Purpose
To describe optical coherence tomography angiography (OCTA) findings in multifocal choroiditis (MFC) with active lesions and to characterize the concordance between the OCTA and other traditional imaging modalities.
Design
Reliability and validity analysis.
Methods
Patients with suspected choroidal neovascularization (CNV) or acute inflammatory lesions associated with MFC were assessed in this study. All participants underwent preliminary traditional multimodal imaging including color fundus photography, fundus autofluorescence, near-infrared reflectance imaging, spectral-domain optical coherence tomography, and fluorescence angiography (FA). The participants were prospectively recruited to perform OCTA. OCTA findings of active lesions were compared with other traditional imaging results. Vascular flow signal representing CNV was identified, and a quantitative analysis of CNV size was performed on OCTA.
Results
Fifty-two eyes of 26 MFC patients (14 were bilaterally affected and 12 were unilaterally affected) were included. Among the 23 active CNV cases, 20 were confirmed on OCTA while the other 3 were invalid owing to severe motion artifacts. OCTA of CNV showed a well-circumscribed vascular network (mean flow area, 0.271 mm 2 [± 0.144 mm 2 ]). Among the 34 inflammatory lesions in 13 eyes, 32 showed no blood flow in the outer retina on OCTA while the other 2 showed blood flow signal.
Conclusions
OCTA has a predominant advantage in differentiating CNV from inflammatory lesions. It also allows the visualization of detailed vascular structure of CNV and the function of flow area measurement realizes the quantitative analysis of CNV. Hence, it could be an alternative option for CNV identification and may better guide treatment.
Multifocal choroiditis (MFC) is a chronic, recurrent, often bilateral, ocular inflammatory disorder of unknown etiopathogenesis that is more common in young or middle-aged myopic women. Patients with MFC may complain of blurred vision, floaters, scotomas, and photopsia. The inflammation results in the formation of multiple grayish or yellow-white, punched-out chorioretinal lesions that may progress to secondary inflammatory choroidal neovascularization (CNV), discrete pigmented fibrotic scars, atrophy, or curvilinear chorioretinal streak. The frequency of CNV is nearly 33.3% in patients with MFC, and the risk of developing CNV is also high, even in patients without clinical evidence of acute inflammation. The presence of CNV may be responsible for severe central visual loss, and hence prompt therapy is very important. In patients with only active inflammation, corticosteroid and immunomodulatory therapy is usually needed. However, in patients with active CNV, an intravitreal injection of an anti–vascular endothelial growth factor (VEGF) agent is required as well. Recognition of pathologic changes is required in treatment decision making. Early detection and monitoring of a new or recurrent active CNV is crucial for preventing progressive and irreversible vision loss in patients with MFC.
Nowadays, multimodal imaging plays an increasingly important role in diagnosing MFC nowadays. However, the distinction between active CNV and other lesions such as inflammation remains unclear in many cases. Color and red-free photography, near-infrared reflectance (NIR) imaging, and fundus autofluorescence (FAF) imaging are of little value in distinguishing active inflammation from CNV. Indocyanine green angiography (ICGA) may fail to identify CNV if the vascular change is insignificant, especially at the beginning of the disease. Spectral-domain optical coherence tomography (SDOCT) has become an important noninvasive method for structural imaging of patients with suspected CNV. However, both the inflammation and CNV may appear as hyperreflective signals on OCT, making it difficult to differentiate one from another. The gold standard for identifying CNV currently is fluorescence angiography (FA), but the possible adverse reactions to the dye, including nausea, allergy, and, rarely, anaphylaxis, limit the use of this invasive method. Therefore, an imaging modality that would help in unequivocal diagnosis of CNV in this population would be invaluable.
Recently a new imaging technology, the optical coherence tomography angiography (OCTA) using an algorithm called “split-spectrum amplitude-decorrelation angiography” (SSADA), has allowed for new insights into the visualization of normal and pathologic vascularization. It detects motion in the blood vessel lumen by measuring the variation in reflected OCT signal amplitude between consecutive cross-sectional scans. The use of the SSADA algorithm improves the signal-to-noise ratio of flow detection and provides a better visualization of the blood flow in selected retinal layers.
The present study evaluated a cohort of patients with MFC using both OCTA and other traditional imaging technologies to describe the concordance in detecting CNV in MFC. Also, the sensitivity and specificity of OCTA were evaluated using FA as the ground truth. This study did not distinguish between punctate inner choroidopathy (PIC) and MFC, which were considered as different entities in the past. According to the multimodal imaging study by Spade and associates, MFC and PIC might be the same disease, since they target the same structures, in the same phenotypic manner, and are treated in the same way.
Methods
This reliability and validity analysis study was approved by the Institutional Review Board and Ethics Committee of Shanghai First People’s Hospital, which allowed recruitment of patients and performing of multimodal imaging examination, including OCTA scans. Informed consents were obtained from all the patients. The research adhered to the tenets of the Declaration of Helsinki.
This study investigated 52 eyes of 26 MFC patients with active lesions seen by 2 physicians (S.Y. and Y.G.) in the ophthalmology department of Shanghai First People’s Hospital from August 1, 2014 to September 30, 2015. The diagnostic criteria of MFC were based on the original descriptions of both MFC and PIC adapted from previous reports by Spaide and associates. Active lesions were thought to be present in the following cases: (1) new symptoms with lesions in the posterior pole of the eyes without any previous diagnosis; (2) new or reactivated old lesions in an eye with a known diagnosis; (3) subretinal fluid (SRF)/intraretinal fluid (IRF) ; or (4) leakage on FA.
Before recruitment, the participants underwent comprehensive examination to identify any underlying systemic inflammatory or infectious etiology. Ophthalmic examination included best-corrected visual acuity (BCVA), anterior segment examination, dilated fundus biomicroscopy, and preliminary multimodal imaging such as color photography (Zeiss Visucam 200 digital fundus camera; Carl Zeiss Meditec AG, Jena, Germany), FAF, NIR, SDOCT, and FA (Spectralis HRA+OCT; Heidelberg Engineering, Heidelberg, Germany). CNV was defined based on (1) the history and clinical signs, (2) the typical FA features, and (3) OCT appearance. Active CNV lesions were differentiated from inactive CNV lesions by progressive pooling of dye leakage on FA, or subretinal/intraretinal hemorrhage and/or fluid accumulation on OCT. Inflammatory lesions were identified by the appearance of multiple yellow-white chorioretinal lesions in the absence of characteristics of CNV on OCT and FA. The lesion identification was conducted by 2 trained readers (S.Y. and G.Y.) independently. The ones that were not agreed upon were sent to a third reader.
After recruitment, all participants underwent OCTA using the instrument from Optovue RTVue XR Avanti (Optovue, Inc, Fremont, California, USA). This instrument captured 2 consecutive B-scans at each fixed position, and SSADA was used to analyze the OCTA information and remove artifacts. The OCTA angiograms were separately projected into en face views in 4 layers (superficial retina, deep retina, outer retina, and choroid capillary) automatically. The outer retina layer was defined from beneath the outer plexiform layer (OPL) to the Bruch membrane (BM). In healthy individuals, the outer retina is normally avascular. Therefore, after accurate segmentation and artifact removal, any flow in this layer can be interpreted as abnormal vessels such as CNV.
The 6 mm × 6 mm OCTA images were used primarily to evaluate each MFC eye to include as many lesions as possible. If any lesion was detected on the 6 mm × 6 mm OCT angiogram, a more detailed evaluation was performed using the 3 mm × 3 mm OCTA. The OCTA images were assessed for the presence of CNV, CNV size, location, structural information, and morphologic appearance through independent evaluation by another 2 trained readers (L.C. and X.C.); again, the ones that were not agreed upon were sent to a third reader. Both of the OCTA readers were masked to the disease-related information and other imaging results. To obtain information about the CNV size, the quantitative analysis tool of the updated AngioVue v2015.100.0.3 system was used to calculate the flow area of each CNV lesion, and the greatest linear dimension was also measured for comparison. In flow area measurement, the outline of CNV was drawn manually and then the flow area was calculated automatically by the system. The appearance of CNV on the OCTA image was classified as well circumscribed or poorly circumscribed. While assessing an OCTA image, manual adjustment was done if the boundaries were misaligned. Besides, the Angio Overlay function of the AngioVue v2015.100.0.3 system, in which the B-scans were overlaid with the blood flow signal, was used to assist in CNV detection on the cross-sectional view.
Concurrently, to estimate the sensitivity and specificity of OCTA in detecting active CNV, the FA images of the same cohort, as the ground truth, were reviewed independently by another 2 trained readers (S.W. and L.M.), who had no idea about the diagnosis and other imaging findings. The sensitivity, specificity, and positive and negative predictive values were calculated for OCTA compared with FA. The statistical analysis was performed using SAS Software Version 9.2 (SAS Institute Inc, Cary, North Carolina, USA).
Results
This study included 52 eyes of 26 MFC patients (22 female and 4 male) with a mean age of 46.3 ± 12.3 years; 12 patients were unilaterally and 14 were bilaterally affected. The median BCVA for the affected eyes at presentation was 20/63 (range, FC [fingers counting]/30 cm to 20/20), while the median BCVA for the unaffected eye was 20/25 (range, 20/32 to 20/20). Of the 40 affected eyes, 20 had CNV lesions, 6 had acute inflammatory lesions, 7 had both CNV and inflammatory lesions, and the remaining 7 eyes had only atrophy lesions. Among the 27 eyes with CNV, 22 had active CNV, 4 had inactive CNV, and 1 had both. The patients’ demographic characteristics and information about lesions are shown in Table 1 .
| Case No. | Sex | Age (y) | Eye | BCVA | Affected Eye (Y/N) | Lesion Type a , b | FA (Y/N) c | OCTA (Y/N) d |
|---|---|---|---|---|---|---|---|---|
| 1 | M | 35 | OD | 20/667 | Y | aCNV/inaCNV | Y/N | Y/Y |
| OS | 20/32 | Y | aCNV | Y | Y | |||
| 2 | F | 47 | OD | 20/25 | Y | INF | N | N |
| OS | 20/100 | Y | aCNV/INF | Y/N | Y/N | |||
| 3 | F | 54 | OD | 20/63 | Y | Atrophy | N | N |
| OS | FC/30 cm | Y | aCNV | Y | Y | |||
| 4 | F | 55 | OD | 20/200 | Y | aCNV | Y | Y |
| OS | 20/50 | Y | Atrophy | N | N | |||
| 5 | F | 29 | OD | 20/20 | Y | Atrophy | N | N |
| OS | FC/30 cm | Y | aCNV/INF ×3 | Y/N | Y/N | |||
| 6 | F | 34 | OD | 20/20 | Y | aCNV/INF ×3 | Y/N | Y/N |
| OS | 20/20 | N | ∖ | N | N | |||
| 7 | F | 33 | OD | 20/20 | N | ∖ | N | N |
| OS | 20/80 | Y | INF×2 | N | Y | |||
| 8 | F | 28 | OD | 20/25 | Y | aCNV/INF ×4 | Y/N | Y/N |
| OS | 20/20 | Y | INF | N | N | |||
| 9 | F | 57 | OD | 20/32 | N | ∖ | N | N |
| OS | 20/667 | Y | aCNV/INF | Y/Y | Y/N | |||
| 10 | F | 64 | OD | 20/667 | Y | Atrophy | N | N |
| OS | 20/50 | Y | aCNV | Y | Y | |||
| 11 | F | 64 | OD | 20/400 | Y | inaCNV | N | N |
| OS | 20/32 | Y | aCNV | Y | Y | |||
| 12 | F | 35 | OD | 20/20 | Y | inaCNV | N | Y |
| OS | 20/200 | Y | aCNV | Y | Y | |||
| 13 | F | 36 | OD | 20/80 | Y | aCNV | Y | N |
| OS | 20/200 | Y | aCNV | Y | Y | |||
| 14 | F | 45 | OD | 20/25 | Y | INF ×4 | N | N |
| OS | 20/20 | N | ∖ | N | N | |||
| 15 | F | 45 | OD | 20/125 | Y | inaCNV/INF ×4 | N/N | Y/N |
| OS | 20/200 | Y | aCNV | Y | N | |||
| 16 | F | 70 | OD | 20/250 | Y | Atrophy | N | N |
| OS | 20/80 | Y | aCNV | Y | Y | |||
| 17 | F | 61 | OD | 20/25 | N | ∖ | N | N |
| OS | 20/200 | Y | aCNV | Y | Y | |||
| 18 | F | 58 | OD | 20/20 | N | ∖ | N | N |
| OS | 20/25 | Y | aCNV | Y | N | |||
| 19 | M | 60 | OD | 20/25 | N | ∖ | N | N |
| OS | 20/20 | Y | Atrophy | N | N | |||
| 20 | F | 59 | OD | 20/63 | Y | Atrophy | N | N |
| OS | 20/40 | Y | aCNV | Y | Y | |||
| 21 | F | 41 | OD | 20/32 | Y | INF ×2 | N | N |
| OS | 20/32 | N | ∖ | N | N | |||
| 22 | M | 40 | OD | 20/63 | Y | aCNV | Y | Y |
| OS | 20/25 | N | ∖ | N | N | |||
| 23 | F | 27 | OD | 20/32 | N | ∖ | N | N |
| OS | 20/63 | Y | aCNV | Y | Y | |||
| 24 | F | 29 | OD | 20/32 | N | ∖ | N | N |
| OS | 20/125 | Y | aCNV/INF ×7 | Y/N | Y/N | |||
| 25 | M | 32 | OD | 20/25 | Y | inaCNV | N | N |
| OS | 20/25 | Y | INF | N | N | |||
| 26 | F | 66 | OD | 20/25 | N | ∖ | N | N |
| OS | 20/200 | Y | aCNV | Y | Y |
a Lesion types are defined as follows: aCNV (active CNV lesions); inaCNV (inactive CNV lesions); INF (inflammatory lesions); Atrophy (atrophy lesions). In the affected eyes in which there are other kinds of lesions, atrophy lesions are not shown.
b There are usually more than 1 active inflammatory lesions in 1 eye; numbers of lesions are noted by “x” (34 active inflammatory lesions in total).
c Y means that there were active CNV lesions according to the FA findings; N means there were no active CNV lesions according to FA.
d Y means that there were blood flow signals representing CNV on OCTA; N means no CNV was detected on OCTA.
Of the 23 active CNV cases, all were demonstrated on FA and 20 (87.0%) were identified on OCTA, while the other 3 (13.0%) could not be identified owing to poor image quality. The sensitivity of OCTA in detecting active CNV abnormalities using FA as the ground truth was 87.0%, and the specificity was 89.7% ( Table 2 ). A total of 34 acute inflammatory lesions from 13 eyes were observed, of which 32 (94.1%) showed no blood flow in the outer retina on OCTA and the other 2 (5.9%) showed blood flow signal (Case 2). Among the 5 inactive CNV lesions, 3 (60.0%) showed blood flow signal on OCTA (Case 5).
| OCTA | FA | Diagnostic Accuracy | |
|---|---|---|---|
| Yes | No | ||
| Yes | 20 | 3 | Positive predictive value 87.0% |
| No | 3 a | 26 | Negative predictive value 89.7% |
| Diagnostic Accuracy | Sensitivity 87.0% | Specificity 89.7% | |
a No choroidal neovascularization was identified on OCTA in these 3 cases owing to poor image quality.
Characteristics of Active Choroidal Neovascularization Lesions on Optical Coherence Tomography Angiography, Fluorescence Angiography, Spectral-Domain Optical Coherence Tomography, and Fundus Autofluorescence
Characteristics of active CNV lesions on OCTA, FA, SDOCT, and FAF are shown in Tables 3–5 .
| Item | Characteristic | Number of Cases | Proportion |
|---|---|---|---|
| OCTA | |||
| En face view | |||
| GLD | <1000 μm | 12 | 52.2% |
| 1000∼2000 μm | 8 | 34.8% | |
| >2000 μm | 0 | 0.0% | |
| No blood flow signal | 3 | 13.0% | |
| Total | 23 | 100.0% | |
| Flow area | <0.3 | 12 | 52.2% |
| 0.3∼0.6 | 8 | 34.8% | |
| >0.6 | 0 | 0.0% | |
| No blood flow signal | 3 | 13.0% | |
| Total | 23 | 100.0% | |
| CNV shape | Well circumscribed | 19 | 82.6% |
| Poorly circumscribed | 1 | 4.3% | |
| No blood flow signal | 3 | 13.0% | |
| Total | 23 | 100.0% | |
| Angio overlay | With blood flow signal | 20 | 87.0% |
| No blood flow signal | 3 | 13.0% | |
| Total | 23 | 100.0% | |
| FA | |||
| Active CNV appearance | Early hyperfluorescence with late leakage | 23 | 100.0% |
| No leakage | 0 | 0% | |
| Total | 23 | 100.0% |
| Item | Characteristic | Number of Cases | Proportion |
|---|---|---|---|
| CNV Type | Type 1 | 3 | 13.0% |
| Type 2 | 13 | 56.5% | |
| Mixed | 7 | 30.4% | |
| Total | 23 | 100.0% | |
| Fluid | Intraretinal fluid | 3 | 13.0% |
| Subretinal fluid | 3 | 13.0% | |
| Both | 5 | 21.7% | |
| No fluid | 12 | 52.2% | |
| Total | 23 | 100.0% | |
| Hemorrhage | Yes | 1 | 4.3% |
| No | 22 | 95.7% | |
| Total | 23 | 100.0% | |
| Ellipsoid zone | Intact | 0 | 0.0% |
| Discontinued | 6 | 26.1% | |
| Disappeared | 17 | 73.9% | |
| Total | 23 | 100.0% | |
| Choroid under CNV | Increased reflectivity | 5 | 21.7% |
| Decreased reflectivity | 18 | 78.3% | |
| No change | 0 | 0.0% | |
| Total | 23 | 100.0% |
| Characteristic | Number of Cases | Proportion |
|---|---|---|
| Hyper-AF | 2 | 8.7% |
| Hypo-AF | 4 | 17.4% |
| Iso-AF | 1 | 4.3% |
| Mixed | ||
| Hyper-AF ringed by hypo-AF | 4 | 17.4% |
| Hyper-AF with central hypo-AF | 1 | 4.3% |
| Hyper-AF with central and surrounding hypo-AF | 3 | 13.0% |
| Hypo-AF ringed by hyper-AF | 7 | 30.4% |
| Normal -AF ringed by hypo-AF | 1 | 4.3% |
| Total | 23 | 100.0% |
Except for the 3 negative cases, the CNV lesions revealed on OCTA shared similar characteristics of well-circumscribed vascular networks in different shapes. The mean greatest linear dimension was 950 μm (± 368 μm; range, 350–1786 μm). A total of 52.2% of the lesions were smaller than 1000 μm and all were within 2000 μm. The CNV areas were also evaluated by the tool of flow area measurement, and the mean area was 0.271 mm 2 (± 0.144 mm 2 ; range, 0.053–0.536 mm 2 ). A total of 52.2% were smaller than 0.3 mm 2 and all were within 0.6 mm 2 . Angio Overlay function could also detect the blood flow signal of CNV on the cross-sectional section and distinguish it from the projection artifact ( Figure 1 ).
In the present study, all of the 23 active secondary CNV lesions had the classic appearance on FA with early hyperfluorescence and late leakage. The lesions presented as hyperreflective heterogeneous material on SDOCT, mostly (56.5%) beneath the neurosensory retina, some (13.0%) beneath the retinal pigment epithelium (RPE), or both (30.4%). SRF or IRF was detected in 11 (47.8%) of the cases, and only 1 case (4.3%) had hemorrhage. The ellipsoid zone was discontinued or disappeared. The FAF appearance was diverse, from decrease to increase of autofluorescence.
Characteristics of Acute Inflammatory Lesions on Optical Coherence Tomography Angiography, Fluorescence Angiography, Spectral-Domain Optical Coherence Tomography, and Fundus Autofluorescence
Characteristics of acute inflammatory lesions on OCTA, FA, SDOCT, and FAF are shown in Tables 6–8 .
| Item | Characteristic | Number of Cases | Proportion |
|---|---|---|---|
| OCTA | |||
| En face view of outer retina layer | With blood flow signal | 2 | 5.9% |
| No blood flow signal | 32 | 94.1% | |
| Total | 34 | 100.0% | |
| Angio Overlay | With blood flow signal | 2 | 5.9% |
| No blood flow signal | 32 | 94.1% | |
| Total | 34 | 100.0% | |
| FA | |||
| Early frame | Hyperfluorescence | 8 | 23.5% |
| Hypofluorescence | 14 | 41.2% | |
| isofluorescence | 12 | 35.3% | |
| Total | 34 | 100.0% | |
| Late frame | With leakage | 34 | 100.0% |
| No leakage | 0 | 0.0% | |
| Total | 34 | 100.0% |
| Item | Characteristic | Number of Cases | Proportion |
|---|---|---|---|
| Location | Outer retina layer | 7 | 20.6% |
| Sub-RPE | 12 | 35.3% | |
| Sub-RPE with outer retinal infiltration | 15 | 44.1% | |
| Total | 34 | 100.0% | |
| Fluid | Intraretinal fluid | 0 | 0.0% |
| Subretinal fluid | 3 | 8.8% | |
| Both | 0 | 0.0% | |
| No fluid | 31 | 91.2% | |
| Total | 34 | 100.0% | |
| Hemorrhage | Yes | 0 | 0.0% |
| No | 34 | 100.0% | |
| Total | 34 | 100.0% | |
| Ellipsoid zone | Intact | 7 | 20.6% |
| Discontinued | 13 | 38.2% | |
| Disappeared | 14 | 41.2% | |
| Total | 34 | 100.0% | |
| Choroid under inflammatory lesion | Increased reflectivity | 27 | 79.4% |
| Decreased reflectivity | 5 | 14.7% | |
| No change | 2 | 5.9% | |
| Total | 34 | 100.0% |
| Characteristic | Number of Cases | Proportion |
|---|---|---|
| Hyper-AF | 8 | 23.5% |
| Hypo-AF | 19 | 55.9% |
| Iso-AF | 0 | 0.0% |
| Mixed | ||
| Hyper-AF with central hypo-AF | 7 | 20.6% |
| Total | 34 | 100.0% |
When analyzing the multimodal imaging characteristics of inflammation, all multiple lesions detected in 1 eye, concurrently or successively, were included.
No blood flow signal in the outer retinal layer was detected for acute inflammatory lesions on OCTA, except for 2 lesions (5.9%) that will be described later in Case 2. The Angio Overlay function demonstrated similar results. On FA, these lesions appeared differently, from hypofluorescence (41.2%) to isofluorescence (35.3%) to hyperfluorescence (23.5%), in the early phase, and all showed leakage in the late phase. One inflammatory lesion was misidentified as CNV by the FA readers ( Figure 2 ). On SDOCT, these lesions presented as sub-RPE or outer retinal hyperreflective homogeneous material. Fluid accumulation was only seen in 3 cases (8.8%), but SRF was a possibility. Damage of the ellipsoid zone band was found in 27 (79.4%) of the cases. Similar to CNV, inflammatory lesions also lack typical characteristics on FAF.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
