Long-term Complications Associated with Glaucoma Drainage Devices and Boston Keratoprosthesis

Page 1

tGtstmr

Long-term Complications Associated with GlaucomaDrainage Devices and Boston Keratoprosthesis

JENNIFER Y. LI, MARK A. GREINER, JAMES D. BRANDT, MICHELE C. LIM, AND MARK J. MANNIS

tdm

● PURPOSE: To evaluate long-term complications relatedto glaucoma drainage devices in patients undergoingBoston type 1 keratoprosthesis surgery.● DESIGN: Retrospective case series.● METHODS: All patients who underwent Boston type 1keratoprosthesis surgery at the University of California,Davis, between 2004 and 2010 were included. Preoper-ative and postoperative data were reviewed. Twenty-fiveeyes with glaucoma drainage devices were highlighted.Visual acuity and postoperative complications weretracked at postoperative months 1, 3, 6, 9, and 12 and atannual intervals thereafter.● RESULTS: Forty eyes of 35 patients were evaluatedwith an average follow-up of 33.6 months. Conjunctivalbreakdown occurred in association with 10 glaucomadrainage device implants in 9 eyes. Eleven eyes hadglaucoma drainage devices placed before keratoprosthesissurgery, 3 eyes underwent glaucoma drainage deviceplacement at the time of surgery, and 2 eyes had aglaucoma drainage device placed after surgery. All butone of the eroded glaucoma drainage devices were placedbefore surgery. Associated complications included endo-phthalmitis, hypotony, and keratoprosthesis extrusion,with 6 glaucoma drainage devices requiring removal.Long-term beset-corrected visual acuity was maintainedbetter in eyes in which glaucoma drainage device erosionsdid not develop.● CONCLUSIONS: One of the main challenges with kera-oprosthesis surgery is treating concurrent glaucoma.laucoma drainage devices have been advocated as a way

o address this long-term complication, but this seriesuggests that glaucoma drainage device-related complica-ions can cause significant vision loss. (Am J Ophthal-ol 2011;152:209–218. © 2011 by Elsevier Inc. All

ights reserved.)

O VER THE PAST 2 DECADES, KERATOPROSTHESIS

surgery has offered new hope for patients withcorneal blindness who otherwise would be poor

candidates for traditional penetrating keratoplasty. In theUnited States, the most commonly used device is the

Accepted for publication Jan 13, 2011.From the U.C. Davis Health System Eye Center, University of

California at Davis, Sacramento, California (J.Y.L., M.A.G., J.D.B.,M.C.L., M.J.M.).

Inquiries to Mark J. Mannis, Department of Ophthalmology & VisionScience, U.C. Davis Health System Eye Center, 4860 Y Street, Suite

2400, Sacramento, CA 95817; e-mail: [email protected]

© 2011 BY ELSEVIER INC. A0002-9394/$36.00doi:10.1016/j.ajo.2011.01.034

Boston type 1 keratoprosthesis, which was approved by theFood and Drug Administration in 1992.1 The complica-ions of the Boston type 1 keratoprosthesis have been wellocumented in the literature and include retroprostheticembrane formation,2–4 persistent epithelial defects,2

sterile vitritis,2,4–6 infectious endophthalmitis,3, 6–8 retinaldetachments,4,9 keratoprosthesis extrusion,3–4 and glau-coma.2,6,10

Glaucoma continues to be one of the most difficultcomplications to manage in patients with a keratoprosthe-sis. The ability to measure and monitor intraocular pres-sures and glaucomatous damage accurately after placementof a Boston keratoprosthesis continues to elude cornealsurgeons as patients lose vision to this disease. Glaucomadrainage devices are a common surgical option to helpintraocular pressure control when medical therapy is in-sufficient.2–4,10–11 We report our experience with long-term complications related to glaucoma drainage devicesin the setting of keratoprosthesis surgery.

METHODS

A RETROSPECTIVE CHART REVIEW OF ALL BOSTON TYPE 1

keratoprosthesis implantation procedures performed at theUniversity of California, Davis (UC Davis) between May2004 and May 2010. The Boston type 1 keratoprosthesiswas obtained from the Massachusetts Eye and Ear Infirmary(Boston, Massachusetts, USA). All surgeries were per-formed by a single surgeon (M.J.M.) using the technique asdescribed in the initial Boston Keratoprosthesis StudyGroup.4,12

After surgery, we routinely placed a soft Konturcontact lens (Kontur Kontact Lens, Co, Inc, Hercules,California, USA) of 16.0-mm diameter and startedpatients on a short course of medroxyprogesterone 1%eye drops in addition to long-term antibiotic and anti-inflammatory eye drops primarily consisting of eitherfortified vancomycin 50 mg/mL eye drops or a fourth-generation fluoroquinolone eye drop 4 times daily andprednisolone acetate 1% eye drops 4 times daily. Ad-justments were made to this postoperative regimendepending on the surgeon’s assessment of the individualpatient’s situation. Visual acuity and postoperative com-plications were recorded at postoperative months 1, 3, 6,

9, and 12 and at annual intervals thereafter.

LL RIGHTS RESERVED. 209

Page 2

4dsmwsdtt

(UsqinkpBp(sfT28(uapt

fttcttrbsclttwtrTccv

p

Uap3ptnb

IstdAlAtsp

● SURGICAL TECHNIQUE FOR GLAUCOMA DRAINAGE

DEVICE PLACEMENT: Of the 16 eyes reviewed in this study,eyes had glaucoma drainage devices (5 glaucoma drainage

evices) placed at institutions outside of UC Davis. Theurgical reports for these glaucoma drainage device place-ents were not available for review. The remaining devicesere placed at UC Davis. The following describes the routine

urgical technique for placement of a glaucoma drainageevice at UC Davis. The decision for the type, location, andiming of placement of the glaucoma drainage device was athe discretion of the glaucoma surgeon.

Patients who received a Baerveldt glaucoma implantBGI; Abbott Laboratories Inc., Santa Ana, California,SA) received a 350-mm2 device placed underneath the

uperior and lateral rectus muscles in the superior temporaluadrant approximately 9 mm posterior to the limbus. Themplant was secured to the sclera with either 8-0 or 9-0ylon sutures through the implant’s anchoring holes. Thenots were rotated into the implant’s anchoring holes torevent erosion of the knot through overlying conjunctiva.ecause of the nonvalved nature of the implant, the tubeortion was ligated with either a 7-0 or 8-0 polygalactinVicryl; Ethicon, Inc, Somerville, New Jersey, USA)uture followed by confirmation of occlusion. The optionor tube fenestration was at the discretion of the surgeon.he tube was placed into the anterior chamber through a3-gauge needle tract and was secured to the sclera with-0 or 9-0 nylon sutures. Tutoplast allograft pericardiumTutoplast; IOP, Inc, Costa Mesa, California, USA) wassed to cover the tube. Subconjunctival injections ofntibiotic and corticosteroids were placed at the end of therocedure. Postoperative care included administration ofopical prednisolone acetate 1% and a topical antibiotic.

Patients had either a limbus-based conjunctival flap or aornix-based conjunctival flap created for the placement ofhe glaucoma drainage device. The limbus-based conjunc-ival flap was created by making an incision throughonjunctiva and the Tenon capsule at least 8 mm posterioro the limbus in the superior temporal quadrant. Closure ofhe incision was performed with 9-0 Vicryl sutures in aunning locking fashion through the Tenon fascia followedy a running suture through conjunctiva with the sameuture material. The fornix-based conjunctival flap wasreated by making an incision through conjunctiva at theimbus with a peritomy extending 4 to 6 clock-hours ofhe limbal circumference. The incision was radialized inhe inferior temporal quadrant. Closure of the conjunctivaas performed with 8-0 or 9-0 Vicryl anchoring sutures at

he wings of the conjunctival incision. Closure of theadialized incision was performed in a running fashion.he conjunctiva along the length of the limbus was notlosed with sutures. There were no specific indications forhoosing either a limbus-based or fornix-based conjuncti-al flap except for surgeon preference for each approach.Eyes that had a pars plana BGI first underwent pars

lana vitrectomy performed by the vitreoretinal service at c

AMERICAN JOURNAL OF210

C Davis. The procedure for placing the glaucoma drain-ge device is as outlined above except that the tube islaced into the posterior chamber through the pars planamm posterior to the limbus. Eyes that had a stage I BGI

laced underwent the same surgical procedure, however,he tube was inserted at a later date when IOP control waseeded. The conjunctival flap is closed with the tubeuried underneath the conjunctiva superior temporally.Ahmed glaucoma drainage devices (New World Medical

nc., Rancho Cucamonga, California, USA) were placed in aimilar manner as BGI implants. The Ahmed implantsypically were placed in the superior temporal quadrant, butid not need to be inserted under the rectus muscles.dditionally, given the valved nature of the implant, no

igation sutures or tube fenestrations were placed. Thehmed tubes also were placed into the anterior chamber

hrough a 23-gauge needle tract and were secured to theclera with 8-0 or 9-0 nylon suture. Tutoplast allograftericardium was used to cover the tubes at the close of the

TABLE 1. Patient Demographics for All Boston Type 1Keratoprosthesis Surgeries

No. of eyes 40

No. of patients 35

Average age (range), years 52.9 (2 to 86)

Average follow-up time (range),

months

33.6 (5.2 to 72.0)

Preoperative diagnosis

Chemical/thermal burn 11 (27.5%)

Aniridia 5 (12.5%)

Stevens-Johnson syndrome 1 (2.5%)

Ocular cicatricial pemphigoid 1 (2.5%)

Rheumatoid arthritis 1 (2.5%)

Multiple failed grafts

Keratoconus 4 (10%)

Herpes simplex keratitis 4 (10%)

Corneal dystrophy 4 (10%)

Other 9 (22.5%)

TABLE 2. Preoperative Glaucoma Data in Patients withBoston Type 1 Keratoprosthesis

Eyes with pre-KPro glaucoma 23 (57.5%)

Average no. of preoperative glaucoma

medications in patients with glaucoma 1.50

Eyes with prior glaucoma filtering surgery or CPC 14 (35.0%)

Eyes with prior GDD placement only 8 (20.0%)

Eyes with prior trabeculectomy only 1 (2.5%)

Eyes with both prior trabeculectomy and GDD 4 (10.0%)

Eyes with multiple GDDs 2 (5.0%)

Eyes that had GDDs placed at the time of KPro

surgery 3 (7.5%)

CPC � cyclophotocoagulation; GDD � glaucoma drainage

device; Kpro � keratoprosthesis.

ase.

OPHTHALMOLOGY AUGUST 2011

Page 3

RESULTS

TABLE 1 SUMMARIZES THE PATIENT DEMOGRAPHICS OF THE

complete study group. Forty eyes of 35 patients wereevaluated. The underlying preoperative diagnoses are out-

TABLE 3. Postoperative Glaucoma Data in Patients withBoston Type 1 Keratoprosthesis

Eyes being treated for glaucoma at last

follow-up visit 32 (80.0%)

Average no. of glaucoma medications at last

follow-up visit for all patients with

glaucoma 1.22

Eyes requiring ECP or CPC during

postoperative course 5 (12.5%)

Eyes requiring placement of GDDs during

postoperative course 2 (5.0%)

Development of end-stage glaucoma 7 (17.5%)

Total number of GDDs 17

Eyes with GDD erosions 9 (22.5%)

Total no. of eroded GDDs (percentage of

total devices) 10 (58.8%)

Total no. of GDDs removed (percentage of

total devices) 6 (35.3%)

CPC � cyclophotocoagulation; ECP � endocyclophotoco-

agulation; GDD � glaucoma drainage device.

TABLE 4. Best-Corrected Visual Acuity Data in Eyes withBoston Type 1 Keratoprosthesis

% of eyes with preoperative visual acuity �20/400

All eyes without glaucoma (n � 17) 76.5%

All eyes with glaucoma (n � 23) 82.6%

Eyes with GDD without erosions (n � 7) 100.0%

Eyes with GDD erosions (n � 9) 77.8%

% of eyes with BCVA � 20/200 at 1 year follow-up

All eyes without preoperative glaucoma (n � 16) 75.0%

All eyes with preoperative glaucoma (n � 20) 55.0%

Eyes with GDD without erosions (n � 5) 60.0%

Eyes with GDD erosions (n � 8) 75.0%

% of eyes with BCVA � 20/200 at final visit and at

least 1 year follow-up

All eyes without preoperative glaucoma (n � 16;

follow-up range, 17.9 to 72.0 mos) 62.5%

All eyes with preoperative glaucoma (n � 20;

follow-up range, 13.3 to 64.4 mos) 45.0%

Eyes with GDD without erosions (n � 5; follow-

up range, 13.3 to 39.3 mos) 60.0%

Eyes with GDD erosions (n � 8; follow-up range,

20.4 to 64.4 mos) 25.0%

BCVA � best-corrected visual acuity; GDD � glaucoma

drainage device; mos � months.

lined in Table 1.

COMPLICATIONS OF GLAUCOMA DRAINVOL. 152, NO. 2

Table 2 summarizes the glaucoma variables in ourpatient group. Twenty-three eyes (57.5%) had preopera-tive glaucoma, defined as the need for intraocular pressure-lowering medications at the time of surgery or a history ofhaving undergone prior glaucoma filtering procedures, cy-clodestructive procedures, or both. Fourteen eyes (35.0%)had prior glaucoma surgery, cyclodestructive procedures, orboth; 1 patient had undergone both procedures. Addition-ally, 3 eyes (7.5%) underwent placement of a glaucomadrainage device concomitantly with placement of theBoston keratoprosthesis. Two of the glaucoma drainagedevices placed were placed as stage I Baerveldt glaucomaimplants (ie, glaucoma drainage device hardware placed atthe time of the surgery but the tube was not hooked up tothe anterior or posterior chamber); these stage I shunts hadnot yet been connected at the last follow-up visit for these2 patients.

Table 3 summarizes the development and progression ofpostoperative glaucoma in our patient group. Long-termcomplications related to glaucoma drainage devices in-cluded 10 tubes in 9 eyes in which glaucoma drainagedevice erosions developed during the postoperative course(Table 3). The 10 glaucoma drainage device erosionsrepresented 58.8% of all devices that were placed at anygiven time in this group of 40 patients. Of those devicesthat eroded, 6 had to be removed.

Table 4 shows best-corrected visual acuity (BCVA)before and after surgery in eyes in which glaucoma drain-age device erosions subsequently developed as comparedwith eyes with glaucoma drainage devices in which theydid not. It also compares preoperative and postoperativeBCVA of eyes with and without glaucoma (includingthose with a glaucoma drainage device) before placementof the keratoprosthesis. Before surgery, BCVA was compa-rable in all groups. At 1 year, those eyes with a history ofpreoperative glaucoma already showed a decline in thepercentage of eyes that maintained BCVA of 20/200 orbetter, whereas there was only a minimal change in theBCVA of those eyes without preoperative glaucoma after 1year. There is a more pronounced difference at the finalfollow-up visit between these different groups. Eyes withpreoperative glaucoma retained BCVA of 20/200 or betterin only 45.0% of eyes as compared with those eyes withoutpreoperative glaucoma (62.5%). More importantly, ofthose eyes in which tube erosions developed, only 25%maintained BCVA of 20/200 or better. In comparison,60% of eyes with tube shunts without erosions maintainedBCVA of 20/200 or better and 62.5% of eyes withoutpreoperative glaucoma maintained that level of visualacuity.

The clinical courses of the 9 eyes and 10 glaucomadrainage devices in which tube erosions developed areoutlined in Table 5. The duration of the shunt beforesurgery, preoperative visual acuity, preoperative intraocu-lar pressure, best vision documented after surgery, duration

after surgery to tube erosion, subsequent complications

AGE DEVICES WITH BOSTON KPRO 211

Page 4

TABLE 5. Patients with Keratoprosthesis Device and Glaucoma Drainage Device Erosions

Patient

No. Underlying Disease

GDD: Type, Location

(covered by)

Duration of

GDD

before Surgery

(mos)

Preoperative

VA

Preoperative IOP

(mm Hg)

Best VA after

Surgery (Time

Recorded; mos)

Time to GDD

Erosion after

Surgery (mos) Complications

Length of

Follow-up

(mos) Final VA

1 Chemical burn IT BGI (unknown) 53 HM 32 20/150 (3) 45 Noted to have a CL fragment near the

erosion site during the initial tube

revision. Initial tube revision with

Tutoplast failed. Subsequent

removal of GDD with concomitant

ECP.

64 HM

2 Keratoconus with multiple

failed PKP

(1) IN Ahmed

(unknown) (2)

Pars plana ST

BGI (unknown)

(1) 10 (2) 19 HM 29 20/25-2 (6) 8 Contact lens eroding the conjunctiva

overlying IN Ahmed initially noted

but reconjunctivalized

spontaneously. Subsequent

exposure again 26 mos later with

associated drop of VA to HM,

removal of IN Ahmed GDD and

CPCx2.

54 HM

3 Multiple failed PKP (1) IN Ahmed

(Tutoplast) (2)

Pars plana ST

BGI (Tutoplast)

(1) 15 (2) 5 20/400 16 20/50 (24) (1) 30 (2) 32 Development of Pseudomonas

aeruginosa endophthalmitis with

concomitant IN Ahmed tube

exposure. Revision with Tutoplast

followed by choroidal detachment,

RD with exposure of second tube 2

mos later requiring removal of ST

BGI. IN Ahmed plate exposure 2

mos subsequently requiring

removal.

52 LP

4 Keratoconus with multiple

failed PKP

ST pars plana

BGI (unknown)

40 CF 15 20/40 (12) 17 Exposure of Hoffman elbow with

revision using Tutoplast. Fourteen

mos after revision, plate became

exposed, requiring second revision.

44 20/50-2

5 Multiple failed PKP ST BGI

(unknown)

Placed 17

mos after

Kpro

HM 5 CF 2 Small area of scleromalacia with

associated adjacent GDD erosion.

Glued to close wound leak.

Subsequently pre phthisical and

lost to follow-up.

20 HM

6 Aniridia ST BGI

(Tutoplast)

30 CF 17 20/20 (3) 4 GDD erosion and repair with

subsequent endophthalmitis,

hypotony, and KPro extrusion. GDD

removed at time of endophthalmitis.

33 HM

(Continued on next page)

AM

ERIC

AN

JOU

RN

AL

OF

OPH

THA

LMO

LOG

Y212

AU

GU

ST2011

Page 5

TA

BLE

5.(C

ontin

ued

)

Pat

ient

No.

Und

erly

ing

Dis

ease

GD

D:

typ

e,lo

catio

n

(cov

ered

by)

Dur

atio

nof

GD

D

bef

ore

Sur

gery

(mos

)

Pre

oper

ativ

e

VA

Pre

oper

ativ

eIO

P

(mm

Hg)

Bes

tV

Aaf

ter

Sur

gery

(Tim

e

Rec

ord

ed;

mos

)

Tim

eto

GD

D

Ero

sion

afte

r

Sur

gery

(mos

)C

omp

licat

ions

Leng

thof

Follo

w-u

p

(mos

)Fi

nalV

A

7A

nirid

iaS

TB

GI

(Tut

opla

st)

3620

/400

920

/150

(3)

9G

DD

rem

oved

.S

ubse

que

nthy

pot

ony,

chor

oid

ald

etac

hmen

ts,

and

RD

.

27N

LP

8M

ultip

lefa

iled

PK

PS

TB

GI

(Tut

opla

st)

37C

F21

20/3

0(3

to12

)11

Ero

sion

occu

rrin

gat

edge

ofC

L.

Rev

ised

with

Tuto

pla

st.

2420

/70-

1

9A

nirid

iaS

Tp

ars

pla

na

BG

I(T

utop

last

)

53LP

10H

M7

Ero

sion

atan

terio

red

geof

Hof

fman

elb

ow.

Rev

ised

/cov

ered

with

corn

ealp

atch

.

9H

M

BG

I�B

aerv

eld

tgl

auco

ma

imp

lant

;CF

�co

untin

gfin

gers

;CL

�co

ntac

tle

ns;C

PC

�cy

clop

hoto

coag

ulat

ion;

EC

P�

end

ocyc

lop

hoto

coag

ulat

ion;

GD

D�

glau

com

ad

rain

age

dev

ice;

HM

�

hand

mov

emen

ts;

IN�

infe

rior

nasa

l;IO

P�

inoc

ular

pre

ssur

e;IT

�in

ferio

rte

mp

oral

;K

pro

�ke

rato

pro

sthe

sis;

LP�

light

per

cep

tion;

mos

�m

onth

s;N

LP�

nolig

htp

erce

ptio

n;P

KP

�

pen

etra

ting

kera

top

last

y;R

D�

retin

ald

etac

hmen

t;S

T�

sup

erio

rte

mp

oral

;V

A�

visu

alac

uity

.

VOL.

related to tube erosion, and final visual acuity at lastfollow-up visit are documented.

Overall, 9 of 10 glaucoma drainage devices in whichtube erosions developed after surgery were placed beforeplacement of the Boston type 1 keratoprosthesis. Theaverage duration that the eroded devices were in placebefore placement of the keratoprosthesis was 31 months(range, 5 to 53 months). Only 1 tube erosion occurred ina device placed after the keratoprosthesis surgery at post-operative month 17 (Patient 5). The glaucoma drainagedevices that were placed at the UC Davis had excellentdocumentation as to whether the tube was covered at theclose of the case and with what material. Six of 10 tubes inthis group were known to have been covered with Tuto-plast allograft pericardium based on operative reports. Theremaining 4 tubes were placed at outside institutions andthe operative records were not available. The location(inferior nasally vs superior temporally) and the site ofinsertion (limbus vs pars plana) also are noted in Table 5.

The average duration between the placement of theBoston keratoprosthesis and the erosion of the glaucomadrainage device was 16.5 months (range, 2 to 45 months).Visual outcomes after erosion of the glaucoma drainagedevice were poor, with only 2 eyes maintaining BCVA of20/200 or better at the final follow-up visit (Patients 4 and8), despite 7 eyes having attained that level of vision atsome point during the postoperative course.

Table 5 also highlights the complications occurring afterdrainage device erosion, including those in 2 eyes in whichendophthalmitis developed at the time of or immediatelyafter the episode of tube erosion (Patients 3 and 6). Patient3 had culture-positive endophthalmitis secondary to Pseu-domonas aeruginosa while receiving routine prophylaxiswith fortified vancomycin drops twice daily. In Patient 6,endophthalmitis developed (presumed Staphylococcus au-reus) also while receiving routine prophylaxis with fortifiedvancomycin drops twice daily. In 3 eyes, hypotony, preph-thisis, or both developed after the erosion and repair orremoval of the glaucoma drainage device (Patients 5, 6,and 7), and 1 of these eyes subsequently extruded thekeratoprosthesis spontaneously (Patient 6). Six of theglaucoma drainage devices had to be removed (Patients 1,2, 3, 6, and 7), with 2 eyes requiring cyclophotocoagula-tion procedures to help with intraocular pressure controlafter glaucoma drainage device removal (Patients 1 and 2).

Table 6 summarizes the remaining 7 eyes with glaucomadrainage devices in which tube erosions did not develop.All of these glaucoma drainage devices were placed at UCDavis, and good records of the operative report wereavailable for review. Only 3 of these devices had beenplaced before surgery (Patients 11, 12, and 13), with anaverage duration before surgery of 15.7 months (range, 7 to27 months). Three devices were placed at the time ofkeratoprosthesis surgery (Patients 10, 15, and 16), and 1device was placed after the keratoprosthesis surgery (Pa-

tient 14). Two patients (Patients 12 and 15) had stage I

COMPLICATIONS OF GLAUCOMA DRAINAGE DEVICES WITH BOSTON KPRO152, NO. 2 213

Page 6

TABLE 6. Patients with Keratoprosthesis Device and Glaucoma Drainage Device without Erosions

Patient

No. Underlying Disease

GDD: Type,

Location (covered by)

Timing of GDD

Placement

Preoperative

VA

Preoperative IOP

(mm Hg)

Best VA after Surgery

(Time Recorded) Comments

Length of

Follow-up (mos) Final VA

10 HLA-B27 uveitis with multiple

failed PKP

ST pars plana BGI

(Tutoplast)

At time of KPro

surgery

CF 33 20/50 (36 mos) Glaucoma progression noted

based on optic nerve

appearance.

39 20/100

11 Multiple failed PKP ST Ahmed

(Tutoplast)

7 mos before KPro

surgery

HM 5 20/70 (3 mos) History of noncompliance with

medications.

13 20/200

12 Peter’s anomaly Stage I ST BGI 27 mos before KPro

surgery

F&F 22 F&F Postoperative course complicated

by corneal melt x3 with

extrusion of the

keratoprosthesis and

subsequent episode of

microbial keratitis.

33 LP

13 Multiple failed PKP ST Ahmed

(Tutoplast)

13 mos before KPro

surgery

CF 17 CF Postoperative retroprosthetic

membrane refractory to

treatment.

29 CF

14 Chemical burn ST Ahmed

(Tutoplast)

18 mos after KPro

surgery

LP Soft 20/40 (1 mo) Noted to have progression of

glaucoma based on optic nerve

appearance. BRVO developed

with drop in visual acuity

subsequently.

20 20/200

15 CHED with multiple failed

PKP

Stage I ST BGI At time of KPro

surgery

HM Firm 20/200 (1 mo) 5 20/400

16 Chemical burn ST BGI (Tutoplast) At time of KPro

surgery

HM 15 CF 5 CF

BGI � Baerveldt glaucoma implant; CF � counting fingers; CHED � congenital hereditary endothelial dystrophy; CPC � cyclophotocoagulation; ECP � endocyclophotocoagulation; F&F �

fixes and follows; GDD � glaucoma drainage device; HM � hand movements; IN � inferior nasal; IOP � intraocular pressure; IT � inferior temporal; Kpro � keratoprosthesis; LP � light

perception; mos � months; PKP � penetrating keratoplasty; ST � superior temporal.

AM

ERIC

AN

JOU

RN

AL

OF

OPH

THA

LMO

LOG

Y214

AU

GU

ST2011

Page 7

doatwa(idpsf

bdbhetdewdTakp

feat(gnot1otfim

sbbhlHaaet

Baerveldt devices placed. Five glaucoma drainage deviceshad Tutoplast placed over the tube. The 2 glaucomadrainage devices that did not have Tutoplast were placedas stage I procedures.

Long-term visual acuity results in this group show that atthe final follow-up visit, 3 eyes (50%) maintained 20/200vision or better. Three eyes (50%) never improved invisual acuity after placement of the keratoprosthesis—thiswas presumed to be related to end-stage glaucomatousdamage before placement of a keratoprosthesis.

DISCUSSION

GLAUCOMA CONTINUES TO BE ONE OF THE MOST DIFFI-

cult postoperative management challenges in patients witha Boston type 1 keratoprosthesis. The challenges to ad-dressing the issue of glaucoma in this patient populationare multiple.

Before surgery, it can be difficult to determine the degree ofglaucomatous damage in the eye under consideration for akeratoprosthesis. Patients who have had either multiple failedcorneal transplantations, anterior segment dysgenesis syn-dromes, or chemical burns likely have underlying damage tothe angle and are more likely to have secondary glaucomaeven before keratoprosthesis surgery. Unfortunately, the vastmajority of these patients have such severe corneal pathologicfeatures before surgery that there is no view to the posteriorpole and the optic nerve. In addition, the other eye often isaffected by the same disease process with a poor view of theiris, and therefore testing for a relative afferent papillarydefect is not possible. It may not be until after the surgery isfinished that one is able to view and determine the extent ofglaucomatous optic neuropathy. Patients 9 and 16 in ourseries here are examples of patients who were noted to havesevere optic nerve cupping only after placement of thekeratoprosthesis, which likely explained the lack of signifi-cant visual improvement after keratoprosthesis placement.

Additionally, monitoring of intraocular pressures also isa challenge. Current methods of tonometry cannot be usedon an eye with a keratoprosthesis. In some patients withadequate vision and a view that allows documentation ofoptic nerve appearance and structure, perimetry and opticnerve imaging can be performed to follow the status ofglaucoma. However, this is not possible in all eyes. In ourseries of patients, several had progression of glaucomadespite close monitoring. This reflects the inaccuracy oftactile pressure measurements and the difficulty of follow-ing optic nerve status in keratoprosthesis eyes even in thehands of experienced glaucoma specialists.

One approach to glaucoma management in these pa-tients has been to implant glaucoma drainage devices atthe time of keratoprosthesis surgery in all eyes requiringtreatment with antihypertensive drops before surgery.13

Although there may seem little downside to implanting a

drainage tube in all patients with marginal intraocular (

COMPLICATIONS OF GLAUCOMA DRAINVOL. 152, NO. 2

pressure control,2 our study found that glaucoma drainageevices eroded through conjunctiva in 9 patients (22.5%f all eyes and 58.8% of all drainage devices); required 14dditional surgeries to revise or remove the device; neededo be explanted in 5 eyes (6 explanted devices total); andere implicated in the development of endophthalmitisnd periprosthetic infection leading to extrusion in 2 eyesPatients 3 and 6). The need to achieve postoperativentraocular pressure control is significant, yet glaucomarainage devices seem to carry significant risks in thisopulation, and alternatives such as transscleral or endo-copic cyclophotocoagulation should be considered care-ully when determining the surgical plan.

Eight of the 9 glaucoma drainage devices were implantedefore the keratoprosthesis surgery (Table 5), with an averageuration in the eye of 31 months (range, 5 to 53 months)efore the keratoprosthesis surgery. In fact, 6 of the 9 shuntsad been in place for several years before surgery, and only 1roded glaucoma drainage device was placed after the kera-oprosthesis surgery. This is in contrast to the glaucomarainage devices in our patients who did not experience anrosion episode. Of the 7 devices that did not erode, only 3ere placed before keratoprosthesis surgery, with an averageuration in the eye of 15.7 months (range, 7 to 27 months).he remaining glaucoma drainage devices that did not hadn erosion episode were placed either at the time of theeratoprosthesis surgery or during the postoperative follow-uperiod.

The average duration of postoperative keratoprosthesisollow-up for our patient in whom glaucoma drainage devicerosions developed was 35.3 months (range, 9 to 64 months),s compared with an average follow-up of 20.6 months inhose patients without glaucoma drainage device erosionrange, 5 to 39 months). Our results may be biased to olderlaucoma drainage devices developing more erosions thanewer glaucoma drainage devices because of the longerverall follow-up time of the older implants. However, theime from keratoprosthesis surgery to erosion is an average of8.3 months (range, 2 to 45 months), with 5 of the 9 erosionsccurring within the first postoperative year after keratopros-hesis placement. Thus, it seems that one important riskactor for tube erosion is the duration that the tube has beenn place before surgery, with older glaucoma drainage devicesore likely to develop erosions.Another risk factor for glaucoma drainage device ero-

ions seen in our series of patients and that has previouslyeen described by Chew and associates is the presence of aandage contact lens.13 All of our keratoprosthesis patientsave a bandage contact lens placed (typically a Kontur

ens with a diameter of 16.0 mm), as recommended byarissi-Dagher and associates and Dohlman and associ-

tes, to protect the ocular surface from evaporative dam-ge.14,15 Of the 9 cases of glaucoma drainage devicerosion, in at least 3 cases (Patients 1, 2, and 8), conjunc-ival breakdown occurred at the edge of contact lens

Figures 1 and 2). Presumably, the focal mechanical

AGE DEVICES WITH BOSTON KPRO 215

Page 8

trauma of the contact lens edge on top of the glaucomadrainage device contributed to the erosions of thesedevices.

These glaucoma drainage devices erode not onlythrough conjunctiva, but also through allograft pericar-dium patch grafts (Tutoplast). For those glaucomadrainage devices that were placed at the UC Davis, wehave documented records of pericardial patch graft use(see Table 5). The type or location of the glaucomadrainage device does not seem to matter— eroded glau-coma drainage devices include both Ahmed glaucomavalves and BGIs in our series and were located in thesuperior temporal, inferior temporal, or inferior nasalquadrants (Figure 3). Although a pars plana insertion ineyes with a keratoprosthesis may seem to offer anadvantage by placing glaucoma drainage device hard-ware further back from the limbus, 3 of our erosionsoccurred in glaucoma drainage devices placed in thepars plana (Patients 3, 4, and 9) where the plate itselferoded through Tutoplast and conjunctiva (Figure 2).Two of these erosions occurred in patients with a BGIthat incorporates a Hoffman elbow for pars planainsertion (Patients 4 and 9). Hoffman elbows have beenlinked to erosion problems in eyes even without kera-toprosthesis and are no longer used widely. We nowbelieve that a pars plana approach in which the tube ispassed directly into the posterior chamber under a thickscleral flap with overlying pericardial patch graft rein-forcement may be the best solution to the issue ofglaucoma drainage device erosions because it puts thedevice as far away from the limbus as possible and placesas much reinforcing material over the device as possible.

Many patients undergoing implantation of a Boston kera-toprosthesis are already predisposed to conjunctival break-

FIGURE 1. Photograph showing a Boston type 1 keratopros-thesis with a glaucoma drainage device eroding through con-junctiva in Patient 8. Note the erosion of the conjunctivaoverlying the tube at the edge of the contact lens.

down because of their underlying ocular surface disease.

AMERICAN JOURNAL OF216

Patients with prior chemical burns are an obvious example ofpatients with poor ocular surface healing, but one mustwonder if these patients who have had multiple failedpenetrating keratoplasties before their keratoprosthesis proce-dure also may have a poor ocular surface contributing to theirfailed grafts. Long-term eye drops, including fortified vanco-

FIGURE 2. Photograph showing a Boston type 1 keratopros-thesis with an exposed pars plana glaucoma drainage device inPatient 9. There is erosion of the pars plana Hoffman elbow atthe edge of the bandage contact lens.

FIGURE 3. Photograph showing a Boston type 1 keratopros-thesis with an exposed glaucoma drainage device in the inferiornasal quadrant in Patient 3. There is significant conjunctivalerosion and exposure of the tube.

mycin drops, also can cause some degree toxicity to the ocular

OPHTHALMOLOGY AUGUST 2011

Page 9

surface, thereby accelerating breakdown of the ocular surfacein the long term. The combination of a poor ocular surfaceand the mechanical trauma of the long-term contact lens inthese patients seems to predispose them to glaucoma drainagedevice erosions.

Finally, the long-term visual consequences of glau-coma drainage device erosions are significant. Of ourpatients in whom glaucoma drainage device erosionsdeveloped, only 2 were able to maintain functionalvision (BCVA �20/200). The other patients hadBCVA of hand movements or worse despite the factthat all but 2 of these patients (Patients 5 and 9) hadsignificant improvement of vision after placement of thekeratoprosthesis. The complications related to the glau-coma drainage device erosions includes associated en-dophthalmitis, hypotony, choroidal detachments, andretinal detachments. These complications contributedto vision loss in these patients, and in comparison withpatients without glaucoma drainage device erosions,long-term vision loss in patients with glaucoma drainagedevice erosions is worse (Table 4). One confoundingfactor to note is that eyes with glaucoma drainage deviceerosions probably were more likely to have had glau-coma for a longer period before surgery based just on thetiming of glaucoma drainage device placement. Thus,worse underlying glaucoma also may be contributing tovision loss, in addition to the complications related tothe glaucoma drainage device erosions.

It is important to point out that endophthalmitisdeveloped in our patients with keratoprosthesis andexposed glaucoma drainage device tubes, despite thefact that these patients were receiving daily prophylaxiswith fortified vancomycin eye drops. Obviously, al-though vancomycin provides good broad-spectrumgram-positive coverage to address the most commonpathogens on the lids and ocular surface, it does notaddress gram-negative pathogens such as Pseudomonas

that developed in one of our patients. An agent such as

experience. Cornea 2009;28(3):321–327.

COMPLICATIONS OF GLAUCOMA DRAINVOL. 152, NO. 2

fortified ceftazidime eye drops or even a broad-spectrumfourth-generation fluoroquinolone would help to addressgram-negative coverage better, but at the expense ofgram-positive coverage. One would have to weigh care-fully the benefits of adding another agent to the pa-tient’s postoperative regimen with the financial cost tothe patient and the effect of an additional agent onpatient compliance. However, in light of the complica-tions and the reality of long-term contact lens use inthese patients, coverage with 2 antibiotic drops in thelong term is advised and the practice patterns at UCDavis also have been adjusted accordingly.

This study has several limitations. First of all, it hasthe inherent limitations of a retrospective study inwhich all data were collected outside of the context ofa prospective protocol. Additionally, the study group issmall and our findings are observational in nature.There are numerous confounding variables in this pa-tient population based on the diversity of underlyingdiseases and ocular comorbidities that may affect visualpotential and long-term complications. The Boston type1 keratoprosthesis is a final option for patients withcorneal blindness who have exhausted other options. Itgives hope to these patients, and for many, good visioncan be achieved at least for a limited time. One of themain concerns remains the long-term management ofglaucoma. Many of the patients have pre-existing glau-coma and in many others, glaucoma develops aftersurgery. Glaucoma drainage devices have been advo-cated as a way to address this long-term complication,but because we have shown in our series that thesedevices are not benign and that related complicationscan cause significant vision loss. As a result, the con-tinued coordination of care and research efforts betweencornea and glaucoma specialists to refine the postoper-ative management of keratoprosthesis surgery is of

utmost importance.

THE AUTHORS INDICATE NO FINANCIAL SUPPORT. DR BRANDT IS A CONSULTANT FOR ABBOT MEDICAL OPTICS, ENDO-Optiks, and Glaukos. Involved in conception and design of study (J.Y.L., M.A.G., M.J.M.); Conduct of study (J.Y.L., M.A.G., M.C.L., J.D.B., M.J.M.);collection of data (J.Y.L., M.A.G., M.C.L., J.D.B., M.J.M.); and approval of manuscript (J.Y.L., M.A.G., M.C.L., J.D.B., M.J.M.). This study wasapproved by the University of California, Davis, Institutional Review Board.

REFERENCES

1. Dohlman CH. Background of the present Boston Kpro Ifor graft failures. Boston Keratoprosthesis Update 2004;1:1–2.

2. Aldave AJ, Kamal KM, Vo RC, Yu F. The Boston type Ikeratoprosthesis: improving outcomes and expanding indica-tions. Ophthalmology 2009;116(4):640–651.

3. Bradley JC, Hernandez EG, Schwab IR, Mannis MJ. Bostontype 1 keratoprosthesis: the University of California Davis

4. Zerbe BL, Belin MW, Ciolino JB. Results from the multi-center Boston Type 1 Keratoprosthesis Study. Ophthalmol-ogy 2006;113(10):1779e1–1779e7.

5. Nouri M, Durand ML, Dohlman CH. Sudden reversiblevitritis after keratoprosthesis: an immune phenomenon?Cornea 2005;24(8):915–919.

6. Khan BF, Harissi-Dagher M, Khan DM, Dohlman CH.Advances in Boston keratoprosthesis: enhancing reten-tion and prevention of infection and inflammation. IntOphthalmol Clin 2007;47(2):61–71.

7. Nouri M, Terada H, Alfonso EC, Foster CS, Durand ML,

Dohlman CH. Endophthalmitis after keratoprosthesis: inci-

AGE DEVICES WITH BOSTON KPRO 217

Page 10

1

1

1

1

dde

i

S

dence, bacterial causes, and risk factors. Arch Ophthalmol2001;119(4):484–489.

8. Tsui I, Uslan DZ, Hubschman JP, Deng SX. Nocardiafarcinica infection of a Baerveldt implant and endophthal-mitis in a patient with a Boston type I keratoprosthesis. JGlaucoma 2010;19(5):339–340.

9. Ray S, Khan BF, Dohlman CH, D’Amico DJ. Managementof vitreoretinal complications in eyes with permanent kera-toprosthesis. Arch Ophthalmol 2002;120(5):559–566.

10. Netland PA, Terada H, Dohlman CH. Glaucoma associatedwith keratoprosthesis. Ophthalmology 1998;105(4):751–757.

11. Rubin PA, Chang E, Bernardino CR, Hatton MP, DohlmanCH. Oculoplastic technique of connecting a glaucoma valveshunt to extraorbital locations in cases of severe glaucoma.Ophthal Plast Reconstr Surg 2004;20(5):362–367.

AJO History of OThe Cyclopes

There are two different traditions about the giantCyclopes of ancient Greece. Polyphemus, theCyclope blinded by Odysseus in the Odyssey, is

epicted as a malevolent monster, but curiously, nowhereoes Homer specifically state that Polyphemus had only oneye in the middle of his forehead.

The other tradition was described by the Greek poet Hesiodn his poem the Theogeny (origin of the gods). There the

ubmitted by Ronald S. Fishman of the Cogan Ophthalmic

AMERICAN JOURNAL OF218

2. Dohlman CAbad JDudenhoefer EGraney J. Keratoprosthesis:beyond corneal graft failure. In: Spaeth G, ed. OphthalmicSurgery: Principles and Practice, 3rd ed. Philadelphia: W.B.Saunders; 2002:199–207.

3. Chew HF, Ayres BD, Hammersmith KM, et al. Bostonkeratoprosthesis outcomes and complications. Cornea 2009;28(9):989–996.

4. Harissi-Dagher M, Beyer J, Dohlman CH. The role of softcontact lenses as an adjunct to the Boston keratoprosthesis.Int Ophthalmol Clin 2008;48(2):43–51.

5. Dohlman CH, Dudenhoefer EJ, Khan FF, Morneault S.Protection of the ocular surface after keratoprosthesissurgery: the role of soft contact lenses. CLAO J2002;28(2):72–74.

halmology Seriesreek Legend

yclopes were sons of Uranus (the sky) and Gaia (the earth)nd did have only a single eye in the center of the forehead.hey were often depicted as blacksmiths at the forge, provid-g thunderbolts to Zeus and the trident to Poseidon, whichere then used to overcome the Titans. Having only one eye

eems a hazardous situation for a blacksmith. Probably duringhe Bronze Age and early Iron Age it was so common for metalorkers to have lost an eye from a spark or foreign body thatyth incorporated this element in it.

tory Society

phtof G

CaTinwstwm

His

OPHTHALMOLOGY AUGUST 2011