Protocol

A prospective, interventional, consecutive case series was performed on patients with corneal neovascularization associated with clinically active microbial keratitis evident on slit-lamp biomicroscopy. A comparison of neovascularization was undertaken before initiation of treatment and following a sustained clinical response. A clinical response was defined as either a reduction in inflammation (reduction in corneal edema and or infiltrate) or healing of the ulceration. The area of corneal neovascularization associated with the active keratitis was defined using slit-lamp biomicroscopy as the region of interest. Because of differences in the delineation of corneal neovascularization using color imaging, compared with FA or ICGA, the size of the region of interest differed between these imaging methods. Patients were excluded if they had an allergy to shellfish, iodides, or fluorescein.

Color Imaging

Acquisition of color images for each patient was performed using a standard slit-lamp–mounted digital camera system (Topcon SL-D Digital Slit Lamp; Topcon, Tokyo, Japan). A 10× magnification image was taken including the region of interest using a 45-degree angled beam of white light projected through a diffuser filter with a variable flash intensity. The same region of interest was photographed at baseline initiation of treatment and at follow-up. The region of interest that encompassed the neovascularization was identified on the selected baseline images for color imaging, FA, and ICGA. The baseline and follow-up images were overlapped, and the same sized region of interest then was used in the follow-up images centered and was aligned on the most prominent vessel, trying to keep image magnification the same. A total of 24 color images were obtained (12 at baseline and 12 at follow-up).

Indocyanine Green and Fluorescein Angiography

The HRA2 Scanning Laser Ophthalmoscope (Heidelberg Engineering, Heidelberg, Germany) was used to generate digital FA and ICGA images of the corneal neovascularization in the region of interest. A 5-mL bolus of indocyanine green (Pulsion Medical Systems, Munich, Germany) at a concentration of 5 mg/mL was injected into a peripheral vein of each patient, and region of interest images were captured with automatic real-time software every 3 to 5 seconds for approximately 3 minutes thereafter. A similar region of interest image capture method was used after a subsequent 3-mL intravenous injection of 20% sodium fluorescein (Martindale Pharmaceuticals, Romford, Essex, United Kingdom). Late-frame FA and ICGA images also were obtained at 5 and 10 minutes.

Analysis of Images

The best-quality images were selected for analysis based on our previously published subjective grading system. The baseline (initiation of treatment) images and the follow-up (clinical resolution) images were exported in TIFF format from the image capture systems, respectively (eg, color fundus images from TOPCON iBase and angiography images from Heidelberg Eye Explorer). An in-house semiautomated computer program was used to detect and assess the corneal neovascularization parameters, including the total area, vessel diameters, and tortuosity, in each image of the region of interest. The program was developed using Matlab R14 (The Mathworks, Inc, Natick, Massachusetts, USA), and the technique was described and validated elsewhere. In brief, a subimage of the region of interest was processed to detect the vessels within it based on a selective filter that highlights linear structures such as blood vessels. After vessel detection, a further geometrical analysis was performed to compute the total area, vessel diameter, and tortuosity. This has been described in detail in our previous work, which demonstrates that this method can assist users to measure and characterize corneal vessels accurately on angiography with good interobserver and intraobserver repeatability and reasonable time efficiency.

Statistical Analysis

A 2-tailed paired Student t test was used to test for differences in mean neovascular parameters (diameter, tortuosity) across the region of interest between baseline and follow-up in each patient and for differences in mean neovascular area, variance, and fluorescein leakage times at baseline and follow-up for the entire patient group. P values less than .05 were considered significant.

Patients

Twelve patients with corneal neovascularization associated with active keratitis between 35 to 74 years of age were included ( Table 1 ). The neovascularization was associated with herpes simplex keratitis (HSK) in 5 patients, with bacterial keratitis in 5 patients, and with presumed HSK after penetrating keratoplasty in 2 patients. Follow-up periods (time between baseline and follow-up imaging) ranged from 1 to 7 months (mean, 2.75 months). Clinical summaries of 2 illustrative patients are outlined below ( Figures 1-4 ).

Color imaging, fluorescein angiography (FA), and indocyanine green angiography (ICGA) image analysis from the illustrative case of herpetic keratouveitis ( Table 1 , Patient 3) at baseline presentation and at posttreatment follow-up. Shown are quantitative corneal neovascularization area changes from color photographs: (Top left) baseline, (Top center) follow-up, and (Top right) area change in region of interest in square millimeters; FA: (Center left) baseline, (Center) follow-up, and (Center right) area change in region of interest in square millimeters; and ICGA: (Bottom left) baseline, (Bottom center) follow-up, and (Bottom right) area change in region of interest in square millimeters.

Frequency distribution histogram of vessel diameters from the illustrative case of herpetic keratouveitis ( Figure 1 ). The distribution (percent) of corneal neovascular vessel diameters (in micrometers) on indocyanine green angiography across the region of interest at baseline (white bars) and at posttreatment follow-up (black bars) are shown. The histogram shows a shift toward a reduction in the percentage of larger vessels and a relative increase in smaller vessels in posttreatment follow-up.

Color imaging, fluorescein angiography (FA), and indocyanine green angiography (ICGA) image analysis from the illustrative case of bacterial keratitis with corneal neovascularization ( Table 1 , Patient 11) at baseline presentation and at posttreatment follow-up. Shown are quantitative corneal neovascularization area changes from color photographs: (Top left) baseline, (Top center) follow-up, and (Top right) area change in region of interest in square millimeters; FA: (Center left) baseline, (Center panel) follow-up, and (Center right) area change in region of interest in square millimeters; and ICGA: (Bottom left) baseline, (Bottom center) follow-up, and (Bottom right) area change in region of interest in square millimeters.

Frequency distribution histogram of vessel diameters from the case of bacterial keratitis ( Figure 3 ). The distribution (percent) of corneal neovascular vessel diameters (in micrometers) on indocyanine green angiography across the region of interest at baseline (white bars) and at posttreatment follow-up (black bars) are shown. The histogram shows a shift toward a reduction in the percentage of larger vessels and a relative increase in smaller vessels in posttreatment follow-up.

Analysis of Images

Corneal neovascular vessel diameter

The spread of vessel diameters across the region of interest changed after treatment, evident, for example, in the frequency histograms of individual vessel diameters on ICGA shown in Figure 2 (Patient 3), Figure 4 (Patient 11), and Figure 5 (all study patients). The histograms show a shift toward a reduction in the percentage of larger vessels and a relative increase in smaller vessels. There was a significant reduction in mean vessel diameter across the region of interest after treatment for each patient, more marked on FA and ICGA than on color image analysis, that is, 29.10 to 25.17 μm (ratio, 1.16; P < .01) for color, 42.74 to 32.52 μm (ratio, 1.31; P < .01) for FA, and 44.77 to 33.29 μm (ratio, 1.34; P < .01) for ICGA ( Table 3 ). There was a significant reduction in the spread (variance) of vessel diameters from baseline to follow-up across the regions of interest on ICGA (range, 0.42 to 0.21; P < .05), but not on color imaging (range, 0.22 to 0.15; P = .15) or FA (range, 0.55 to 0.32; P = .14). For 2 patients, corneal neovascularization that was not apparent in the region of interest at follow-up on color imaging was evident with ICGA. Although FA gave a better delineation of vessels than on color images, vessel diameter was affected by leakage. This was evident in Patients 4, 6, and 7, in whom there was an apparent increase in mean vessel diameter after treatment because of the presence of leakage on the FA image used for analysis.

Frequency distribution histogram for the distribution (percent) of corneal neovascular vessel diameters (in micrometers) on indocyanine green angiography across the region of interest for all 12 patients at baseline (white bars) and at posttreatment follow-up (black bars). As in Figures 3 and 4 , the pattern of reduced numbers of large diameter vessels in posttreatment follow-up continues to be seen in the analysis of all 12 patients.

TABLE 3

Corneal Neovascularization Vessel Diameter (in Micrometers) within the Region of Interest at Baseline (Initiation of Treatment) and at Follow-up (Resolution of Microbial Keratitis) Using Color Photography, Fluorescein Angiography, and Indocyanine Green Angiography for Each of the 12 Patients

Patient No.	Time Point	Color Imaging	FA	ICGA
1	Baseline	15.30	77.94	71.25
	Follow-up	12.13 a	39.00 a	35.26 a
2	Baseline	24.68	41.08	37.51
	Follow-up	29.02 a	25.28 a	31.58 a
3	Baseline	29.80	38.18	39.66
	Follow-up	25.16 a	26.48 a	35.77 a
4	Baseline	39.50	57.87	63.35
	Follow-up	32.89 a	58.75	56.85 a
5	Baseline	18.97	30.04	27.25
	Follow-up	NV	29.94	24.41
6	Baseline	25.83	26.75	55.68
	Follow-up	23.31 a	27.97 a	32.06 a
7	Baseline	N/A	34.63	42.03
	Follow-up	26.90	43.63 a	33.86 a
8	Baseline	30.73	66.11	68.43
	Follow-up	33.51 a	35.51 a	33.19 a
9	Baseline	23.59	63.50	58.76
	Follow-up	26.76 a	29.20 a	34.54 a
10	Baseline	28.51	61.17	68.43
	Follow-up	NV	31.17 a	28.28 a
11	Baseline	12.67	30.48	46.04
	Follow-up	NV	32.23	32.04 a
12	Baseline	28.34	35.62	36.14
	Follow-up	NV	NV	29.94 a
Mean	Baseline	29.10	42.74	44.77
	Follow-up	25.17 a	32.52 a	33.29 a

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Association of Toll-like Receptor 4 Gene Polymorphisms in Japanese Subjects With Primary Open-Angle, Normal-Tension, and Exfoliation Glaucoma
2. Photodynamic Therapy With or Without Intravitreal Bevacizumab for Polypoidal Choroidal Vasculopathy: Two Years of Follow-Up
3. Cataract Surgery in Eyes With Nanophthalmos and Relative Anterior Microphthalmos
4. Book review

Stay updated, free articles. Join our Telegram channel

Join