Background/aims To investigate the utility of using montaged optical coherence tomography (OCT) thickness maps to monitor perivascular thickness as a marker of vasculitic activity in patients with large-vessel retinal vasculitis.
Methods This is a retrospective cohort study of 22 eyes of 11 patients with a history of retinal vasculitis associated with birdshot chorioretinopathy (BCR). Patients had serial spectral domain 6×6 mm cube OCT scans centred on the fovea, optic nerve and proximal branches of the superior and inferior retinal vessels. OCT thickness change maps for each respective region were analysed. Changes in perivascular thickness were confirmed by assessing vasculitic activity on fluorescein angiography (FA), when clinically indicated.
Results In three patients, montaged OCT scans were acquired at diagnosis and serially through initial treatment. In all three patients, montaged OCT demonstrated reduced perivascular thickening with oral prednisone treatment, which was confirmed by FA showing reduced vascular leakage in both eyes. Eight patients had serial montaged OCT scans after diagnosis and initial treatment of BCR. Four of these patients showed fluctuations in perivascular thickness during flares and treatment that were confirmed by either increased or decreased vascular leakage on FA. The other four patients remained quiet on their immunosuppressive treatment regimens, and no changes in perivascular thickness were detected.
Conclusions Evaluating large-vessel perivascular thickness on OCT scans may be a useful method for non-invasively monitoring posterior pole large-vessel retinal vasculitis.
- birdshot chorioretinopathy
- fluorescein angiography
- optical coherence tomography
- retinal vasculitis
Statistics from Altmetric.com
Optical coherence tomography (OCT) has revolutionised the diagnosis and management of posterior segment disease, especially conditions involving the macula.1 Recent advances in OCT technology include the introduction of OCT angiography (OCTA), high-speed swept-source scanning technology and wide-field imaging beyond the macula.2 OCTA has already produced a better understanding of retinal vascular disease by allowing distinct imaging of the superficial and deep retinal vascular plexuses.3 In addition, OCTA has been used to demonstrate increased microvascular changes and reduced capillary density in patients with uveitis.4 5 However, one drawback of OCTA is the inability to detect leakage from retinal blood vessels, which is crucial to evaluating inflammatory eye disease.
Retinal vasculitis may be associated with known uveitic conditions, such as birdshot chorioretinopathy (BCR) and Behçet’s disease, associated with other systemic diseases, such as systemic lupus erythematosus, or it may be idiopathic in aetiology.6 Retinal vasculitis is traditionally diagnosed and monitored with fluorescein angiography (FA).7 While generally safe, FA is an invasive procedure with a small but real risk of life-threatening anaphylactic reaction.8 Patients with a known allergy to fluorescein cannot be followed using FA. Furthermore, fluorescein has been found in breast milk after intravenous administration and should be used with caution in nursing mothers to avoid the potential for a phototoxic reaction in breastfed neonates.9 Therefore, a non-invasive means of monitoring retinal vascular leakage is highly desirable.
Previous studies have identified increases in macular as well as retinal nerve fibre layer thickness on OCT in association with active uveitis, and oedema of the macula and optic nerve is routinely monitored with OCT.10–13 We hypothesised that retinal thickness in localised areas adjacent to large retinal vessels would increase at times of active large-vessel retinal vasculitis and decrease with quiescence. To investigate this hypothesis, we compared spectral domain (SD)-OCT scans covering the large posterior pole vessels in patients with retinal vasculitis associated with BCR at times of disease quiescence and activity. Vasculitis activity was confirmed with FA. Here we demonstrate that standard SD-OCT scans can be used to monitor retinal vasculitis affecting the posterior pole.
Materials and methods
Eleven patients with a history of retinal vasculitis associated with BCR who were seen under an institutional review board-approved protocol in the Uveitis and Ocular Immunology Clinic at the National Eye Institute (NEI) were included in this retrospective cohort study. Patient characteristics, including age, sex and clinical examination, were obtained from the NEI electronic health record. The study was conducted in accordance with the Health Insurance Portability and Accountability Act (HIPAA) regulations and adhered to the tenets of the Declaration of Helsinki. All patients provided written informed consent.
Both eyes of each patient had serial 6×6 mm 512×128 A scans acquired on a Zeiss Cirrus SD-OCT machine (Cirrus HD-OCT; Carl Zeiss Meditec, Dublin, California, USA). Overlapping cube scans centred on the fovea, optic nerve and proximal branches of the superior and inferior retinal vessels were acquired (figure 1). The arcade scans were standardised for each patient by placing the corner of the scanned cube in the centre of the optic disc. The time required to take the four images per eye was around 30 s per image with a compliant patient. The technicians who acquired the scans reported no major difficulties in acquisition. Thickness change maps were produced in the Zeiss Cirrus Viewer based on automated segmentation of the total retina thickness (internal limiting membrane to retinal pigment epithelium). Cool colours (green and blue) represent decreased thickness, while warm colours (yellow, orange and red) represent increased thickness. If automatic registration failed, manual registration was used. Montaged images of the different retinal areas from the same eye were produced in Adobe Photoshop CS6 (San Jose, California, USA). The term ‘montaged OCT’ will be used to refer to these images for the remainder of this report. Standard FA was performed when clinically indicated based on increased inflammatory activity (ie, increased vitreous cellular activity, new intraretinal haemorrhages) or to assess response to a change in medication regimen. Late frames of the angiogram were acquired out to 10 min. Vasculitic activity on FA was used to corroborate changes in perivascular thickness identified on montaged OCT.
For the quantitative analysis, retinal thickness over corresponding superior and inferior arcade vessels was measured at 1500 µm and 3000 µm from the optic disc. A line of best fit was drawn from the boundary of the optic disc to 1500 µm and 3000 µm over each arcade vessel evaluated (typically two vessels per quadrant). To facilitate the detection of the vessel course, the fundus view was used. Every effort was made to replicate this line between corresponding scans of the same patient. The ETDRS grid was snapped to the distal boundary of the line and the retinal thickness obtained from the central subfield value. Retinal thickness was compared in the newly diagnosed patients pretreatment and post-treatment, and statistical significance was assessed using a paired t-test.
Twenty-two eyes of 11 patients with a history of retinal vasculitis associated with BCR were followed with serial OCT cube scans centred on the fovea, optic nerve and proximal branches of the superior and inferior retinal vessels. The mean age at first scan was 55.5 years (range 38–64 years). Nine patients were female, and 100% of patients were human leukocyte antigen (HLA)-A29+. The mean time between diagnosis and first OCT scan for this study was 2.4 years (range 0–12 years). The mean follow-up time with images for this study was 1.4 years (range 0.2–2.2 years). The mean number of montaged OCT scans per patient was 11 (range 4–17).
Three patients had serial montaged OCT scans beginning within 1 month of diagnosis and were followed through their initial treatment. All three patients had large retinal vessel leakage on FA in both eyes at the time of diagnosis. With oral prednisone treatment, all three patients showed substantial reduction in perivascular retinal thickness on montaged OCT associated with reduced large-vessel leakage on FA in both eyes (figure 2). As seen on the OCT B-scans, the change in retinal thickness was predominantly due to reduction in inner retinal thickness adjacent to retinal vessels.
For the quantitative assessment of perivascular retinal thickness, a total of 47 measurements (eight measurements per eye in 5/6 eyes; seven measurements in 1/6 eyes) were taken both pretreatment and post-treatment. The mean retinal thickness of the six eyes from three patients at diagnosis was 397.4 µm (range 323–485 µm, SD 35.1 µm). The mean thickness following prednisone treatment was 346.9 µm (range 301–400 µm, SD 26.0 µm), with a mean decrease of 50.6 µm (SD 25.8, SE of the mean 3.8 µm, p<0.00001).
The other eight patients had serial montaged OCT scans after diagnosis and initial control of their disease (range 0.25–12 years). In four of these patients, montaged OCT detected fluctuations in retinal perivascular thickness, which were corroborated as either increased or decreased vascular leakage on FA in at least one eye. For example, a patient with BCR treated with mycophenolate mofetil 1 g twice daily by mouth and tapering doses of oral prednisone experienced increased floaters, vitreous cellular reaction and perivascular thickness on montaged OCT in his right eye when the prednisone dose was reduced from 10 to 7.5 mg daily (figure 3). Seven months later, following another course of oral prednisone 30 mg per day with taper and increasing the dose of mycophenolate mofetil to 1.5 g twice daily, his perivascular thickness returned to baseline on montaged OCT and retinal vascular leakage resolved on FA.
For the quantitative assessment of perivascular retinal thickness in patients who were initially inactive, then flared and subsequently returned to quiescence with treatment, a total of 46 measurements (eight measurements per eye in 4/6 eyes and seven measurements per eye in 2/6 eyes) were taken preflare, during and postflare. The mean retinal thickness of the six eyes from three patients at baseline was 323.4 µm (range 271–393 µm, SD 28.3 µm). The mean thickness during activity was 331.6 µm (range 271–398 µm, SD 29.1 µm). The mean thickness during quiescence after treatment was 315.8 µm (range 260–365 µm, SD 23.5 µm). There was a mean increase of 8.2 µm (SE of the mean 2.8 µm) during flare from baseline and a mean decrease of 15.8 um (SE of the mean 2.9 µm) after treatment of the flare.
In another example, a patient with BCR, who had been stably quiet for over 3 years on oral ciclosporin A 100 mg twice daily, developed increased serum creatinine levels, and the ciclosporin dose was reduced to 50 mg twice daily. Six weeks after reducing the dose of ciclosporin, increased perivascular thickness was noted on montaged OCT (figure 4). FA at the same visit confirmed large retinal vessel staining and leakage. The right eye of this patient also had an area vitreoretinal adhesion in the superior arcade confounding retinal thickness measurements in this area.
For three of the four patients who flared, the date of the FA showing activity corresponded to the date of the first montaged OCT showing increased perivascular thickness. In the other patient, mild leakage on FA was first noted approximately 2 months prior to increased perivascular thickness on montage OCT. When increased perivascular thickness was noted on montaged OCT, the FA showed increased leakage.
The remaining four patients remained quiet on their immunosuppressive regimens. No substantial changes in perivascular thickness were detected with serial montaged OCT, and no further vasculitic activity was observed with FA (figure 5). For the quantitative assessment of perivascular retinal thickness in the patients who remained stable, a total of 64 measurements (eight measurements per eye in 8/8 eyes) were taken both at baseline and after a period quiescence (mean 1.75 years). The mean retinal thickness of the eight eyes from four patients at baseline was 268.1 µm (range 211–338 µm, SD 27.3 µm). The mean thickness at the last visit was 250.7 µm (range 198–318 µm, SD 24.8 µm), with a mean decrease of 17.8 µm (SE of the mean 3.0, p<0.00001).
OCT provides non-invasive three-dimensional reconstructions of the retina, and many of the retinal anatomic structures can be deciphered in these reconstructions.1 In the current report, we demonstrate that perivascular retinal thickness as measured on montage OCT increases during active retinal vasculitis and decreases with quiescence, and can serve as a surrogate to retinal large-vessel leakage on FA. Similarly, prior studies have shown that peripapillary retinal nerve fibre thickness may increase in active uveitis and decline with control of inflammation.11 13
Montage OCT is not an absolute substitute for FA, especially for initial diagnosis of retinal vascular leakage. However, once retinal vasculitis has been diagnosed, this method may be used along with FA to monitor disease activity by following perivascular retinal thickness assessed by OCT scans encompassing the posterior pole vasculature. For example, if perivascular retinal thickness increases on OCT, then an FA can be obtained to confirm vascular leakage and evaluate the extent of vasculitis. All of the patients in our study had retinal vasculitis associated with BCR; therefore, care should be taken when translating this method to other forms of retinal vasculitis. Furthermore, patients may have small-vessel leakage leading to generalised increased retinal thickness or macular oedema, which is better evaluated by traditional analysis of OCT B-scans and global retinal thickness measurements.
The described method of using OCT to monitor retinal vasculitis has several limitations compared with FA. While wide-field FA is routinely available and important for monitoring peripheral leakage in patients with uveitis,14 imaging outside the posterior pole is not feasible with current commercially available SD-OCT machines limiting OCT analysis to the major posterior pole vessels. However, this will likely change in the near future as wide-field OCT becomes more available. Furthermore, other potential complications of retinal vasculitis, such as retinal non-perfusion, cannot be directly assessed with traditional OCT. However, long-standing non-perfusion will eventually cause thinning of retinal thickness on OCT, and macular ischaemia can now be assessed by OCTA.15 In addition, three eyes of two patients had vitreoretinal adhesions over retinal blood vessels. In one of these eyes, the tractional retinal thickening precluded reliable OCT perivascular thickness analysis in the superior arcade. Adhesion of the vitreous and other vitreoretinal interface abnormalities, such as epiretinal membrane, and their effect on retinal thickness must be considered when using this methodology.
While no major fluctuations in montaged OCT perivascular thickness were detected in patients who remained stable during the study period, overall decrease in thickness was detected after an average of 1.75 years of follow-up. This suggests a possible progressive loss of neuroretinal tissue over time in patients with BCR even in those without overt intraocular inflammation, and stresses the importance of additional functional testing, such as electroretinography (ERG) and visual field testing, during surveillance of these patients.16 Indeed, one of these patients had progressive loss of function on ERG and visual field testing.
Evaluating perivascular retinal thickness with OCT allows the ability to non-invasively monitor patients with large-vessel retinal vasculitis. With the advent of wide-angle OCT imaging, OCT monitoring of perivascular thickening may be performed outside of the proximal large retinal vessels without the need to montage separately acquired images. This method may detect increased vascular leakage, and therefore inflammatory activity, which may have otherwise gone unnoticed (eg, in the absence of increased vitreous cell or haze, or presence of vascular sheathing), prompting further evaluation with FA to confirm the OCT findings. While thickness measurements can be obtained on selected areas along the vasculature, use of objective thickness measurements in clinical practice is not practical and is unlikely to be used. We find that the qualitative changes on colour topography along the vasculature are helpful and are more likely to be used in clinical practice. The absolute sensitivity of this method compared with FA for detecting large-vessel leakage requires further study. However, this may be a useful tool for monitoring large-vessel vasculitis when FA is not otherwise being obtained.
We thank Robert B Nussenblatt, MD, MPH, for intellectual discussions regarding this study.
JEK and WT contributed equally.
Contributors JEK, WT and HNS designed the study. JEK and WT collected the data. JEK, WT, SK, MA and HNS analysed the data. JEK drafted the manuscript, and all authors critically reviewed and approved the manuscript.
Funding This work was supported by the National Eye Institute Intramural Research Program, National Institutes of Health.
Competing interests None declared.
Ethics approval National Eye Institute, National Institutes of Health.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.