Color Fundus Photography, Optical Coherence

Tomography, and Fluorescein Angiography in

Diagnosing Polypoidal Choroidal Vasculopathy

VORAPORN CHAIKITMONGKOL, PREEYANUCH KHUNSONGKIET, DIREK PATIKULSILA,

MANSING RATANASUKON, NAWAT WATANACHAI, CHAISIRI JUMROENDARARASAME,
CATHERINE B. MAYERLE, IAN C. HAN, CONNIE J. CHEN, PAWARA WINAIKOSOL,
CHUTIKARN DEJKRIENGKRAIKUL, JANEJIT CHOOVUTHAYAKORN, PARADEE KUNAVISARUT, AND
NEIL M. BRESSLER

PURPOSE: To determine sensitivity and specificity of 0.81 (95% CI: 0.73–0.88), 0.83 (95% CI: 0.76–0.90),
polypoidal choroidal vasculopathy (PCV) diagnosis using 0.79 (95% CI: 0.73–0.87), 0.82 (95% CI: 0.74–0.88).
color fundus photography (CFP), optical coherence to- Discrepancies between Thai and US graders existed
mography (OCT), and fundus fluorescein angiography through sets A, C, and D.
(FFA) without indocyanine green angiography (ICGA).
CONCLUSIONS: These data suggest that without

DESIGN: Validity analysis. ICGA, fundus photography combined with OCT pro-

METHODS: Treatment-naı̈ve eyes with serous/serosan- vides high sensitivity and high specificity to diagnose
guinous maculopathy undergoing CFP, OCT, FFA, and PCV; adding FFA does not improve accuracy. (Am J
ICGA imaging before treatment at a university hospital Ophthalmol 2018;192:77–83. Ó 2018 Elsevier Inc. All
in Thailand (January 1, 2013 to June 30, 2015) were rights reserved.)
identified. Images of each subject were categorized into

P
4 sets (set A: CFP; set B: CFPDOCT; set C: OLYPOIDAL CHOROIDAL VASCULOPATHY (PCV) IS AN
CFPDFFA; set D: CFPDOCTDFFA). Six graders, 3 important cause of central vision loss among indi-
from Thailand (PCV endemic area) and 3 from the viduals of African and Asian descent.1–5 It is still
United States (nonendemic area), individually reviewed controversial whether PCV is a subtype of neovascular
each set (without ICGA), and determined if the presumed age-related macular degeneration (NVAMD) or a different
diagnosis was PCV. In parallel, 2 other graders confirmed disease entity. The 2 conditions share similar presenting
if each case had PCV or not using EVEREST criteria characteristics, including submacular fluid, exudates, or
(including ICGA). Sensitivity and specificity of a PCV hemorrhage.6–9 If left untreated, both can lead to
diagnosis with each set (without ICGA) were analyzed permanent visual loss. However, it is important to
compared with diagnoses including ICGA. differentiate PCV from typical NVAMD because their

RESULTS: Of 119 study eyes (113 subjects, 57% male, natural history, treatment response, and prognosis are
mean age ± SD 59.9 ± 13.8 years), definite PCV diag- different.
nosis was 40.3%. Sensitivity of sets A, B, C, D: 0.63 Indocyanine green angiography (ICGA), a vascular
(95% confidence interval [CI]: 0.47–0.76), 0.83 (95% imaging technique that allows physicians to visualize
CI: 0.69–0.92), 0.54 (95% CI: 0.39–0.68), 0.67 (95% choroidal vasculature and its abnormality, has been
CI: 0.51–0.79); specificities: 0.93 (95% CI: 0.84– accepted as a gold standard to diagnose PCV.9 The charac-
0.97), 0.83 (95% CI: 0.72–0.91), 0.97 (95% CI: teristics of PCV on ICGA include the presence of hyper-
0.89–0.99), 0.92 (95% CI: 0.82–0.97); accuracies: fluorescent polypoidal vascular filling with or without
branching vascular network (BVN).10 Despite its advan-
tages, ICGA is an invasive imaging procedure. It requires
Accepted for publication May 3, 2018. special dye and equipment, which are available only at
From the Retina Division, Department of Ophthalmology, Chiang Mai large ophthalmic institutions. Moreover, ICGA is contra-
University, Chiang Mai, Thailand (V.C., P.K., D.P., N.W., J.C., P.K.); indicated in patients with impaired kidney or liver func-
Retina Division, Department of Ophthalmology, Prince of Songkla
University, Songkhla, Thailand (M.R.); Retina Division, Department of tions, those with history of allergy to iodine-based
Ophthalmology, Phramongkutklao College of Medicine, Bangkok, contrast dye, and those during pregnancy. Therefore,
Thailand (C.J.); Retina Division, Wilmer Eye Institute, Johns Hopkins ICGA is not always a practical tool to assist physicians to
University School of Medicine, Baltimore, Maryland, USA (C.B.M.,
I.C.H., N.M.B.); Virginia Mason Medical Center, Seattle, Washington, diagnose PCV, especially those working in countries with
USA (C.J.C.); and Department of Ophthalmology, Faculty of Medicine, limited resources, including Thailand.
Chiang Mai University, Chiang Mai, Thailand (P.W., C.D.). In clinical practice, there are some clinical clues suggest-
Inquiries to Neil M. Bressler, Retina Division: Maumenee 752, Johns
Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21287-9277; e-mail: ing the diagnosis of PCV, including the presence of subre-
nmboffice@jhmi.edu tinal orange nodules or massive subretinal hemorrhage on

0002-9394/$36.00 © 2018 ELSEVIER INC. ALL RIGHTS RESERVED. 77

https://doi.org/10.1016/j.ajo.2018.05.005
FIGURE 1. Flowchart of study methods. *A total of 135 cases were excluded owing to receiving previous treatments (88), poor image
quality (20), incomplete imaging (missing 4 imaging types) (17), and presence of coexisting vision-threatening conditions (10). The
EVEREST criteria for polypoidal choroidal vasculopathy (PCV) diagnosis include the presence of indocyanine green angiography
(ICGA) hyperfluorescence (appearing within the first 6 minutes of ICGA dye injection) and at least 1 of the following diagnostic
criteria: nodular appearance of the polyp on stereoscopic viewing, hypofluorescent halo around the nodule, abnormal vascular chan-
nel(s) supplying the polyps, pulsatile filling of polyps, orange subretinal nodules corresponding to the hyperfluorescent area on ICGA,
and massive submacular hemorrhage. CFP [ color fundus photography; FFA [ fundus fluorescein angiography; NPV [ negative
predictive value; OCT [ optical coherence tomography; PPV [ positive predictive value; US [ United States.

color fundus photography (CFP), the presence of thumb- receiver operating characteristic curve [AUC]) of using
like pigment epithelial detachment (PED), notched PED, CFP, OCT, and FFA in diagnosing PCV without ICGA
or double layer signs on optical coherence tomography information, and to determine if there is any difference in
(OCT), or the presence of occult choroidal neovasculariza- the results when images are evaluated by graders working
tion on fundus fluorescein angiography (FFA). When some inside vs outside a PCV endemic area.
of these clues are identified on clinical examination, a pro-
visional diagnosis of PCV often can be made, even without
ICGA information.
A few studies have reported the high sensitivity and high METHODS
specificity of OCT in detecting PCV.10–13 However, when
evaluating patients, physicians usually get information from THIS WAS A VALIDITY ANALYSIS OF PATIENTS WITH NEWLY
fundus examination along with OCT rather than OCT diagnosed serous or serosanguinous maculopathy, including
alone. Physicians sometimes also use FFA, an imaging PCV, NVAMD, or central serous chorioretinopathy
tool that is more available worldwide than ICGA. (CSC) in 1 or both eyes seen at Chiang Mai University
Therefore, this study aims to determine sensitivity, Hospital between January 1, 2013 to June 30, 2015
specificity, and predictive accuracy (area under the (2.5-year period). The study was prospectively approved

78 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2018

FIGURE 2. Example image from each imaging set evaluated by graders: set A (color fundus photography; CFP), set B (CFP and op-
tical coherence tomography; OCT), set C (CFP and fundus fluorescein angiography; FFA), and set D (CFP, OCT, and FFA). Note
that graders had an access to review multiple scans of OCT and multiple FFA images in early, mid, and late phases. This case example
demonstrates typical PCV findings. CFP shows transudates or an exudative maculopathy with a pigment epithelial detachment (PED,
arrow); OCT shows a double layer sign (black arrows), sharply peaked PED (white arrows), and a hyperreflective ring underneath the
PED (asterisk); and a late-phase frame of a FFA shows a pattern of occult choroidal neovascularization.

by the Research Ethics Committee, Faculty of Medicine, no official grading standardization was performed, as part
Chiang Mai University and adhered to the Declaration of of the objectives was to determine differences, if they
Helsinki. Individuals who underwent 4 imaging types, existed, among graders from inside vs outside an endemic
including CFP (VX-10i, Kowa Company Ltd., Nagoya, area of PCV.
Japan), spectral-domain OCT (Spectralis; Heidelberg En- For the images sent to graders, CFP included images of
gineering Inc., Heidelberg, Germany), and simultaneous both the study eye and fellow eye, OCT included at least
FFA and ICGA (HRA Spectralis; Heidelberg Engineering 25 cross-sectional B-scan images of the study eye, and
Inc., Heidelberg, Germany) at their first presentation prior FFA included multiple images from early, mid, and late
to any treatment were identified. Patients with coexisting phases of the study eye and fellow eye. Majority rule was
retinal abnormalities or those with ocular media opacity applied to determine the group decision of graders. Opin-
preventing adequate image quality for analysis were ions from at least 4 of 6 graders were determined as the final
excluded (Figure 1). results of all graders, and opinions of at least 2 of 3 graders
De-identified images of each subject were classified into 4 were determined as final results from each subgroup of
sets: set A, CFP only; set B, CFP and OCT; set C, CFP and graders.
FFA; and set D, CFP, OCT, and FFA (Figure 2). Images of In parallel, 4 types of retinal imaging (CFP, OCT, FFA,
each set were randomly sequenced and sent to 6 retinal and ICGA) of each case were de-identified and sent to
specialist graders (3 from Thailand [PCV endemic area]: another 2 retinal specialist graders (D.P., V.C.; from
M.S., N.W., C.J.; 3 from the United States [US, outside Thailand, PCV endemic area) to determine whether the
PCV endemic area]: C.B.M., C.J.C., I.C.H.) to review, inde- definite diagnosis of each case was PCV, according to the
pendently, and determine if their provisional diagnosis of PCV definition used in the EVEREST study—that is, the
each case was PCV or not. Examples of clinical clues for presence of subretinal focal ICGA hyperfluorescence
PCV on color fundus photography include subretinal orange within 6 minutes after injection of ICGA and at least 1 of
nodule, hemorrhagic or fibrovascular PED, massive subreti- the following additional criteria: (1) association with
nal hemorrhage, peripapillary location, multifocal lesions, abnormal vascular channel supplying the polyps, (2) pres-
or no large drusen in the fellow eye; on OCT, sharply peaked ence of pulsatile polyps, (3) nodular appearance in stereo-
RPE protrusion, notched PED, hyperreflective ring sur- scopic CFP, (4) presence of hypofluorescent halo around
rounding hyporeflective halo underneath the PED, or the nodules (in first 6 minutes), (5) orange subretinal nod-
double-layer sign; and on FFA, an occult CNV leakage ules in stereoscopic CFP corresponding to ICGA lesion, or
pattern. Graders might use this information in combination (6) association with massive submacular hemorrhage (at
with their own background knowledge or experience to least 4 disc areas).10 Any disagreement between these 2
determine whether each case was PCV or not. Note that graders was reviewed and discussed using open adjudication.

VOL. 192 FUNDUS PHOTO, OCT, & FA FOR POLYPOIDAL CHOROIDAL VASCULOPATHY 79
 high PPV (0.77; 95% CI: 0.63–0.87), high NPV (0.88;
TABLE 1. Demographic and Clinical Characteristics of Study 95% CI: 0.77–0.94), and highest AUC (0.83; 95% CI:
Eyes 0.76–0.90). Imaging set C (CFP and FFA) provided the
highest specificity (0.97; 95% CI: 0.89–0.99) and highest
Study Patients Results (N ¼ 116 Patients)
PPV (0.93; 0.75–0.99), but lowest sensitivity (0.54; 95%
Age, mean 6 SD (y) 59.9 6 13.8 CI: 0.39–0.68), lowest NPV (0.76; 95% CI: 0.66–0.84),
Ethnicity, n (%) and lowest AUC (0.76; 95% CI: 0.68–0.83). Imaging set
Thai 115 (99.1%) A (CFP only) and set D (CFP, OCT, and FFA) provided
Hill tribe 1 (0.9%)
results ranking between imaging set B and C. Compared
Sex, n (%)
to set B, imaging set A (CFP only) provided higher speci-
Male 66 (56.9%)
Female 50 (43.1%)
ficity (0.93; 95% CI: 0.84–0.97) and higher PPV (0.86;
Laterality, n (%) 95% CI: 0.69–0.95), but lower sensitivity (0.63; 95% CI:
Unilateral 113 (97.4%) 0.47–0.76), lower NPV (0.79; 95% CI: 0.68–0.86), and
Bilateral 3 (2.6%) lower AUC (0.78; 95% CI: 0.70–0.85). Imaging set D
(CFP, OCT, and FFA) provided higher specificity (0.92;
Study Eyes Results (N ¼ 119 Eyes)
95% CI: 0.82–0.97) and higher PPV (0.84; 95% CI:
Diagnosis, n (%) 0.68–0.93), but lower sensitivity (0.67; 95% CI: 0.51–
PCV (not typical NVAMD) 48 (40.3%)
0.79), lower NPV (0.80; 95% CI: 0.70–0.88), and lower
CSC 35 (29.4%)
AUC (0.79; 95% CI: 0.72–0.86) (Table 2).
Typical NVAMD 34 (28.6%)
For subgroup analysis, there were some discrepancies be-
Othersa 2 (1.7%)
tween results from the Thai and US graders. Results from
CSC ¼ central serous chorioretinopathy; NVAMD ¼ neovascu- the US graders showed relatively higher sensitivity and
lar age-related macular degeneration; PCV ¼ polypoidal higher NPV across all imaging sets, while results from
choroidal vasculopathy. Thai graders showed relatively higher specificity and higher
a
Other diagnoses included serous pigment epithelial detach- PPV across all imaging sets. These differences were
ment and retinal angiomatous proliferation. apparent in imaging sets A, C, and D but no differences
were apparent in imaging set B. For the US subgroup, im-
aging set B (CFP and OCT) achieved the highest value
A provisional diagnosis of each case, in each imaging set across all analyses: sensitivity (0.90; 95% CI: 0.81–0.98),
(without ICGA), made by 6 graders was then compared specificity (0.76; 95% CI: 0.66–0.86), PPV (0.72; 95%
with the definite diagnosis of each case (using ICGA infor- CI: 0.60–0.83), NPV (0.92; 95% CI: 0.84–0.99), and
mation). Sensitivity, specificity, positive predictive value AUC (0.82; 95% CI: 0.75–0.88) for PCV diagnosis. For
(PPV), negative predictive value (NPV), and predictive the Thai subgroup, imaging set B (CFP and OCT) achieved
accuracy (AUC) in PCV diagnosis of each imaging set the highest sensitivity (0.81; 95% CI: 0.70–0.93), highest
were analyzed. A flowchart of study methods is shown in NPV (0.86; 95% CI: 0.78–0.95), and highest AUC (0.80;
Figure 1. 95% CI: 0.73–0.87), while imaging set C provided the
highest specificity (0.99; 95% CI: 0.96–1.00) and highest
PPV (0.94; 95% CI: 0.82–1.00). However, for predictive
accuracy (AUC), there was no obvious difference between
RESULTS results from Thai vs US graders across all imaging sets
(Table 3).
TWO HUNDRED RECORDS WERE REVIEWED, AND 116 CASES
(119 eyes) met the inclusion criteria. One hundred
thirty-five cases were excluded owing to having a history
of previous treatments in the study eye (88), poor image DISCUSSION
quality (20), incomplete imaging (missing all 4 types)
(17), or presence of coexisting vision-threatening condi- INDOCYANINE GREEN ANGIOGRAPHY IS CONSIDERED THE
tions (10). Of 116 participants, all were Asian, 99% were gold standard for PCV diagnosis. However, it is an invasive
Thai, and 2.7% had bilateral study eyes. Of 119 study procedure that requires special equipment and is not avail-
eyes, definite diagnoses were PCV in 40.3%, central serous able at every eye center. Color fundus photography, optical
chorioretinopathy in 29.4%, typical NVAMD in 28.6%, coherence tomography, and fundus fluorescein angiog-
and others in 1.7% (Table 1). raphy are more common retinal imaging tools, but little
Among the imaging sets and all graders, imaging set B is known about whether information from these images
(CFP and OCT) provided the best overall results, including without ICGA could help physicians to diagnose PCV.
the highest sensitivity (0.83, 95% confidence interval [CI]: This study evaluates the sensitivity, specificity, and pre-
0.69–0.92), high specificity (0.83; 95% CI: 0.72–0.91), dictive accuracy (AUC) in PCV diagnosis by using fundus

80 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2018

 TABLE 2. Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Value, and Accuracy of Each Imaging Set

Set A (CFP only) Set B (CFPþOCT) Set C (CFPþFFA) Set D (CFPþOCTþFFA)

Sensitivity (95% CI) 0.63 (0.47–0.76) 0.83 (0.69–0.92) 0.54 (0.39–0.68) 0.67 (0.51–0.79)
Specificity (95% CI) 0.93 (0.84–0.97) 0.83 (0.72–0.91) 0.97 (0.89–0.99) 0.92 (0.82–0.97)
PPV (95% CI) 0.86 (0.69–0.95) 0.77 (0.63–0.87) 0.93 (0.75–0.99) 0.84 (0.68–0.93)
NPV (95% CI) 0.79 (0.68–0.86) 0.88 (0.77–0.94) 0.76 (0.66–0.84) (0.70–0.88)
AUC (95% CI) 0.78 (0.70–0.85) 0.83 (0.76–0.90) 0.76 (0.68–0.83) 0.79 (0.72–0.86)

AUC ¼ area under the receiver operating characteristic curve; CFP ¼ color fundus photography; CI ¼ confidence interval; FFA ¼ fundus
fluorescein angiography; NPV ¼ negative predictive value; OCT ¼ optical coherence tomography; PPV ¼ positive predictive value.

photography alone (imaging set A), combined fundus help improve the accuracy. This may be owing to the inferi-
photography with OCT (imaging set B), combined fundus ority of FFA in visualization of choroidal vasculature, and
photography with FFA (imaging set C), or combined therefore, FFA may not provide useful additional informa-
fundus photography with OCT and FFA (imaging set D). tion in PCV diagnosis. These results suggest little role of
Results showed that combined fundus photography with FFA to differentiate PCV from typical NVAMD or CSC
OCT (imaging set B) provided the highest sensitivity, when PCV is defined by ICGA and when fundus photog-
high specificity, high PPV, and highest AUC in PCV diag- raphy (or fundus examination) and OCT are available.
nosis, while other imaging sets provided high specificity but Subgroup analyses included a comparison of results from
low sensitivity, or vice versa. This finding highlights the graders working in an endemic area (Thailand) and outside
role of CFP and OCT, which are rapid, noninvasive imag- an endemic area (the US). Although there were no differ-
ing tools, in helping physicians make a diagnosis of PCV ences identified in accuracy for each imaging set, some dif-
with good accuracy when ICGA is not available. ferences between subgroups were observed. Results from
These results are consistent with previous studies on the US graders showed higher sensitivity, with differences
OCT in differentiating PCV from typical NVAMD. De between Thai graders across all imaging sets except set B
Salvo and associates11 reported 94.6% sensitivity and (CFP and OCT). In contrast, results from Thai graders
92.9% specificity while Liu and associates12 reported showed higher specificity, with differences between US
89.4% sensitivity and 85.3% specificity. Recently, Chang graders through all imaging sets except set B (CFP and
and associates13 reported 89.8% sensitivity and 84.5% OCT). For the US subgroup, imaging set B (CFP and
specificity when using OCT with choroidal thickness crite- OCT) achieved the highest value across all analysis (sensi-
rion. Our results showed relatively lower sensitivity and tivity, specificity, PPV, NPV, and AUC). For the Thai sub-
lower specificity than those reports, which might be owing group, imaging set B (CFP and OCT) achieved the highest
in part to different sample selections. For example, our sensitivity, NPV, and AUC, but imaging set C (CFP and
study included a broader group of serous or serosanguinous FFA) provided the highest specificity and highest PPV.
maculopathies, including CSC as well as typical NVAMD These results suggest that additional information from
and PCV, whereas previous studies included eyes with PCV FFA seems to increase the specificity of PCV diagnosis by
or typical NVAMD only. Our rationale to include eyes Thai graders but decrease the specificity by the US graders.
with CSC in this study was to make the study have greater These differences between subgroups may be explained
clinical relevance, as PCV sometimes presents as serous by an absence of grading standardization in this study.
maculopathy mimicking CSC. Also, note that this study This study was designed to determine if there were any dif-
evaluated OCT in combination with color fundus photog- ferences in PCV diagnosis when the same images were eval-
raphy, not OCT alone, as in previous reports. In real prac- uated by retina specialist graders working in different parts
tice, we believe that physicians usually evaluate the fundus of the world, specifically inside and outside PCV endemic
of their patients before, or along with, reviewing OCT im- areas, in the absence of prespecified standardization of
ages. Therefore, we used a combination of color fundus grading, to mimic what the physicians currently are using
photography (as a proxy for the fundus examination) and in their practice. Therefore, each grader evaluated the im-
OCT rather than OCT alone. ages using his or her own opinion, clinical experience, and
Regarding FFA, an invasive imaging technique, this study background knowledge. The discrepancies found in this
showed that when fundus photography and OCT are avail- study suggest that, without standardization, the use of
able (imaging set B), additional information from FFA (im- CFP, OCT, and FFA in detecting PCV may result in varied
aging set D) did not increase the accuracy of PCV diagnosis. sensitivity and specificity according to grader differences,
Also, compared to fundus photography alone (imaging set but overall, there were no statistically significant differ-
A), additional FFA information (imaging set C) did not ences in diagnostic accuracy. However, imaging set B

VOL. 192 FUNDUS PHOTO, OCT, & FA FOR POLYPOIDAL CHOROIDAL VASCULOPATHY 81
 (CFP and OCT) showed no discrepancy between the 2 sub-
groups of graders, with high sensitivity and high specificity
TABLE 3. Comparison of Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Values, and Accuracy of Each Imaging Set Between Results From Thai and US Graders

0.31 (0.18–0.44) 0.88 (0.78–0.97) <.001 0.46 (0.32–0.60) 0.83 (0.73–0.94) <.001

0.99 (0.96–1.00) 0.69 (0.58–0.80) <.001 0.97 (0.93–1.00) 0.70 (0.60–0.81) <.001
P Value

AUC ¼ area under the receiver operating characteristic curve; CFP ¼ color fundus photography; CI ¼ confidence interval; FFA ¼ fundus fluorescein angiography; NPV ¼ negative predictive value;
>.99
.003
.016
for both subgroups. This supports the role of fundus photog-
raphy (or fundus examination) and OCT as screening tools

0.92 (0.81–1.00) 0.66 (0.54–0.77)

0.68 (0.59–0.77) 0.89 (0.81–0.97) <.001 0.73 (0.64–0.82) 0.86 (0.77–0.95)
0.81 (0.73–0.88) 0.77 (0.70–0.84) >.99 0.82 (0.75–0.89) 0.76 (0.68–0.83)
to assist physicians worldwide to detect PCV when ICGA
Set D (CFPþOCTþFFA)

is not available.
US

Limitations of this study included the small number of

graders in each subgroup and the lack of grading stan-
dardization, as previously mentioned. Also, owing to
the retrospective nature of this study, standardized imag-
ing protocols for image acquisition for each study eye
TH

could not be applied. In addition, this study did not

include OCT angiography (OCTA), a relatively new
noninvasive imaging technology that allows visualiza-
P Value

.02

tions of retinal and choroidal vasculature without dye

injection, as OCTA was not yet available at our center
0.94 (0.82–1.00) 0.66 (0.54–0.77)

during the study period. However, the definitive role of

OCTA in PCV diagnosis remains under investigation.
Set C (CFPþFFA)

US

Compared with ICGA, OCTA potentially allows equal

or better visualization of branching vascular networks
but poorer visualization of polypoidal lesions16; the latter
is essential for a diagnosis of PCV.10 Therefore, we are
uncertain whether additional information from OCTA
TH

would help in diagnosing PCV diagnosis, with or without

OCT ¼ optical coherence tomography; PPV ¼ positive predictive value; TH ¼ Thai graders; US ¼ US graders.

ICGA.
The major strength of this study includes the study
P values were adjusted according to Bonferroni correction accounting for 20 multi-hypothesis testing.
P Value

>.99

>.99

>.99
>.99
>.99

design, which closely mimics clinical situations, that is,

evaluating fundus photography (as a surrogate for ophthal-
moscopy during clinical examination) along with OCT or
0.81 (0.70–0.93) 0.90 (0.81–0.98)

0.97 (0.93–1.00) 0.69 (0.58–0.80) <.001a 0.80 (0.71–0.89) 0.76 (0.66–0.86)

0.93 (0.84–1.00) 0.65 (0.53–0.76) <.001b 0.74 (0.62–0.85) 0.72 (0.60–0.83)

0.86 (0.78–0.95) 0.92 (0.84–0.99)
0.85 (0.78–0.91) 0.75 (0.68–0.83) >.99c 0.80 (0.73–0.87) 0.82 (0.75–0.88)

FFA or both (not OCT alone or FFA alone), such that

Set B (CFPþOCT)

US

these results may be practical and applicable to real prac-

tice. Furthermore, this study included retina specialist
graders working both within and outside endemic areas of
PCV to evaluate the images.
In conclusion, when ICGA is not available, color fundus
TH

photography in combination with OCT can provide high

sensitivity and high specificity to diagnose PCV. Both are
rapid and noninvasive imaging techniques that can be
P value from generalized score test by Leisenring et al.14

used, in combination, as a screening tool with high accu-

P Value

.46b
.016

racy for PCV detection by physicians working either in or

outside of an endemic area of PCV wherein ICGA may
0.56 (0.42–0.70) 0.83 (0.73–0.94)

0.77 (0.68–0.85) 0.86 (0.77–0.95)

not be feasible. This study did not suggest that fluorescein

angiography was necessary to help differentiate PCV from
Set A (CFP Only)

US

P value calculated from DeLong test.15

typical NVAMD or CSC when fundus photography and

OCT are available. Indocyanine green angiography
remains a gold standard to diagnose PCV in clinical trial
P value from McNemar test.

settings; however, its invasiveness and risks warrant

attempts to consider less invasive imaging if feasible. Future
TH

studies may aim to identify specific, clinically relevant fea-

tures on fundus photography or OCT and determine their
sensitivity and specificity in PCV diagnosis. Such informa-
AUC (95% CI)
NPV (95% CI)
PPV (95% CI)

tion, along with the development of standardized grading of

(95% CI)

(95% CI)
Specificity
Sensitivity

CFP and OCT, may allow physicians with different clinical

backgrounds to have a standardized way to detect PCV
b

c
a

with high accuracy when ICGA is not available or not

obtained.

82 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2018

FUNDING/SUPPORT: FUNDING OF THIS STUDY WAS PROVIDED BY RESEARCH COMMITTEE, FACULTY OF MEDICINE, CHIANG MAI
University, and unrestricted donations to Johns Hopkins University for retina research. Financial Disclosures: Voraporn Chaikitmongkol reported
receiving research grants from Bayer and ThromboGenics; and travel expenses from Allergan, Bayer, and Novartis. Direk Patikulsila reported working
as a consultant, receiving honoraria and travel expenses from Bayer and Novartis; and honoraria and travel expenses from Alcon. Mansing Ratanasukon
reported receiving honoraria and travel expenses from Bayer and Novartis. Nawat Watanachai reported receiving honoraria and travel expenses from
Alcon, Allergan, Bayer, and Novartis. Chaisiri Jumroendararasame reported receiving travel expenses from Allergan, Bayer, and Novartis. Janejit Choo-
vuthayakorn reported receiving honoraria and travel expenses from Alcon, Allergan, Bayer, and Novartis. Paradee Kunavisarut reported receiving hon-
oraria and travel expenses from Novartis. Neil M. Bressler reported receiving grants to his employer, the Johns Hopkins University School of Medicine,
from Bayer, Novartis, Roche (Genentech), and Samsung Bioepis. The following authors have no financial disclosures: Preeyanuch Khunsongkiet, Cath-
erine B. Mayerle, Ian C. Han, Connie J. Chen, Pawara Winaikosol, and Chutikarn Dejkriengkraikul. All authors attest that they meet the current ICMJE
criteria for authorship.

REFERENCES 10. Koh A, Lee WK, Chen LJ, et al. EVEREST STUDY Efficacy
and safety of verteporfin photodynamic therapy in combina-
1. Yannuzzi LA, Wong DW, Sforzolini BS, et al. Polypoidal tion with ranibizumab or alone versus ranibizumab monother-
choroidal vasculopathy and neovascularized age-related mac- apy in patients with symptomatic macular polypoidal
ular degeneration. Arch Ophthalmol 1999;(117):1503–1510. choroidal vasculopathy. Retina 2012;32:1453–1464.
2. Lafaut BA, Leys AM, Snyers B, et al. Polypoidal choroidal 11. De Salvo G, Vaz-Pereira S, Keane PA, et al. Sensitivity and
vasculopathy in Caucasians. Graefes Arch Clin Exp Ophthalmol specificity of spectral-domain optical coherence tomography
2000;238(9):752–759. in detecting idiopathic polypoidal choroidal vasculopathy.
3. Scassellati-Sforzolini B, Mariotti C, Bryan R, et al. Polypoidal Am J Ophthalmol 2014;158(6):1228–1238.e1.
choroidal vasculopathy in Italy. Retina 2001;21(2):121–125. 12. Liu R, Li J, Li Z, et al. Distinguishing polypoidal choroidal
4. Ladas ID, Rouvas AA, Moschos MM, et al. Polypoidal vasculopathy from typical neovascular age-related macular
choroidal vasculopathy and exudative age-related macular degeneration based on spectral domain optical coherence to-
degeneration in Greek population. Eye 2004;18(5):455–459. mography. Retina 2016;36:778–786.
5. Imamura Y, Engelbert M, Iida T, et al. Polypoidal 13. Chang YS, Kim JH, Kim JW, et al. Optical coherence
choroidal vasculopathy: a review. Surv Ophthalmol 2010; tomography-based diagnosis of polypoidal choroidal vascul-
55(6):501–515. opathy in Korean patients. Korean J Ophthalmol 2016;30(3):
6. Sho K, Takahashi K, Yamada H, et al. Polypoidal choroidal 198–205.
vasculopathy: incidence, demorgraphic features, and clinical 14. Leisenring W, Alonzo T, Pepe MS. Comparisons of predictive
characteristics. Arch Ophthalmol 2003;121(10):1503–1510. values of binary medical diagnostic tests for paired designs.
7. Yannuzzi LA, Sorenson J, Spaide RF, Lipson B. Idiopathic Biometrics 2000;56(2):345–351.
choroidal vasculopathy. Retina 1990;10:1–8. 15. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing
8. Uyama M, Wada M, Nagai Y, et al. Polypoidal choroidal the areas under two or more correlated receiver operating
vasculopathy: natural history. Am J Ophthalmol 2002;133: characteristic curves: a nonparametric approach. Biometrics
639–648. 1988;44(3):837–845.
9. Koh AH, Chen LJ, Chen SJ, et al. Polypoidal choroidal 16. Tan CS. The role of optical coherence tomography angiog-
vasculopathy evidence-based guidelines for clinical diagnosis raphy in polypoidal choroidal vasculopathy. JAMA Ophthal-
and treatment. Retina 2013;33:686–716. mol 2017;135(12):1316–1317.

VOL. 192 FUNDUS PHOTO, OCT, & FA FOR POLYPOIDAL CHOROIDAL VASCULOPATHY 83