Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.
com
BJO Online First, published on August 20, 2015 as 10.1136/bjophthalmol-2015-307390
1
Singapore National Eye
Centre, Singapore, Singapore
2
Histopathology, Pathology
Department, Singapore General
Hospital, Singapore, Singapore
3
Singapore Eye Research
Institute, Singapore, Singapore
4
Ophthalmology & Visual
Sciences Academic Clinical
Program, Duke-NUS Graduate
Medical School, Singapore,
Singapore
5
Nanyang Technological
University, Singapore,
Singapore
6
Department of
Ophthalmology, Yong Loo Lin
School of Medicine, National
University of Singapore,
Singapore, Singapore
Correspondence to
Professor Donald TH Tan,
Singapore National Eye Center,
11 Third Hospital Avenue,
Singapore 168751, Singapore;
donald.tan.t.h@snec.com.sg
Received 2 July 2015
Revised 27 July 2015
Accepted 2 August 2015
To cite: Chan ASY,
Mehta JS, Al Jajeh I, et al.
Br J Ophthalmol Published
Online First: [ please include
Day Month Year]
doi:10.1136/bjophthalmol-
2015-307390
Copyright Article author (or their employer) 2015. Produced by BMJ Publishing Group Ltd under licence.
Clinical science
Histological features of Cytomegalovirus-related
corneal graft infections, its associated features
and clinical significance
Anita SY Chan,1,2,3,4 Jodhbir S Mehta,1,3,4,5 Issam Al Jajeh,2 Jabed Iqbal,2
Arundhati Anshu,1,3 Donald TH Tan1,3,4,5,6
ABSTRACT
Aim To describe the histological features of
Cytomegalovirus (CMV)-related corneal graft infections,
its associated features and clinical significance.
Methods This was a retrospective histological study of
48 consecutive cases of failed repeat penetrating
keratoplasty cases with a clinical diagnosis of allograft
rejection from 2011 to 2013. CMV infection was
confirmed with CMV antibody immunohistochemistry
(IHC) and electron microscopy. Additional CD163 and
CD68 IHCs for macrophages were also performed.
Clinical data and previous graft histology were then
reviewed.
Results Mean incidence of CMV infection in corneal
graft rejection buttons was 6.3% per year. 3/48 graft
buttons were CMV antibody positive. Histological
features of CMV graft infection include: (1) stromal
keratocytes with cytopathic changes; (2) lack of
inflammation, only occasional macrophages present and
(3) absence of vascularisation. None of the patients had
a history of active CMV infection.
Conclusion CMV infection is not limited as
endotheliitis, but extends into the corneal stroma, and is
a potential reservoir for graft infection, especially in
partial thickness endothelial surgery. Clinical features are
often non-specific, although glaucoma was present in
our patients. CMV-infected grafts showed CD163-
positive M2 macrophages in close association with the
infected keratocytes, suggesting that the macrophage
may be important in CMV graft infection. Histological
examination with CMV IHC is a useful method to detect
CMV infection postoperatively. Post penetrating
keratoplasty, CMV systemic treatment with valganciclovir
can prevent graft infection and failure. Boston
keratoprosthesis may be a potential alternative surgery in
active CMV infections that obviates the need for systemic
therapy.
INTRODUCTION
Cytomegalovirus (CMV) organ transplant-related
infection is a well-known complication of solid
organ transplant surgery with significant morbid-
ity.1 Previously, corneal transplant rejection, unlike
its solid organ counterparts, was not thought to be
associated with CMV infection. However, CMV
endotheliitis in the immunocompetent host can
mimic graft rejection, and has been identified as an
important cause of graft failure.2–7 Differentiation
between corneal endothelial rejection and CMV
graft infection can be difficult due to similarities in
presentation, although the presence of epithelial
and/or corneal stromal oedema, pigmented keratic
Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390
precipitates (KPs) and linear Descemet’s membrane
folds8 would alert one to suspect CMV infection.
CMV corneal histopathology has not been previ-
ously characterised, and the incidence of undiag-
nosed corneal CMV infection as a cause of graft
rejection and failure is unknown. The purpose of
this study was to determine its histological features,
related clinical features and incidence. As CMV
remains in a latent state within blood monocytes,
which subsequently differentiate into tissue macro-
phages when activated, we also postulated that
macrophages may play a role in CMV infection in
the cornea, and hence, studied the presence of
these cells by immunohistochemistry (IHC).9 10
METHODS
This was a retrospective study of 48 consecutive
cases of failed repeat penetrating keratoplasty (PK)
cases with a clinical diagnosis of allograft rejection
between January 2011 and December 2013.
Clinical cases were identified within the Singapore
Corneal Transplant Study database, an institutional
review board (IRB)-approved prospective database
tracking all corneal transplants from 1991 to
present day at the Singapore National Eye Centre
(SNEC). Failed PK donor buttons from these cases
were analysed by the SNEC Ophthalmic Pathology
Collaborative Service, with approval by the local
IRB.
H&E sections of the cornea were reviewed for
viral cytopathic features, vascularisation and inflam-
mation. CMV infection was confirmed with CMV
IHC and electron microscopy (EM), which were
performed according to standard diagnostic proto-
cols.11–13 In brief, 4 mm sections (Microsystems
Plus Slides, Leica) were cut and incubated overnight
at 80°C. Deparaffinisation, heat epitope retrieval
and blocking of endogenous peroxidase activity
were performed. Primary antibodies for CMV,
CD68 and CD163 (Dako, USA) were applied, fol-
lowed by secondary antibody conjugated with
horseradish peroxidase, diaminobenzidine and/or
alkaline phosphatase-anti-alkaline phosphatase
(Vision Biosystems, USA).11 12 For EM, the region
of viral inclusions seen on H&E section was identi-
fied, and the tissue was dewaxed in xylene, washed
in ethanol, trimmed to a 1 mm cube and placed in
0.1 M cacodylate buffer for 10 min. Postfixation in
1% aqueous osmium tetroxide at room temperature
for 1 h was performed and then the section was
placed in 1:1 propylene oxide:812 resin for 2 h
and 1:3 propylene oxide:812 resin for 2 h. It was
embedded in fresh pure 812 resin and cured in a
1
 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com
Clinical science
60°C oven for 24 h. Ultra-thin sections were cut, stained and
mounted on copper grids for study.13
When CMV was detected, the previous corneal specimens
and medical records were reviewed.
Statistical analysis was performed with a Student’s t test with
p values <0.05 considered significant.
RESULTS
Three grafts from three patients demonstrated histological evi-
dence of CMV infection out of 48 graft buttons reviewed from
2011 to 2013 for presumed graft rejection. The incidence of
CMV graft infection was 6.7% (1/15), 7.1% (1/14) and 5.2%
(1/19) in 2011, 2012 and 2013, respectively, with the mean
incidence per year of 6.33% (±SD 0.01%). Retrospective
review of previous graft buttons from these three individuals
included five buttons removed prior to 2011. Total number of
grafts reviewed was 53.
The median number of regrafts in the non-CMV-infected
failed grafts (n=45) was two (average 2.2±SD 0.5) grafts/
patient in comparison with three (average 3.0±SD 1.0) grafts/
patient in the CMV-infected failed grafts (n=3), and was statis-
tically significant ( p=0.001). In this study, none of the patients
with histologically CMV-negative failed grafts were subsequently
diagnosed with CMV graft-related infection in the initial post-
operative period.
Histological findings
All three graft buttons showed typical features of graft failure
with stromal oedema and loss of endothelial cells (figures 1–3).
Stromal viral cytopathic features comprising cytomegaly
(defined as infected keratocytes enlarged up to two times the
size of a normal keratocytes), with eosinophilic intranuclear and
intracytoplasmic inclusions (figures 1–3C) were also seen. In the
corneas of patients 1 and 3, viral cytopathic changes were con-
centrated in the posterior stroma, immediately anterior to the
Descemet’s membrane. CMV IHC demonstrated CMV-positive
staining in these cells and also highlighted CMV infection in
normal-sized keratocytes (figure 1F). Our second patient
showed anterior stromal infection near the graft–host junction
rather than the posterior stroma (figures 2B, C). The residual
atrophic endothelial cells showed no evidence of CMV endothe-
liitis on light microscopy and IHC in these three cases, but was
seen in a previous graft of patient 1 (figure 1H, insert).
Significant acute inflammation (neutrophils), chronic inflam-
mation (lymphocytes and plasma cells) and granulomas (epithe-
lioid histiocytes) were absent in all three cases (figures 1–3B),
although macrophage markers showed occasional
CD163-positive M2 macrophages and CD68-positive macro-
phages adjacent to the cytopathic change. No colocalisation of
CD163-positive and CD68-positive macrophages with CMV
antibody IHC was seen (figures 1–3F).
Vascularisation of the cornea was absent in all three cases.
Descemet’s membrane was also intact (figures 1B–3B).
Electron microscopy
EM was performed in all three cases, and confirmed the intracy-
toplasmic and intranuclear viral capsids with an average size of
112–120 (±SD11) nm consistent with CMV viral capsids
(figure 3G–I) within the cells with the morphology of
keratocytes.
Clinical history, management and outcome
These patients had no previous diagnosis of active ocular or sys-
temic CMV infection or documented clinical features of
2 Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390
endotheliitis prior to surgery. All were immunocompetent with
normal full blood counts and negative HIV serologies prior to
surgery, although all three received systemic corticosteroid and/
or immunosuppression. Table 1 summarises the graft surgeries
and time to graft rejection/failure for each patient.
Patient 1: deep CMV stromal infection in the third failed graft.
Previous graft also showed CMV stromal infection that was
unrecognised
Patient 1, an elderly Chinese man underwent a left eye Boston
keratoprosthesis (K-Pro) in 2011. Previously, he had bilateral
uncomplicated cataract surgery. No Fuchs’ endothelial corneal
dystrophy was noted. He developed pseudophakic bullous kera-
topathy in the left eye in 2001, and he underwent an uncompli-
cated left PK in 2003. He also developed glaucoma in the left
eye. A total of three left PKs were performed before K-Pro
surgery when each graft showed signs of ‘endothelial rejection’
and failed despite systemic immunosuppression. Postoperatively,
he showed no signs of inflammation in the residual cornea
around the K-Pro and the anterior chamber at 28 months of
follow-up (figure 1D), despite no treatment for the CMV infec-
tion. Retrospective review of his corneal tissue from his second
PK (table 1) in 2007 showed CMV infection that was unrecog-
nised (figure 1G–I). Unfortunately, tissue from his first PK was
not sent for histology analysis (table 1).
Patient 2: CMV anterior stromal infection at graft–host interface
in fourth corneal graft button
Patient 2, a middle-aged Malay man underwent a right eye
Boston K-Pro procedure in 2012 (figure 2). A previous history of
bilateral penetrating ocular trauma with surgical repair left him
phthisical in the left eye and aphakic in the right. He developed
aphakic bullous keratopathy with secondary glaucoma, and a
right PK was performed. The graft developed ‘endothelial rejec-
tion’ and failed. A total of four PKs in the right eye (table 1) for
presumed endothelial graft rejection (figure 2A) and graft failure,
despite systemic immunosuppression, were performed before the
Boston K-Pro procedure. Postoperatively, he was not treated for
the CMV stromal infection as he showed no signs of inflamma-
tion in the corneal bed around the K-Pro and his anterior
chamber up to 20 months of follow-up (figure 2D).
Retrospective review of his previous failed graft buttons in 2008,
2009 and 2010 showed no evidence of CMV infection on hist-
ology and IHC.
Patient 3: deep stromal CMV infection with stromal
and epithelial keratitis
Patient 3, an elderly Chinese man underwent a second PK
surgery in 2013. Previously, he had undergone a left phacoemul-
sification surgery in 2010. He presented with corneal oedema
and glaucoma 4 months after his cataract surgery, and was
found to have epithelial dendrites. A presumptive diagnosis of
herpes simplex viral (HSV) keratitis was made. Despite treat-
ment with systemic acyclovir, the cornea decompensated, and
he underwent a PK. Five months postoperatively, the graft
showed partial corneal oedema with pigmented KPs, suggestive
of allograft rejection and subsequently failed despite full topical
immunosuppression and antiviral cover (figure 3A). He under-
went a second PK in 2013 (table 1). Intraoperatively, aqueous
humour had been sent for CMV, HSV, varicella-zoster virus and
toxoplasma real time (RT)-PCR analysis, and it returned positive
for CMV only. He was treated with oral valganciclovir 900 mg
twice daily, but defaulted after 2 weeks of therapy. Two weeks
later, the CMV endotheliitis recurred, but resolved after
 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com

Clinical science

Figure 1 Patient 1: (A) Patient 1’s third failed graft with cornea oedema and stromal haze prior to keratoprosthesis (K-pro) surgery. (B) H&E stain,
×10 showing third graft button with viral cytopathic changes, seen predominantly in the posterior stroma, but extending to the mid-stroma. Note
the lack of stromal vascularisation and inflammatory infiltrate (arrow). (C) H&E stain, ×40 showing cells with cytomegaly and cytoplasmic
eosinophilic inclusions. The Descemet’s membrane is intact and shows loss of endothelial cells. (D) Postoperatively, Boston K-pro remains clear
without evidence of inflammation at 28 months’ follow-up. (E) Cytomegalovirus (CMV) immunohistochemistry (IHC) (brown diaminobenzidine (DAB),
×10) showing that the CMV infection extends to the anterior stroma. (F) Dual IHC (×40) with macrophage marker CD163 (brown DAB stain) and
CMV (red chromogen) showing the CD163 macrophage (arrows) adjacent to the CMV-infected keratocytes (stained with red chromogen). (G) Earlier
failed graft in 2007 shows pigmented keratic precipitates, Descemet’s folds and oedema. (H) Histology of this graft button also shows the
eosinophilic cytoplasmic inclusions and cytomegaly, which are positive for CMV antibody on IHC (insert, red chromogen). (I) Dual IHC with CMV (red
chromogen) with CD68 macrophage (brown DAB) demonstrates CMV endotheliitis (red) and CD163-positive macrophages in the stroma (arrow).

restarting systemic valganciclovir. His graft remained clear at
10 months of follow-up (figure 3D). Retrospective histological
review of his previous graft did not show evidence of CMV
stromal infection.
DISCUSSION
CMV—a member of the Herpesviridae family—is a double-
stranded DNA virus.9 10 Recently, it has been recognised as a
cause of primary corneal decompensation and mimics graft
rejection.4 8 Our current study revealed the mean incidence of
CMV graft infections mimicking corneal graft rejection, by the
retrospective review of histology, to be 6.3% per year. The dif-
ference in number of regrafts between the patients with
CMV-infected grafts and non-CMV-infected grafts was statistic-
ally significant. A higher number of regrafts of ≥3 may be useful
to highlight the possibility of a CMV-related graft infection;
however, the authors emphasise that increased awareness of this
entity and preoperative diagnosis with aqueous PCR analysis is
preferred over the reliance on number of regrafts.
Histological features of CMV graft infection and its
clinical significance
Confocal studies have suggested that CMV infection causes
‘owl’s eyes’ inclusions in the endothelium, but hardly in the
corneal stroma. Furthermore, these findings have not been cor-
related with histology.14 15 In our study, only 1 of 53 grafts
(1.9%) showed CMV endotheliitis (figure 1I), detected only on
retrospective review of patient 1’s previous graft, and it likely
reflects the advanced endothelial cell loss by the time a surgical
intervention is required.
The histological features of CMV-related corneal infection
have not been described previously. Advanced CMV corneal
infection resulting in graft failure in our study was characterised
by (1) stroma keratocytes with cytopathic changes, (2) lack of
acute and chronic inflammation and (3) absence of vascularisa-
tion. These features have not been previously reported by con-
focal or other studies.5 7 15–17
The stromal distribution of the CMV-infected keratocytes was
seen predominantly in deep posterior stroma and next to
Descemet’s membrane (patients 1 and 3) and at the anterior
graft–host stromal interface in patient 2. The posterior distribu-
tion of CMV-infected cells is not surprising since CMV is
thought to infect the endothelial cells causing endotheliitis,
which is supported by the histological evidence of endotheliitis
in patient 1’s previous graft (figure 1I). In contrast, the CMV
infection at the donor–host interface (figure 2C) raises the suspi-
cion that CMV infection could possibly arise from the donor
cornea. Finding CMV-infected cells throughout the stroma is

Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390 3

 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com

Clinical science

Figure 2 Patient 2: (A) Patient 2’s fourth failed graft with corneal oedema and Descemet’s folds prior to keratoprosthesis (K-pro) surgery.
(B) Histology of fourth corneal graft (×10, H&E stain) showing viral cytopathic changes in the anterior stroma at the graft–host interface. Note the
lack of stromal vascularisation and inflammatory infiltrate (arrow). (C) Higher magnification of the cells with cytomegaly and cytoplasmic
eosinophilic inclusions (×40, H&E stain) at the graft–host interface, under the Bowman’s layer and fibrous pannus. (D) Postoperatively, Boston K-pro
remains clear without evidence of inflammation at 20 months’ follow-up. (E) Cytomegalovirus (CMV) immunohistochemistry (IHC) (brown DAB, ×10)
positive staining confirms the CMV infection at the graft–host junction. (F) Dual IHC with CMV (red chromogen) with CD163 macrophage (brown
DAB) demonstrates CD163-positive macrophages (arrows) adjacent to CMV-infected keratocytes (red) in the stroma.

also significant as it may serve as a reservoir for early activation
of CMV endotheliitis in Descemet’s stripping automated endo-
thelial keratoplasty (DSAEK) surgery since this CMV-infected
stroma would not be removed. As suggested by our finding of
only 1/53 grafts with CMV endotheliitis, histological analysis of
Descemet’s membrane may not be useful for detecting CMV
infection. In cases of recurrent DSAEK failure from presumed
endothelial rejection, presumed late endothelial failure or from
an unexplained sudden drop in postoperative endothelial cell
density from unknown causes, it might be prudent to exclude
CMV infection with RT-PCR aqueous analysis prior or at the
time of repeat surgery.4
Clinically, vascularisation is not seen in CMV-infected grafts,8
which we confirmed on histology. This lack of vessels and vascu-
lar endothelial cells within the cornea suggests that the patho-
genesis of corneal CMV infection is different from CMV
retinitis. In CMV retinitis, the virus has been shown to infect
vascular endothelial cells leading to infection of the surrounding
glial, neuronal cells and retinal pigment epithelium.18
M2 macrophages have been suggested to be more susceptible
to CMV infections.9 We used CD163 M2 macrophage and pan-
macrophage CD68 markers to colocalise with CMV antibodies
to determine if macrophages were also infected by CMV cells,
but no colocalisation was detected (figures 1–3F). Only occa-
sional CD163 M2 macrophages were detected adjacent to the
CMV-infected keratocytes (figures 1–3F). CD163 M2 macro-
phages are known to favour a Th2 immunosuppressive inflam-
matory response.19 Their presence and the lack of other
inflammatory cells (neutrophils, lymphocytes or plasma cells)
suggest an immunocompromised milieu (figures 1–3B) that may
contribute to the activation or persistence of CMV in the
4 Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390
cornea. This correlates with our previous findings that local
immunosuppression with topical corticosteroids predisposes to
CMV activation.5 8 16
Clinical features of CMV graft infections in our patients
Primary corneal CMV infection rarely mimics HSV keratitis and
presents with epithelial dendrites as seen in our third patient,
and is the likely cause for his corneal decompensation and sub-
sequent graft failures. This highlights that cases of ‘presumed
HSV’, which recur despite adequate antiviral coverage, could be
due to CMV infection. Our first two patients had no documen-
ted inflammation or KPs during initial presentation, suggesting
that a primary CMV infection was unlikely. No other signs were
present to alert the clinician of a CMV infection. Histology may
be the only method to detect CMV graft infection postopera-
tively as CMV infection is currently only diagnosed with
RT-PCR analysis of aqueous fluid for CMV DNA prior or at the
time of corneal surgery.3 17 20 The lack of signs of primary
CMV infection and the detection of CMV after graft surgery
seem to suggest that the CMV may be acquired. Yet, when we
found CMV infection in patient 1’s first donor graft removed
during the second PK, the duration between the grafting of that
donor cornea to rejection was 36 months (table 1) in compari-
son with the much shorter time to rejection of 3 months after
the second PK (when the presence of CMV was missed).
Similarly for patient 2, the time from surgery to rejection for his
fourth graft was 9 months, despite the presence of CMV in this
removed donor graft. This suggests a possibility that CMV may
not be acquired from the donor graft, and may arise from reacti-
vation of a latent host infection. Although histologically, anter-
ior stromal involvement by CMV is seen, there is a gradient of
 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com

Clinical science

Figure 3 Patient 3: (A) Patient 3’s second failed graft with opaque cornea prior to penetrating keratoplasty (PK). (B) Histology of the second failed
corneal graft (×10, H&E stain) showing viral cytopathic changes. Note the lack of stromal vascularisation and inflammatory infiltrate (arrow). The
Descemet’s membrane is intact and shows loss of endothelial cells. (C) Higher magnification of the cells with cytomegaly and cytoplasmic
eosinophilic inclusions (×40, H&E stain) concentrated in the posterior stroma. (D) Postoperatively, the PK remains clear without evidence of
inflammation at 4 months’ follow-up. (E) Cytomegalovirus (CMV) immunohistochemistry (red chromogen, ×10) confirming the CMV infection within
the posterior stroma up to the mid-stroma. (F) Dual staining (macrophage marker CD163, brown DAB stain and CMV, red chromogen ×40) shows
the close association of the CD163 macrophage (arrow) with the CMV-infected keratocytes (red). (G–I) Electron microscopy showing cytoplasmic
capsids in the cytoplasm and nucleus (G). Nuclear viral capsids (H) and targetoid viral capsids in the cytoplasm (I).

distribution with the epicentre predominantly at the edge of the
graft or involving the posterior stroma/Descemet’s membrane
(figure 1), suggesting that the CMV infection is arising from the
host cornea or from the anterior chamber towards the centre of
the donor cornea. Although the authors favour the host reacti-
vation theory, it has yet to be determined if the virus can be
acquired via the donor or is due to the reactivation of a host
CMV infection. For corneal transplantation, the CMV status of
the donor is not routinely evaluated preoperatively as donor
CMV serology positivity is not a contraindication. Latent CMV
infection is endemic in Singapore, with a seropositivity rate of
87%.21 Our study suggests that further studies evaluating the
donor cornea for CMV involvement as well as the donor and
recipient CMV serology would be necessary to determine if
similar CMV serological surveillance as in solid organ trans-
plants should be performed for corneal transplants.

Table 1 Summary of graft surgeries and time to graft rejection/failure

Time to Time to Time to Time to
rejection rejection rejection rejection Fifth Boston
and failure and failure and failure Fourth PK/Boston and failure K-Pro
Case First PK (months) Second PK (months) Third PK (months) K-Pro procedure (months) procedure

1 Host cornea 36 Donor cornea from 3 Donor cornea from 4 Donor cornea from
not sent for first second PK not sent third
histology PK-CMV-positive for histology PK-CMV-positive
2 Host cornea 84 Donor cornea from 10 Donor cornea from 10 Donor cornea from 9 Donor cornea
not sent for first second third from fourth PK-
histology PK-CMV-negative PK-CMV-negative PK-CMV-negative CMV-positive
3 Host 5 Donor cornea from
histology first
negative for PK-CMV-positive
CMV
CMV, cytomegalovirus; K-Pro, keratoprosthesis; PK, penetrating keratoplasty.

Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390 5

 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com
Clinical science
Clinically, all three patients had glaucoma at the time of graft
rejection or prior to the initial surgery. Although post kerato-
plasty, glaucoma is not uncommon, a raised intraocular pressure
(IOP) is often associated with an active CMV infection,5 and
hence, a sudden rise in IOP associated with ‘rejection’ or its
presence prior to graft surgery may be useful in raising suspicion
for CMV.
Postoperatively, the K-Pros and graft remained clear (figures
1–3D). Both patients with K-Pro surgery did not require anti-
viral therapy. This is not surprising as a good visual outcome in
K-Pro surgery is predicated by the optics of the plastic kerato-
prosthesis. The emerging use of the Boston K-Pro worldwide in
recent years, and the good outcomes of our two patients are,
therefore, encouraging as a possible solution to corneal graft
patients with active CMV infection. However, the long-term
risks (device retention, secondary infection) related to the
Boston K-Pro need to be considered.22 In contrast, when a PK
was performed (in our third patient), systemic valganciclovir is
essential as demonstrated by the reactivation of CMV when
treatment was stopped prematurely. Currently, treatment of
CMV-related graft rejections are comanaged with infection
disease physician, and consists of 3 weeks of oral valganciclovir
(900 mg twice daily) followed by a 3-week maintenance dose of
900 mg every morning. After 6 weeks of therapy, a repeat
aqueous tap is performed. Cessation may be considered if it is
negative, although we often continue treatment with topical val-
ganciclovir.4 If positive, the treatment is continued for another
6 weeks before a repeat aqueous tap is performed. Our third
patient was on systemic valganciclovir for a year prior to
cessation.
In summary, this paper highlights the histology of CMV graft
infections, which would be useful for the postoperative diagno-
sis of unsuspected infections. CMV should be excluded in recur-
rent graft failures especially in DSAEK/Descemet’s membrane
endothelial keratoplasty as there may be a stromal reservoir of
infected cells.
Acknowledgements The authors would like to thank Dr Alwin Loh and
Dr Norman Chan from the Singapore General Hospital Department of Pathology for
their contributions on the electron microscopy work.
Contributors Design of the study: ASYC and DTHT; conduct of the study and data
collection: ASYC, IAJ and JI; management: DTHT, JSM and AA; analysis and
interpretation of data: ASYC, IAJ, JSM and DTHT; review of manuscript: JSM,
AA and DTHT; approval of manuscript: JI, IAJ, JSM and DTHT.
Funding This study was supported by a grant from the SingHealth Research
Foundation (SHF/FG526S/2011).
Competing interests None declared.
Ethics approval SingHealth Centralised Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
6 Chan ASY, et al. Br J Ophthalmol 2015;0:1–6. doi:10.1136/bjophthalmol-2015-307390
Data sharing statement Requests for data sharing can be made to the
corresponding author.
REFERENCES
1 Ramanan P, Razonable RR. Cytomegalovirus infections in solid organ
transplantation: a review. Infect Chemother 2013;45:260–71.
2 Wehrly SR, Manning FJ, Proia AD, et al. Cytomegalovirus keratitis after penetrating
keratoplasty. Cornea 1995;14:628–33.
3 Sonoyama H, Araki-Sasaki K, Osakabe Y, et al. Detection of cytomegalovirus DNA
from cytomegalovirus corneal endotheliitis after penetrating keratoplasty. Cornea
2010;29:683–5.
4 Ang M, Sng CC, Chee SP, et al. Outcomes of corneal transplantation for irreversible
corneal decompensation secondary to corneal endotheliitis in Asian eyes. Am J
Ophthalmol 2013;156:260–6.e2.
5 Anshu A, Chee SP, Mehta JS, et al. Cytomegalovirus endotheliitis in Descemet’s
stripping endothelial keratoplasty. Ophthalmology 2009;116:624–30.
6 Shimazaki J, Harashima A, Tanaka Y. Corneal endotheliitis with cytomegalovirus
infection of corneal stroma. Eye (Lond) 2010;24:1105–7.
7 Koizumi N, Suzuki T, Uno T, et al. Cytomegalovirus as an etiologic factor in corneal
endotheliitis. Ophthalmology 2008;115:292–7.e3.
8 Chee SP, Jap A, Ling EC, et al. Cytomegalovirus-positive corneal stromal edema
with keratic precipitates after penetrating keratoplasty: a case-control study. Cornea
2013;32:1094–8.
9 Bayer C, Varani S, Wang L, et al. Human cytomegalovirus infection of M1 and M2
macrophages triggers inflammation and autologous T-cell proliferation. J Virol
2013;87:67–79.
10 Sinclair J, Sissons P. Latent and persistent infections of monocytes and
macrophages. Intervirology 1996;39:293–301.
11 Tan WJ, Thike AA, Bay BH, et al. Immunohistochemical expression of
homeoproteins Six1 and Pax3 in breast phyllodes tumours correlates with
histological grade and clinical outcome. Histopathology 2014;64:807–17.
12 Tan LH, Tan SY, Tang T, et al. Relationship between atypical T- and B-cell size
predicts survival in peripheral T-cell lymphomas with large B-cells. Pathology
2013;45:28–37.
13 Harada O, Hoe R, Lin J, et al. Intranuclear inclusions in epithelial cells of benign
proliferative breast lesions. J Clin Pathol 2011;64:776–80.
14 Shiraishi A, Hara Y, Takahashi M, et al. Demonstration of “owl’s eye” morphology
by confocal microscopy in a patient with presumed cytomegalovirus corneal
endotheliitis. Am J Ophthalmol 2007;143:715–17.
15 Kobayashi A, Yokogawa H, Higashide T, et al. Clinical significance of owl eye
morphologic features by in vivo laser confocal microscopy in patients with
cytomegalovirus corneal endotheliitis. Am J Ophthalmol 2012;153:445–53.
16 Chee SP, Bacsal K, Jap A, et al. Corneal endotheliitis associated with evidence of
cytomegalovirus infection. Ophthalmology 2007;114:798–803.
17 Suzuki T, Hara Y, Uno T, et al. DNA of cytomegalovirus detected by PCR in aqueous
of patient with corneal endotheliitis after penetrating keratoplasty. Cornea
2007;26:370–2.
18 Read RW, Zhang JA, Ishimoto SI, et al. Evaluation of the role of human retinal
vascular endothelial cells in the pathogenesis of CMV retinitis. Ocul Immunol
Inflamm 1999;7:139–46.
19 Locati M, Mantovani A, Sica A. Macrophage activation and polarization as an
adaptive component of innate immunity. Adv Immunol 2013;120:163–84.
20 Koizumi N, Yamasaki K, Kawasaki S, et al. Cytomegalovirus in aqueous humor from
an eye with corneal endotheliitis. Am J Ophthalmol 2006;141:564–5.
21 Wong A, Tan KH, Tee CS, et al. Seroprevalence of cytomegalovirus, toxoplasma and
parvovirus in pregnancy. Singapore Med J 2000;41:151–5.
22 Robert MC, Pomerleau V, Harissi-Dagher M. Complications associated with Boston
keratoprosthesis type 1 and glaucoma drainage devices. Br J Ophthalmol
2013;97:573–7.
 Downloaded from http://bjo.bmj.com/ on November 14, 2015 - Published by group.bmj.com

Histological features of Cytomegalovirus

-related corneal graft infections, its
associated features and clinical significance
Anita SY Chan, Jodhbir S Mehta, Issam Al Jajeh, Jabed Iqbal, Arundhati
Anshu and Donald TH Tan

Br J Ophthalmol published online August 20, 2015

Updated information and services can be found at:

http://bjo.bmj.com/content/early/2015/08/20/bjophthalmol-2015-30739
0

These include:

References This article cites 22 articles, 3 of which you can access for free at:
http://bjo.bmj.com/content/early/2015/08/20/bjophthalmol-2015-30739
0#BIBL
Email alerting Receive free email alerts when new articles cite this article. Sign up in the
service box at the top right corner of the online article.

Topic Articles on similar topics can be found in the following collections

Collections Ophthalmologic surgical procedures (1163)
Angle (962)
Glaucoma (946)
Intraocular pressure (959)

Notes

To request permissions go to:

http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to:

http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to:

http://group.bmj.com/subscribe/