(single-
`
`14,960
`
`Cheung et al.
`
`Ophthal 2012
`
`[28]
`
`Multi-centre,
`
`Aug 2010
`Jan 2005 to
`
`Canada
`
`15,144
`
`Shimada et al.
`
`Graefes 2013
`
`[27]
`
`Single-centre,
`
`July 2012
`July 2009 to
`
`Japan
`
`15,925
`
`Mithal et al.
`
`BJO 2013
`
`[26]
`
`Multicentre,
`May 2012
`Jan 2007 to
`
`India
`
`suspected IE
`Rate of clinically
`
`injections
`n = 
`
`Authors
`
`and location
`Period
`
`Table 1 continued
`
`Novartis Exhibit 2011.005
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 6 of 19
`
`? These points were not mentioned in the manuscript, and the corresponding authors could not be contacted.
`* These points were not explicitly mentioned in the manuscript, but were clarified through personal communication with the corresponding author of the paper.
`Rani ranibizumab, Bev bevacizumab, Aflib aflibercept and Peg pegaptanib.
`Location: “office-based” includes studies where the injections were performed in a clean procedure room, or in the consulting room itself, within an office or outpatients department.
`Anesthetic: “Drops” refers to drops given alone or with the aid of a cotton bud or pledget.
`Steroid and other injections have been excluded.
`
`Office*
`
`Variable*
`
`Variable
`
`Subconj.
`Drops
`
`Yes, 5%
`
`Variable
`
`Variable*
`
`Variable
`
`Variable*
`
`Office*
`
`Nil or non-
`
`sterile*
`
`not used”
`attire was
`
`No
`
`Drops
`
`5%
`
`No*
`
`No*
`
`Variable (used
`
`No*
`
`2,011)
`until Oct
`
`Office
`
`? “Surgical
`
`Variable
`
`Drops
`
`5%
`
`No
`
`No
`
`Yes
`
`No
`
`Bev: 1/3,501 (0.029%;
`(0.016%; 1/6,347)
`Rani: 1/6,347
`By Agent
`1/5,127)
`
`2/10,254 (0.020%;
`Overall
`
`3/10,140 (0.030%,
`Overall
`
`Bev: 2/3,834 (0.052%;
`
`(0.016%;1/6,330)
`
`1/1,917)
`
`Rani: 1/6,330
`By Agent
`(0.030%; 1/3,388)
`3/10,164
`Overall
`Peg: 0/406
`1/3,501)
`
`Rani: 3/7,768 (0.039%;
`By Agent
`1/3,380)
`
`Peg: 0/57
`Bev: 0/2,315
`
`1/2,589)
`
`10,140
`
`et al. [33]
`Englander
`
`BJO 2013
`
`Single-centre,
`31 Dec 2011
`1 Jan 2007 to
`
`USA
`
`10,164
`
`Fineman et al.
`
`Retina 2013
`
`[32]
`
`Single-centre,
`
`Dec 2011
`Nov 2010 to
`
`USA
`
`10,254
`
`AJO 2008
`Pilli et al. [31]
`
`USA
`(multi-site),
`Single-centre
`18 Oct 2007
`5 Jan 2005 to
`
`Location
`
`Gloves
`
`speculum
`Sterile lid
`
`agents used
`Anaesthetic
`
`tration
`iodine concen-
`povidone-
`Conjunctival
`
`Drape
`
`Mask
`
`antibiotics
`Post injection
`
`antibiotics
`injection
`Pre-
`
`suspected IE
`Rate of clinically
`
`injections
`n = 
`
`Authors
`
`and location
`Period
`
`Table 1 continued
`
`Novartis Exhibit 2011.006
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 7 of 19
`
`can add substantial additional cost to patients and the
`healthcare system. It has been suggested that simply being
`in an OR may alter behavior with more careful attention to
`asepsis [23]. While three series describing OR-based injec-
`tions have reported low rates of endophthalmitis, this may
`reflect publication bias. Similar low rates can be achieved
`with strict asepsis in an office setting.
`
`Hand antisepsis
`The aim of surgical hand antisepsis is to reduce the bacte-
`rial load at the commencement of a procedure. Broadly
`speaking there are two main types of antisepsis solutions:
`aqueous scrubs (povidone iodine, chlorhexidine and tri-
`closan) and liquid or gel alcohol rubs (with or without
`additional ingredients) [40].
`Alcohol-based rubs have been found to have superior
`antimicrobial efficacy compared with aqueous scrubs [40,
`41]. The reduction in microbial counts with alcohol rubs
`is rapid, and inhibition of bacterial regrowth to baseline
`levels can take more than 6 h [42]. However, unlike chlo-
`rhexidine, alcohol does not bind to the skin imparting a
`true residual effect, so chlorhexidine or other agents are
`often added to alcohol rubs [40].
`While chlorhexidine induces less allergic reactions
`than povidone-iodine, skin irritation, dryness and irritant
`contact dermatitis still occurs more frequently with chlo-
`rhexidine scrubbing than with alcohol-based rubs [40,
`43]. Liquid alcohol-based rubs are superior to gels both
`in terms of tolerability and efficacy [44].
`Alcohol-based rubs are not without limitations. They do
`not remove surface dirt because they do not contain sur-
`factants or have a foaming action, and may have limited
`effectiveness if the hands are heavily contaminated [40,
`45]. Rubs may also leave a residue on the hands after use.
`There is no specific evidence regarding the role of hand
`antisepsis in the context of intravitreal injection and
`most of the retrospective series do not even mention if
`or how hand antisepsis was performed. If sterile gloves
`are employed, the need for hand antisepsis could be
`questioned given the low risk of glove perforation during
`procedures of very short duration. In our opinion, atten-
`tion to hand antisepsis is important for all invasive pro-
`cedures in medicine. The initial antisepsis at the start of
`an injecting session should ideally include hand washing
`with soap or an aqueous scrub to mechanically remove
`any surface dirt or heavy bacterial contamination, espe-
`cially if gloves are not worn [40, 46]. For subsequent
`antisepsis, alcohol-based rubs are ideal given their rapid
`action and superior dermal tolerance.
`
`Gloves
`The purpose of wearing gloves during invasive proce-
`dures is to protect both the patient and the surgeon [47].
`
`In a survey of retinal specialists in the USA, only 254/762
`(33%) reported wearing sterile gloves for intravitreal
`injections, while 323/762 (42%) did not wear gloves at all
`[48]. In another smaller survey of 158 retinal specialists,
`46% reported that they do not wear gloves [49]. In con-
`trast, 88% of 112 surveyed retinal specialists in the United
`Kingdom reported using sterile gloves [50]. If non-sterile
`gloves are used, perforations may be more common with
`vinyl compared with latex gloves [51–53]. Non-sterile
`gloves may be more prone to fungal contamination com-
`pared with individually sealed sterile gloves [54].
`It has been argued that sterile gloves are not required
`as long as the tip of any instrument touching the eye
`remains sterile [48, 55, 56]. By definition, this is no longer
`“aseptic technique”, a key principle of which is that any
`part of an instrument if touched directly or indirectly
`could result in infection [57]. Sterile gloves are required
`for aseptic procedures, while non-sterile gloves suffice
`for clean procedures [57]. A “no-touch” technique with-
`out gloves at all was advocated for cataract surgery over
`50 years ago, but has fallen out of favor [58].
`Wound infection rates have been shown to be no
`higher with the use of non-sterile compared with sterile
`gloves when suturing the skin [59, 60] However, ster-
`ile gloves are recommended for the insertion of central
`venous catheters and spinal anesthesia procedures [51,
`61, 62]. Like the CSF, the vitreous is an immune-privi-
`leged site and a small inoculum of low virulence bacteria
`may be sufficient to cause endophthalmitis.
`There are no studies directly examining the role of ster-
`ile gloves in reducing the risk of post-injection endoph-
`thalmitis, and they are often used in conjunction with
`other aseptic measures such as face masks and sterile
`drapes. We believe that intravitreal injection should be
`regarded as an aseptic procedure given that it involves
`penetration into an immune-privileged, nutrient-rich
`cavity. Similar to other aseptic procedures in medicine,
`the use of sterile gloves should be considered.
`
`Masks
`The main purpose of wearing a surgical face mask is to
`reduce bacterial contamination of the surgical field from
`the surgeon’s mouth or nasopharynx [63]. In the 2013
`ASRS PAT survey, just 14% of US-based ophthalmolo-
`gists reported wearing a mask and asking the patient not
`to speak [34].
`In his meta-analysis of 105,536 injections, McCannel
`found that eight of the 26 culture-positive cases (31%)
`were due to Streptococcus and noted that this was 3-fold
`higher than earlier studies of endophthalmitis after cat-
`aract surgery [8]. Others have found similar results and
`have highlighted the poor visual outcomes associated
`with this virulent pathogen, with an increased likelihood
`
`Novartis Exhibit 2011.007
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 8 of 19
`
`of a final VA of counting fingers or less and an increased
`likelihood of enucleation [14, 16, 17, 21, 64].
`It is thought that the preponderance of Streptococ-
`cal isolates may relate to droplet dispersal of organ-
`isms while performing intravitreal injections. Viridans
`group Streptococcus species are normal commensals
`of the upper respiratory tract and oral cavity [65, 66].
`As they are uncommonly found as part of the normal
`conjunctival flora, it has been suggested that they may
`contaminate the conjunctival surface or needle via aero-
`solization, leading to endophthalmitis [8, 21]. In studies
`where conjunctival cultures were taken in treatment-
`naïve eyes, the most commonly cultured organisms
`were coagulase negative Staphylococci accounting for
`65–83% of isolates, while 0–7% of isolates were Strepto-
`cocci [6, 67–69].
`In an experimental study designed to simulate the
`conditions during intravitreal injections, Wen et al. [70]
`found that wearing a face mask or remaining silent sig-
`nificantly decreased culture plate contamination com-
`pared with not wearing a face mask or turning the face
`away. They also showed that a significant number of
`colonies grew when the reclined volunteer (simulating a
`patient) continued talking. While no Streptococcal spe-
`cies were isolated from the groups wearing a face mask
`or remaining silent, the proportion of bacterial colonies
`represented by oral Streptococcal species ranged from 67
`to 83% in the other groups.
`In a similar study, Doshi et al. [71] also found almost
`no bacterial growth when a mask was not worn and
`silence was not maintained, if the agar plates were pre-
`treated with povidone-iodine (PI). In practice however, if
`the conjunctival surface were to be contaminated imme-
`diately prior to needle entry, there may not be adequate
`time for the PI to take effect.
`Friedman et  al. [72] recently demonstrated no differ-
`ence in the needle contamination rates when speaking
`compared with maintaining silence.
`Oral commensals have been found in cases of iatro-
`genic meningitis following dural puncture procedures
`and in some cases have been molecularly matched with
`strains found in the oropharynx of the proceduralist [73–
`76]. Absence of a face mask has also been implicated in
`iatrogenic septic arthritis after intraarticular injections
`[77, 78]. In these cases there is a strong suggestion that
`airborne transmission of a proceduralist’s oropharyngeal
`flora onto a needle or patient’s skin is followed by inocu-
`lation into a sterile compartment.
`The presence of a viral upper respiratory tract infec-
`tion (URTI) has been shown to markedly increase the
`airborne dispersal of methicillin-resistant Staphylococ-
`cus aureus (MRSA), which can be prevented by wearing
`a mask [79].
`
`Several studies have shown that wearing a mask does
`not lower the risk of surgical wound infection [80–82].
`Wen et  al. argue that oral Streptococcal species are of
`relatively low virulence in the immunocompetent host,
`which is why they are infrequently found in surgical site
`infections. However the vitreous and cerebrospinal fluid
`are immune-privileged sites where these usually less vir-
`ulent strains can flourish [70, 77, 83]. Dural puncture is
`a procedure similar to intravitreal injection in that both
`involve needle penetration into a nutrient-rich body cav-
`ity that can serve as a culture medium [8].
`Of course, there may be other explanations for the pre-
`ponderance of Streptococcal infections after intravitreal
`injections. Delayed-onset bleb-related endophthalmitis
`is also associated with a disproportionately higher rate of
`Streptococcal infection. This may be the result of altera-
`tions in the resident flora or structural changes in the eye
`wall, and such changes may also occur in some eyes after
`multiple intravitreal injections [30, 84–86].
`While the use of face masks has not been proven to
`reduce the risk of post-injection endophthalmitis [87],
`they have been associated with a reduction in post-oper-
`ative endophthalmitis [88]. Although maintaining total
`silence may be equivalent to the wearing of a mask, it is
`often important to give patients reassurance and instruc-
`tions while performing the procedure [89–91]. A mask
`may also offer protection in the event of an inadvertent
`cough or sneeze. The needle should remain capped until
`immediately before the injection [21]. Patients should be
`instructed to minimize talking before or during the pro-
`cedure. Assistants involved in setting up the instruments,
`drug and sterile trays should maintain silence or wear
`a face mask, and keep the trays covered until the com-
`mencement of the procedure. Patients’ relatives should
`be encouraged to wait outside the procedure room.
`
`Antisepsis
`The aim of antisepsis is to reduce the bacterial load on
`the ocular surface and the periocular structures including
`the eyelids and eyelashes, without inducing antimicro-
`bial resistance or selecting for more virulent organisms
`[92]. No study to date has found a correlation between
`the number of bacteria on the ocular surface and the risk
`of developing endophthalmitis [93]. Antibiotic resistance
`does not appear to impair the utility of PI and chlorhex-
`idine, the two most commonly used antiseptics [94].
`
`Povidone‑iodine (PI)
`PI is a complex of iodide and a solubilizing carrier, pol-
`yvinylpyrrolidone, which acts as a reservoir of “free”
`iodine, which is the active component [95]. The iodine
`penetrates bacterial cell membranes and inactivates key
`cytosolic proteins, fatty acids and nucleotides. PI does
`
`Novartis Exhibit 2011.008
`Regeneron v. Novartis, IPR2020-01317
`
`

`

`Merani and Hunyor. Int J Retin Vitr (2015) 1:9
`
`Page 9 of 19
`
`not have to be allowed to dry or evaporate to have a bac-
`tericidal effect [96]. It has a broad spectrum of antimi-
`crobial activity with negligible bacterial resistance [97].
`A recent survey found that over 99% of retinal specialists
`use PI before intraocular injections [48].
`
`Efficacy
`In a small randomized study, 5% PI instilled
`into the conjunctival sac prior to ophthalmic surgery
`reduced the number of bacterial colonies by 91%, com-
`pared with a 33% reduction in control eyes [98].
`In a subsequent open-label non-randomized parallel
`trial, Speaker and Menikoff found that the incidence of
`culture-positive endophthalmitis was 0.06% in an oper-
`ating suite using 5% PI, compared with 0.24% in another
`suite using silver protein solution (P < 0.03) [99].
`
`Safety and toxicity Adverse reactions to PI are usually
`the result of an irritant effect that is proportional to the
`duration of exposure, and consequently many patients
`report post-injection pain [100, 101]. A study in rabbits
`demonstrated significant epithelial fluorescein staining
`with 5% PI [102]. It should be irrigated thoroughly post-
`injection to minimize discomfort [100]. Less commonly,
`contact dermatitis may develop after repeated exposure
`[100, 101].
`If PI is applied to the surface of the eye just before the
`needle is inserted through the pars plana, a small amount
`of PI may be introduced into the vitreous cavity [103].
`Animal studies have shown that intravitreal injection of
`a small volume of low-concentration PI is well tolerated
`[104, 105].
`Anaphylaxis to PI is rare, and there have been no
`reports of anaphylaxis following the topical ophthalmic
`use of PI [100, 103]. Furthermore, seafood allergy is not
`a contraindication to the use of topical PI, nor is reported
`allergy to iodinated contrast media [100]. Iodine is not
`the allergenic component of shellfish or contrast media,
`even though both contain iodine [100, 101, 106].
`
`Method of instillation A 10  mL flush of PI onto the
`conjunctival surface and fornices has been shown to lower
`conjunctival bacterial counts more than simply instilling
`a few drops [93, 107]. Flushing is thought to dislodge bac-
`teria from the fornices, where the conjunctiva has many
`deep crypts, allowing PI to kill the organisms [93]. Selec-
`tively flushing one quadrant of the conjunctival surface
`while avoiding the cornea has not been compared with
`bathing the entire ocular surface. While PI may be applied
`to the eyelid margins and eyelashes, eyelid scrubbing
`should be avoided [108, 109].
`
`Concentration and  contact/kill‑time Half-strength
`(5%) PI is commonly used on the ocular surface to reduce
`
`its epithelial toxicity, but the most effective concentration
`is debatable [110].
`Berkelman et al. [111] demonstrated that diluting full-
`strength (10%) PI paradoxically increased its bactericidal
`activity against S. aureus in vitro. After a 15 s exposure to
`PI, no organisms were recovered using concentrations of
`0.1, 0.2 or 1% PI, but sterility was not achieved with such
`a short exposure using 5% or 10% PI. A 1–2 min exposure
`to 5% PI and a 4 min exposure to 10% PI was required to
`achieve sterility.
`Ta et al. [112] found no difference in the conjunctival
`culture rates using 5% PI for 1 min, compared with 10%
`PI for 5 min. Van Rooij et al. [113] reported no increase in
`their rate of post-cataract surgery endophthalmitis, after
`switching from 5 to 1% PI. Shimada et al. [27] described a
`series of 15,144 injections using 0.25% PI, without a sin-
`gle case of endopht