Patient characteristics
Control group
Fifty aqueous humors from 50 individuals operated for cataract surgery were analyzed to evaluate correlation of corticoids ocular levels with age, sex and time of surgery. There were 25 men and 25 women. The mean age was 73 ± 14.7 years [29–104 years]. Among this cohort, 27 patients (14 women and 13 men had simultaneous serum and aqueous humor sampling. Their mean age was 72 ± 13 years [29–92]. To evaluate whether levels of corticoids in the aqueous humor correlate with levels in the vitreous, nine patients (5 men and 4 women) who underwent combined cataract and membrane peeling surgery were included. Their mean age was 62.2 ± 12.7 years [45–82].
CSCR group
Fourteen patients (12 men and 2 woman) with long-lasting central serous chorioretinopathy (CSCR) (> 10 years, mean 23 ± 14 years) underwent cataract surgery in one eye, and one patient underwent cataract surgery in both eyes at 2 months interval. The CSCR patients were long-lasting complex cases with alteration of the retinal pigment epithelium associated with signs of pachychoroid and recurrent episodes of subretinal fluid during the follow-up period. Seven CSCR patients had a history of GC intake. Three patients had been treated with GCs for asthma. One patient received systemic GC in a context of allergy. Two patients received intra articular injection of GC (1 patients for discopathy and 1 patient for shoulder arthritis). One patient had an history of nasal inhalation of GC. Three patients had a history of argon-laser photocoagulation on leaky spots several years ago and had been patients been treated by verteporfin photodynamic therapy at least 1 year before the surgery. Another patient had been treated with verteporfin photodynamic therapy several years before the cataract surgery. Type 1 quiescent choroidal neovascularization was suspected in two eyes based on flat irregular pigment epithelium detachments associated with flow on optical coherence tomography angiography but without late phase hyper fluorescence on indocyanine green angiography. No sign of CNV activity was present. The mean subfoveal thickness was 475 ± 100 µm [377–693 µm]. None of the patient had subretinal fluid at the time of cataract surgery. These patients can thus be classified as complex CSCR36, not active at the time of AH sampling.
Serum steroidome
All the analyzed corticoids were measured in the serum (Table 2) and the results are summarized in Fig. 2.
Corticoids in the serum. (A) Correlation between cortisol (F) and cortisone (E) levels (ng/ml) in the serum. (B) Correlation between corticosterone (B) and 11-dehydrocorticosterone (A) levels (ng/ml) in the serum. (C) Correlation between cortisol (F) level (ng/ml) and the time of the day (hours). (D) Correlation between cortisone (E) level (ng/ml) and the time of the day (hours).
As expected, the systemic steroidome shows variation of cortisol levels with the time of sampling and a positive correlation between the active and inactive forms of the gluco (F,E) and mineralocorticoids (B, A) which represents the activity of the 11β-hydroxysteroid dehydrogenase type 1 and 2 enzymes.
There was a positive statistically significant correlation between serum cortisone (E) and serum cortisol (F) levels (Spearman r = 0.70, p < 0.0001) and between serum 11-dehydrocorticosterone (A) and corticosterone (B) (Spearman r = 0.86, p < 0.0001) (Fig. 2A,B). There was a correlation between the time of sampling and cortisol level in the serum (r = − 0.52, p = 0.0043), with higher levels measured in the morning (Fig. 2C). Serum cortisone, the inactive form of cortisol, did not correlate significantly with the time of sampling (r = − 0.33, p = 0.08) (Fig. 2D). In the serum, mean aldosterone level was 0.075 ± 0.065 ng/ml [0.015–0.264] and it was not correlated with the time of the day (Spearman r = 0.38, p = 0.1). The systemic steroidome is shown in Table 2.
Aqueous humor steroidome
In the aqueous humor, aldosterone, progesterone (P4), 11-deoxycorticosterone (DOC), 11-deoxycortisol (S), testosterone (T), DHEA dehydoepiandrosterone (DHEA) and Sulfate of dehydroepiandrosterone (SDHEA) were below limit of quantification in all samples.
The composition of the ocular steroidome is therefore limited compared to that of serum, demonstrating selective mechanisms in the passage of corticoids from the circulation to the aqueous humor as shown at the blood–brain barrier37, 38.
Table 3 shows the ocular steroidome measured in the aqueous humor. The 17 hydroxyprogesterone (17-OHP) was measurable in AH with a mean level of 0.048 ± 0.016 ng/ml [0.035–0.083]. Delta4-androstenedione (∆4A) was above detectable level in 70% of the samples with a mean level of 0.031 ± 0.026 ng/ml.
In the aqueous humor, the mean E level was 0.987 ± 0.438 ng/ml [0.01–1.71] and the mean F level was 5.05 ± 1.34 ng/ml [1.15–7.80] (n = 50) and intraocular cortisol (F) and cortisone (E) levels correlated positively (Spearman r = 0.55, p < 0.0001). The F/E ratio which reflects the activity of the 11β-HSD enzymes demonstrates a prevalence of the active cortisol form that activates the GR pathway. The mean corticosterone (B) level was 0.18 ± 0.09 ng/ml [0.05–0.38] and the 11-dehydrocorticosterone (A) level was 0.10 ± 0.07 ng/ml [0.008–0.294] but there was no correlation between A and B levels in the AH (Spearman r = 0.1, p = 0.4) (Fig. 3A,B) indicating that MR pathway can be activated by corticosterone in the eye like in the brain39.
Metabolism of corticoids in the aqueous humor (AH). (A) Correlation between cortisol (F) and cortisone (E) levels (ng/ml), (A) correlation between corticosterone (B) and 11-dehydrocorticosterone (A) levels (ng/ml), (C) correlation between the time of the day and the level of cortisol (F) in the AH, (D) significant and positive correlation between cortisone (E) level (ng/ml) and the time of the day (r = 0.3, p = 0.03), (E) significant influence of the time of the day on the E/F in the AH, (F) significant influence of the time of the day on the F/E ratio in the AH.
Influence of the circadian rhythm on ocular corticoid levels
Reflecting the circadian clock, systemic cortisol levels fluctuate between a peak in the early hours of the morning to a minimal at midnight, as observed in the declined serum levels of individuals sampled in the evening as compared to the morning. In the aqueous humor, there was no significant correlation between the time of the day and the level of cortisol (F) [Fig. 3C] (r = 0.04, p = 0.9), but there was a positive correlation between time of the day and the cortisone levels (Spearman r = 0.313, p = 0.03) [Fig. 3D], with surprisingly higher cortisone levels in the evening as compared to the morning. Consequently, E/F ratio significantly increased with the time of the day (Spearman r = 0.38, p = 0.009) and F/E decreased with the time of the day (r = − 0.54, p = 0.04) (Fig. 3E,F) showing that the ratio of active cortisol is lower in the evening not because of decreased cortisol levels like in the serum but due to higher either activity of the 11β-HSD2 enzyme that converts cortisol into inactive cortisone or to higher cortisone entry at later times in the day. There was no influence of the time of sampling and the aqueous levels of A, B and B/A (p > 0.005) (not shown).
Influence of age on corticoids ocular levels
There was no correlation between the age of the patients at the time of surgery and the level of cortisol (F) (Spearman r = 0.03, p = 0.833) or cortisone (E) in the aqueous humor (Spearman r = − 0.006, p = 0.966). There was no correlation between the age of the patients and the E/F ratio in the aqueous humor (Spearman r = − 0.07, p = 0.64) [not shown]. No correlation was found between A, B and A/B and the age of the patient as well (p > 0.05).
Influence of sex on corticoids ocular levels
There was no significant difference between cortisol levels measured in the aqueous humor of women as compared to men (p = 0.39) (Fig. 4A), no difference was measured between cortisone levels (p = 0.89) (Fig. 4B) and E/F ratio and sex (p = 0.9) (Fig. 4C). No significant difference was found in A, B and A/B levels in the aqueous humor of male and females (not shown). Thus, sex did not influence the levels of gluco and mineralocorticoid hormones in the aqueous humor.
Influence of sex on the metabolism of glucocorticoids in the AH. No significant difference between male and female in the levels of cortisol (ng/ml) (A), cortisone (ng/ml) (B) and E/F ratio (C).
The 17-OHP (17-hydroxyprogesterone) levels were not significantly different between men and women (p = 0.28) but ∆4A, levels were significantly lower in women as compared to men (0.01 ± 0.04 vs. .019 ± 0.015 ng/ml, p = 0.0075) since it is an adrenal androgen. Pogesterone and testosterone were below detectable levels in all samples and, ∆4A was the only androgen hormone measured in the aqueous humor.
Correlation between corticoids levels in aqueous humor and vitreous (V)
In a specific group of patients, we collected AH and undiluted V during the same surgery to evaluate whether levels of corticoids could differ between the two compartments. There was no significant difference between the level of cortisol (F) in the AH or in the V (4.4 vs. 4.8 ng/ml, p = 0.5) and the levels of F in the AH and V of a same patient were correlated (Spearman r = 0.77, p = 0.01) (Fig. 5A,B). There was also no significant difference between the level of cortisone (E) in the AH and in the V (Fig. 5A) (1.2 vs. 1.9 ng/ml), but there was no significant correlation between E level in the AH and V of the same patient (p = 0.25). The levels of corticosterone (B) and of 11-dehydrocorticosterone (A) were not significantly different in the AH and V of the same patient (Fig. 5C). A/B and E/F ratio were not significantly different in the AH and V. Levels of corticoids did not differ significantly in the AV and V from the same patient suggesting that sampling the AH is representative of V levels, at least in eyes without significant ocular pathology.
Comparison between gluco and mineralocorticoids in the AH and vitreous (V) of the same individual.
Correlation between serum and ocular corticoids levels
To evaluate the link between serum and aqueous humor steroidome, we sampled simultaneously the aqueous humor and the blood of patients undergoing uncomplicated cataract surgery. There was no correlation between serum cortisol and aqueous humor cortisol levels (Spearman r = 0.138, p = 0.49) (Fig. 6A). No more than 10% of serum cortisol was measured in the aqueous humor, indicating that active mechanisms control the entry of cortisol from the serum into the eye, preventing the cortisol level from following the increase in circulating cortisol. Thus, spikes in circulating cortisol do not result in an increase in ocular cortisol.
Comparison of corticoids metabolism in the AH and in the serum. (A) No correlation between cortisol (F) level (ng/ml) in the serum and in the AH (r = 0138, p = 0.49), (B) significant and positive correlation between cortisone (E) level (ng/ml) in the serum and in the AH (r = 0.57, p = 0.0016). (C) No correlation between the E/F ratio in the serum and in the AH. (D) No correlation between the F/E ratio in the serum and in the AH. (E) No correlation between corticosterone (B) level (ng/ml) in the serum and in the AH (r = 0.42 p = 0.06) (F). Significant and positive correlation between 11-dehydrocorticosterone (A) level (ng/ml) in the serum and in the AH (r = 0.52, p = 0.003). G. No correlation between the A/B ratio in the serum and in the AH.
On the other hand, there was a positive and significant correlation between serum and aqueous humor cortisone (E) levels (Fig. 6B) (Spearman r = 0.577, p = 0.0016), although levels of cortisone in the aqueous humor also remained low, around 5% of serum cortisone levels. This could indicate that the inactive form of cortisol (F), cortisone (E), could be transferred more easily from the serum into the aqueous humor, although being also a ligand of efflux transporters. The E/F ratio witnesses the activity of 11β-HSD enzymes and the metabolism of cortisol. There was no correlation between corticoid metabolism in the serum and in the AH (Fig. 6C,D).
Levels of corticosterone (B) in the AH and in the serum did not correlate (Spearman r = 0.42, p = 0.06), but the levels of dehydrocorticosterone (A), the inactive form correlated positively in the AH and serum (r = 0.527, p = 0.003) although less than 10% of serum A and B levels were measured in the AH, indicating that A and B entry into the eye is also regulated by efflux proteins (Fig. 6E,F). Like the E/F ratio, the A/B ratio in the serum and the AH did not correlate, indicating the activity of the 11-bHSD1 and HSD2 enzymes in the AH is independent from that in the serum (Fig. 6G).
Corticoid levels in eyes with CSCR
As cortisol (F), corticosterone (B) and 11 deoxycorticosterone (A) levels in the aqueous humor correlate neither with the age or the sex of the individual, nor with the time of the day, we have compared F, B and A levels in the AH of 15 eyes with complex CSCR, not presenting subretinal fluid at the time of sampling, to those of the corticoids in the controls eyes (Fig. 7).
Comparison ocular steroidome in patients with CSCR and controls. (A) Levels in ng/ml of cortisol (F) (****p = 0.0001) and cortisone (E) (ns, p > 0.05) in control and CSCR eyes. (B) F/E ratio in control and CSCR eyes (**, p = 0.023). (C) Levels in ng/ml of corticosterone (B) (*p = 0.03) and 11-hydroxycorticosterone (A) in control and CSCR eyes. (D) B/A ratio in control and CSCR eyes (**, p = 0.0044).
Since cortisone and the F/E ratio varied with time of the day, we compared first E and F/E with all control samples but also compared the 15 AH from CSCR eyes, that were all collected between 12 and 3PM with control samples, collected at the same time of the day (n = 25 time-matched controls). Cortisol level (F) was lower in CSCR eyes as compared to controls eyes (2.9 ± 1.2 vs. 5.0 ± 1.3 ng/ml, p = 0.0001) (Fig. 7A). Corticosterone (B) was significantly lower in CSCR than in controls eyes (0.11 ± 0.07 vs. 0.18 ± 0.09 ng/ml, p = 0.03). There was no significant difference in the E and A levels between CSCR and controls (0.99 ± 0.44 vs. 0.98 ± 0.46 ng/ml and 0.07 ± 0.06 vs. 0.10 ± 0.07 ng/ml, respectively, p = 0.5 for both). The B/A ratio was 3.36 ± 5.1 [0.6–10] in CSCR eyes and significantly higher than in control eyes (0.99 ± 1.56 [0.014–7], p = 0.018) (Fig. 7B). The F/E ratio was significantly lower in the AH of CSCR eyes (3.5 ± 2.3) as compared to all controls (6.1 ± 3.2) (p = 0.023) and it remained significantly lower as compared to time-matched controls (5.1 ± 2.3) (p = 0.01) (Fig. 7). We did not measure significant difference in the level of 17-OHP and no difference in the level of ∆4A when comparing only men CSCR (n = 12) with control males (n = 25).
Finally, since the mean age of CSCR patients was significantly lower than the one of controls and although we did not find any influence of age on the steroidome of control individuals, we then performed a sub analysis comparing the ocular steroidome of CSCR patients with the one of age-matched controls (p = 0.7, n = 15). In this subanalysis, cortisol (F) level was significantly lower in the CSCR group as compared to the age-matched controls (p = 0.0001), the cortisone (E) levels was not different (p = 0.8), the F/E ratio was significantly decreased (p = 0.04) showing similar results using this smaller group of controls (n = 15) and a trend towards a lower glucocorticoid pathway activation in eyes with CSCR.
Correlation between intraocular corticoids and subfoveal choroidal thickness in CSCR patients
In the group of patients with CSCR, subfoveal choroidal (SFCT) thickness was measured in the regular follow up protocol using EDI SD-OCT less than 8 days before the AH was collected. There was no correlation between subfoveal choroidal thickness and ocular F level (Spearman r = − 0.20, p = 0.46), E level (r = 0.11, p = 0.67), F/E (r = − 0.166, p = 0.55), A level (r = − 0.26, p = 0.37) and B level (r = 0.53, p = 0.06).
Case 1
Caucasian women, 32-year-old with a history of asthma since childhood, having being treated for years with inhaled and oral corticosteroids until she was 20 years old. She then periodically took inhaled corticoids. She did not present any other risk factors. In 2018, after psychological familial stress, she was diagnosed with bilateral CSCR and treated with half-fluence verteporfin photodynamic therapy (PDT) in both eyes in 2018 with transient efficacy as she reported recurrence less than 2 months after treatment. She was referred in November 2020 with bilateral serous macular detachment, similar to images she provided from 2018 (Fig. 8C).
Multimodal imaging of a 32-year-old woman with bilateral chronic CSCR. (A) Intermediate phase at 4 min shows mild staining in the inferior foveal zone and moderate stippled hyperfluorescence at the level of the superior arcade on fluorescein angiography of the right eye, corresponding to a hyperfluorescent plaque on indocyanine green angiography (ICGA). The intermediate cliché at 7 min of the left eye, (B) shows a mild FA staining surrounding the foveal SRF pocket, corresponding to granular stippled hyperfluorescence on ICGA (C–G). spectral-domain optical coherence tomography (SD-OCT) of the left and right eye during follow-up, respectively (C). The exam recorded in 2018 already showed a foveal neurosensory dome-shaped detachment with photoreceptor elongation and thickened choroid (D–F). Progressive SRF resolution after starting systemic mineralocorticoid antagonist therapy. The effect of the treatment on choroidal thickness cannot be evaluated without the enhanced depth imaging focus (EDI) (G). Additional topical dexamethasone treatment increased SRF reduction. Note the vascular enlargement with choriocapillary attenuation and signal hypertransmission underneath the residual neurosensory detachment and slight RPE elevation. No macular neovascularization was observed at that timepoint in both eyes (OCTA not shown).
At presentation, best corrected visual acuity was 8/10 on both eyes and the patient complained of micropsia and loss of contrast. On spectral domain optical coherence tomography (SD-OCT) B scan, subfoveal choroidal thickness was 680 µm in the RE and 516 µm in the LE with massive dilation of choroidal veins and effacement of the choriocapillaris. Macular subretinal fluid was present in both eyes (Fig. 8D). FA and ICG angiography performed in January 2021 showed a faint leaky site nasal to the fovea in the right eye and no leaky site on the left eye with mid-phase hyperfluorescent plaques more pronounced in the left than in the right eye along temporal vessels, at 7 min (Fig. 8A,B). No sign of choroidal neovascularization was found on ICG-A or on optical coherence tomography angiography (OCTA). Eplerenone 50 mg was introduced in January 2021 resulting in reduction of SRF at 1 and 3 months but without complete resolution of the SRF (Figs. 8E,F). Preservative free dexamethasone drops (twice per day) were introduced in July 2021 resulting in almost complete resolution of SRF at 15 days in both eyes (Fig. 8G). The patient was followed with complete resolution of SRF in both eyes and she was maintained on dexamethasone 1 drop/day without any raise in intraocular pressure for 2 months (Fig. 9A). The vision was 10/10 when she was switched to Softacort® 1 drop/day and remained free from recurrence until June 2022 (Fig. 9B,C), where an initial minimal increase in SRF appeared on OCT upon reduction of drop to 1 every 3 days (Fig. 9D). Recurrence of SRF was observed in the left eye 3 weeks after she had stopped all treatments (Fig. 9E).
SD-OCT follow-up after glucocorticoid switch. (A–C) After hydrocortisone switch and slow tapering, a complete absence of CRF was documented, together with a good photoreceptor (PR) layer reconstitution and mild RPE irregularities. (D) Upon furtherly tapering hydrocortisone to 1 drop every 3 days, a minimal PR detachment was observed (white arrow) in the left eye. Glucocorticoid treatment cessation and close follow-up at 3 weeks (E) revealed an initial recurrence of the disease in the left eye corresponding to the same perifoveal area (white arrow).
Case 2
58-years old Asian man, who is referred for unresolved and stable subfoveal fluid in the last 4 years, despite one PDT, 3 years ago. Visual acuity is 1/10. In the left eye, ICG-angiography showed no hyperfluorescent plaques or hyperpermeability signs and fluorescein angiography (FA) showed two juxta foveal and nasal leakage spots, not accessible to laser treatment (Fig. 10A–C). On SD-OCT, the dilated choroidal veins are overhung by a bulge of the pigmentary epithelium, but the choriocapillaris remains visible (Fig. 10D). Subfoveal choroidal thickness was 390 µm in the left eye and 350 in the right eye. In the left eye, foveal subretinal fluid was associated with thinning of the outer nuclear layer and shortening of the photoreceptor segments attesting of chronicity of the disease. At 3 weeks after a combination of eplerenone 50 mg and dexamethasone preservative free drop (3 times per day), there was a reduction of the subfoveal choroidal thickness from 390 to 350 µm. However, there was still SRF [Fig. 10E]. The patient discontinued all treatment by himself and came back 1 month later with increased SRF and increased subfoveal choroidal thickness to 380 µm (Fig. 10F). Six weeks after restoration of the initial treatment, SRF almost resolved and subfoveal choroidal thickness decreased to 290 µm, demonstrating the on–off effect of the treatment both on central macula thickness (CMT) and on choroidal thickness and veinous dilation (Fig. 10G). Vision remained low at 1/10.
Multimodal imaging of the left eye (LE) of a 58-year-old male. (A) Fundus autofluorescence (FAF) showing perifoveal and inferior foveal hyperfluorescence. Intermediate FA (B) and ICGA (C) clichés at 5 min, showing stippled foveal hyperlfuorescence and granular irregular hyperfluorescence, respectively. (D–G) SD-OCT B-scans and infrared fundus imaging during the follow-up period.