Quantitative analysis

The corneal thickness (CT) on presentation ranged from 444-954 µm with a mean of 710.88 µm. Infiltrate thickness (IT) ranged from 112-443 µm with a mean of 288.76 µm. The overall CT in the infiltrated area on day 0 had a mean value of 710.88 µm. On days 3, 7, 14, 28 and 6 weeks mean CT had decreased to 665.12 µm, 650.24 µm, 584.35 µm, 549 µm, 507.47 µm respectively. IT on presentation had a mean value of 288.76 µm. On days 3, 7, 14, 28 and 6 weeks mean IT was 287.24 µm, 244.41 µm, 197 µm, 174.59 µm and 154.82 µm respectively. Serial AS-OCT imaging and change in CT and IT of a representative case, on treatment, are illustrated in Figure 1. Both parameters decreased as the corneal edema and inflammation subsided clinically.

CT showed a statistically significant decrease in the early (day 0-7), middle (day 7-14), and late phase (day 14-6 weeks) of treatment with daily change of 3.65% (p=0.023), 1.4% (p=0.0319) and 0.47% (p<0.0001) respectively. IT decreased at a daily rate of 2.24% (p=0.0267) in the middle phase and 0.38% (p<0.0001) in the late phase. The overall decrease in CT and IT over the course of treatment was found to be statistically significant with an average percentage reduction at 0.67% per day (p<0.0001) and 1.01% per day (p<0.0001). Graphical representation of these changes is shown in Figure 2.

Qualitative analysis

Corneal infiltration appeared hyperreflective as seen in Figure 1. Normal cornea was visualized as green colored. With increasing infiltration, the color changes from green to yellow, orange and red, with maximum hyperreflectivity denoted by red and normal reflectivity by green. Hyporeflectivity is denoted by black signals.

Besides this, unique morphological AS-OCT patterns were seen in the corneal stroma such as cystic spaces and microcystic edema. In the epithelium, a well- defined defect could be seen. Endothelial features included keratic precipitates and retrocorneal inflammatory plaque.

Stromal features are seen in Figure 3a to e. Both stromal infiltration and scarring appear hyperreflective. Stromal cystic spaces are hyporeflective cavity like areas in the stroma (Figure 2 a). A differentiating feature between infiltration (Figure 3 b) and scarring (Figure 3 e) is that infiltration has ill-defined borders compared to scarring which shows more clearly defined borders and an intact epithelium above it. Stromal edema appears as diffuse areas of decreased reflectivity (Figure 3 c). Microcystic edema was a unique feature seen on the AS-OCT in viral keratitis. It was localized to the epithelium and/or anterior stroma (Figure 3 d).

Figure 3f shows an ulcer with sloping edges. Hypereflective area of infiltrate is seen in the base of the ulcer. Some degree of re-epithelization of the defect can also be seen.

Endothelial features are shown in Figure 4, Figure 5. Figure 4 shows an inflammatory plaque attached to the corneal endothelium. Figure 5 shows the presence of a hyperreflective aggregation of inflammatory cells which are clinically seen as keratic precipitates. The latter were seen in case of disciform keratitis whereas endothelial plaque was seen in fungal keratitis.

Figure 1

A representative case showing progressive decrease in CT and IT in resolving microbial keratitis

Figure 2

Graphical change of CT(blue line) and IT(red line) over 6 weeks

Figure 3

Unique morphological changes occurring in the corneal microstructure in resolving microbial keratitis

Figure 4

Endothelial inflammatory plaque

Figure 5

Keratic precipitate seen sticking to posterior endothelial surface

Quantitative assessment of corneal parameters by AS-OCT

We performed serial scans of the cornea and measured the CT and IT at every follow up.

Konstantopoulos et al5 recorded a maximum CT of 1490 µm and IT of 840 µm. In an earlier case series4 on microbial keratitis by the same author, the maximum CT recorded was 1560 µm and maximum IT was 590 µm. In the same study, there was a single case of combined bacterial and fungal infection where the CT was not recordable and the maximum infiltrate recorded was 3640 µm. Sometimes, measurements of CT and IT in cases of fungal or combined infections are difficult due to presence of thick infiltrate which does not allow proper demarcation of endothelium.

The only quantitative study on fungal keratitis by Martone et al6 showed a decrease in central corneal thickness from 584 µm to 532 µm in a single case of Alternaria keratitis. In our study, within three days of initiation of treatment in 13 cases of fungal keratitis, CT had decreased in 8 cases and IT had decreased in 11 cases. The overall daily reduction in CT and IT of fungal keratitis over a course of 6 weeks was 0.67%(p<0.0001) and 1.09%(p<0.0001) respectively.

In the four cases of viral keratitis, CT on day 3 had decreased in two cases (740 µm to 627 µm and 450 µm to 396 µm) whereas IT showed a decrease (426 µm to 337 µm) in only one case. This reflects the more chronic course of viral keratitis. Only the daily reduction in IT over 6 weeks was statistically significant. Hixson et al8 demonstrated a decrease in CT over a period of 25 and 33 days, in two cases of disciform keratitis with resolution of corneal edema on AS-OCT as well as slit lamp.

In a similar study by Kobayashi et al.,9 they demonstrated high reflectivity in the endothelium and deep stromal layers in 10 of 13 cases which decreased throughout the course of antiviral treatment. They, however, did not perform a quantitative analysis.

Konstantopoulos et al5 demonstrated an early phase reduction in IT and CT and a late phase reduction in IT in resolving bacterial keratitis. However, they exclusively analyzed resolving bacterial keratitis.

Our rate of reduction in CT was maximum in the late phase whereas IT decreased more rapidly in the middle and late phase. The likely mechanism for reduction in CT could be due to reduction of corneal edema; whereas resolution of inflammatory cells could lead to a reduction in IT. The late phase reduction in CT can be attributed to the fact that 13 out of 17 cases were fungal in nature and antifungal drugs have longer onset of action.

Qualitative assessment of corneal parameters by AS-OCT

Qualitative assessment of the infiltration is also possible, because the intensity of hyperreflectivity corresponds to the density of the infiltration on slit-lamp examination.4

In AS-OCT, corneal infiltrates and scar both appear hyperreflective. This was in accordance with previous studies. Soliman et al,7 have distinguished infiltrate and scar in that, infiltration has ill-defined borders with overlying defective epithelium. Scarring, on the other hand, has well defined borders with clear well delineated epithelium above. With the resolution of keratitis, the infiltrate is replaced by scar tissue. It was noted that infiltration was associated with increased CT whereas scarring led to a decrease in CT. Konstantopoulos also reported that in later stages of the disease, when scar tissue develops, the affected cornea may become thinner than adjacent healthy tissue.4 A similar observation was reported by Meredith et al.9 in a study on contact lens related corneal infiltration.

AS-OCT can detect changes in corneal microarchitecture occurring during inflammation. Stromal cystic spaces were seen in cases of fungal keratitis and usually limited to anterior stroma (Figure 3 a). These are suggestive of ongoing stromal necrosis. Soliman et al7 demonstrated the presence of localized small stromal cystic spaces and full thickness large stromal cystic spaces (necrotic stroma) as unique patterns for fungal keratitis. Thinning may occur in the substance of the stroma with ongoing necrosis. This may be associated with an overlying defective epithelium as seen in Figure 3b. Stromal thinning was observed by both Soliman et al7 and Naik et al3 in cases of bacterial and fungal keratitis.

Stromal edema is seen as diffuse hyporeflective spaces in the stroma, loss of convexity of corneal endothelium, an increase in corneal thickness or in severe, cases, ruffles or folds in Descemet’s membrane (Figure 3c). Stromal edema was seen by Soliman et al8 in both fungal and bacterial keratitis and by Naik et al7 in bacterial, fungal and viral keratitis. Edema may also be localized in the form of microcystic edema, not seen clinically (Figure 3 d). This type of microcystic edema was also found by Hixson et al8 in monitoring resolution of disciform keratitis.

Stromal scarring is seen as hyperreflective areas with sharp margins and an intact epithelium on top (Figure 3e), in contrast to infiltration. Stromal scarring was also seen by Soliman et al,7 Konstantopoulos et al4 and Naik et al.3

Our findings correlate with those of slit lamp which showed resolution of congestion, corneal edema and subsequent corneal scarring. At the end of the follow up period, 12 of the 17 cases were healed. Thus, AS-OCT can be a useful adjunct to slit lamp examination in diagnosing the exact histological/pathological stage of corneal ulceration.

Figure 3 f shows a defect in the epithelium and anterior stroma, clinically recognized as an ulcer. The floor of the ulcer shows active infiltration (hyperreflective) and there is ongoing epithelization on top. Such features cannot be demonstrated by slit lamp and this highlights the adjunctive role of AS-OCT in microbial keratitis.

Figure 4 shows the presence of an endothelial plaque. Although the exact width of the plaque could not be measured due to lack of horizontal caliper in our AS-OCT, the width of this plaque was found to decrease as the infection resolved. This again highlights the role of AS-OCT in seeing retrocorneal features. Konstantopoulos2 demonstrated a similar plaque on the endothelium and measured the width of this plaque.

Figure 5 was a case of viral keratitis and shows that the infiltrate is mostly confined to the anterior part of the stroma with overlying intact epithelium. The posterior surface of the cornea shows the presence of keratic precipitates which are accumulations of inflammatory cells. In monitoring the resolution of disciform keratitis, stromal edema, microcystic edema and keratic precipitates were seen by Hixson et al8 and Konstantopoulos et al.4

In one case not responding to treatment, neither CT nor IT could be measured. This was probably because of dense infiltration and AS-OCT device with a wavelength of 1310 nm may be more useful in measuring corneal parameters in such cases. In the case series by Konstantopoulos,4 one case with mixed bacterial and fungal infection showed progressive increase in CT for almost a week and had to be finally taken up for a therapeutic keratoplasty. Hence, we concluded that dense infiltration, progressive increase in CT or thinning of the cornea with active infiltration should be taken as dangerous signs. These patients should be intensively followed for early therapeutic keratoplasty if descemetocele formation occurs.

Scans must be interpreted in the context of the clinical situation and examination. AS-OCT cannot replace slit-lamp examination, but it does provide additional information to help in assessment.

This study had several limitations. First, was the inability to see the posterior limit of the cornea in cases of thick infiltrate. This was because of the use of 830 nm wavelength light, unlike the 1310 nm that is used in AS-OCT and which allows deeper tissue penetration by reducing scattering. Secondly, the OCT machine used by us did not have a horizontal caliper tool such that the width of the infiltrate, scar or defect could not be measured. Thirdly, our SD-OCT images only a finite section of the cornea; for widespread keratitis or diffuse corneal edema, SD-OCT devices that image the cornea from limbus to limbus with better resolution would be more useful. Fourthly, scanning the exact desired location of the cornea at every follow up visit is subjective in nature. Lastly, a follow up period of 6 weeks was insufficient, as certain cases of fungal keratitis with thick stromal infiltrate can take longer to heal. We could document complete epithelial healing in only 12 out of 17 cases at the end of this period.