Early detection and correlation of tear fluid inflammatory factors that influence angiogenesis in premature infants with and without retinopathy of prematurity

Abstract

Purpose:

To measure the levels of inflammatory factors in tear fluid of pre-term infants with and without retinopathy of prematurity (ROP).

Methods:

The cross-sectional pilot study included 29 pre-term infants undergoing routine ROP screening. Pre-term infants were grouped as those without ROP (no ROP; n = 14) and with ROP (ROP; n = 15). Sterile Schirmer's strips were used to collect the tear fluid from pre-term infants. Inflammatory factors such as interleukin (IL)-6, IL-8, MCP1 (Monocyte Chemoattractant Protein 1; CCL2), RANTES (Regulated on Activation, Normal T Cell Expressed and Secreted; CCL5), and soluble L-selectin (sL-selectin) were measured by cytometric bead array using a flow cytometer.

Results:

Birth weight (BW) and gestation age (GA) were significantly (P < 0.05) lower in pre-term infants with ROP compared with those without ROP. Higher levels of RANTES (P < 0.05) and IL-8 (P = 0.09) were observed in the tear fluid of pre-term infants with ROP compared with those without ROP. Lower levels of tear fluid IL-6 (P = 0.14) and sL-selectin (P = 0.18) were measured in pre-term infants with ROP compared with those without ROP. IL-8 and RANTES were significantly (P < 0.05) higher in the tear fluid of pre-term infants with stage 3 ROP compared with those without ROP. Tear fluid RANTES level was observed to be inversely associated with GA and BW of pre-term infants with ROP and not in those without ROP. Furthermore, the area under the curve and odds ratio analysis demonstrated the relevance of RANTES/BW (AUC = 0.798; OR-7.2) and RANTES/MCP1 (AUC = 0.824; OR-6.8) ratios in ROP.

Conclusions:

Distinct changes were observed in the levels of tear inflammatory factors in ROP infants. The status of RANTES in ROP suggests its possible role in pathobiology and warrants further mechanistic studies to harness it in ROP screening and management.

Retinopathy of prematurity (ROP) is the leading cause of preventable infant blindness in the world. India leads with the highest number of surviving premature births annually at approximately 3.5 million.[1] Severe disease is now reported even from rural India where there are still inadequate ROP screening programs and variable levels of neonatal care.[1] The gold standard of ROP is shifting from indirect ophthalmoscopy to wide-field photo documentation; however, both require multiple sessions before a clinical conclusion is reached. In addition to clinical imagining, determining key molecular factors in ROP pathogenesis would be beneficial in the prediction of the onset and progression of the disease, as well as in the development of therapeutic strategies against specific disease-causing factors. Toward this, a range of molecular factors has been measured in a variety of sample types including serum, plasma, vitreous humor, aqueous humor, tear fluid, urine, cord blood, and amniotic fluid to elucidate their relevance in ROP pathobiology.[2] Pathological angiogenesis, the primary feature in ROP, is driven by aberrant angiogenic factors that are regulated by inflammatory factors.[3,4] In addition to the classical factors, including oxidative stress, inflammation has evolved to be one of the key determinant factors in the development of ROP.[5] Perinatal or postnatal inflammation has been associated with ROP pathogenesis and its severity.[6,7] In addition, exposure to infection with the production of inflammatory mediators may also increase the risk of ROP.[8] Hence, identifying inflammatory factors that are critical in ROP pathogenesis in preterm infants is a clinical need to further improve the management of ROP.

The association of various cytokines with ROP has been studied in vitreous, serum, and cord blood.[2,9,10,11] Cytokine levels of IGF-1 (Insulin-like growth factor 1) and VEGF (Vascular endothelial growth factor) have been found useful in predicting the risk of ROP.[12] However, invasive procedures are required to acquire these body fluids. The need for a non-invasive easy screening test to determine at-risk babies is warranted to aid in the management of ROP. Tears have long been used as a biomarker in various systemic diseases such as multiple sclerosis, Parkinson's disease, Tay Sachs disease, and even in breast cancer.[13,14,15,16] Tears have been used to study the role of various cytokines in keratoconus and dry eye disease as well and have helped in the modulation of the treatment strategy.[17,18] In diabetic retinopathy, tear levels of Tumor necrosis factor (TNF)-alpha have been used to predict the severity of the disease.[19] The tears of infants have not been extensively used to evaluate the risk of ROP. We reported the first association of proangiogenic factors in the tear fluid of preterm infants with and without ROP and observed the association between angiogenin and ROP.[20] The ratio of angiogenin level with birth weight, gestational age, and/or VEGF could serve as a potential non-invasive screening biomarker for ROP.[20] Findings from similar approaches exploring the relevance of inflammatory factors in the tear fluid of ROP infants are yet to be reported. However, a couple of recent studies have studied the status of inflammatory factors in tear fluid of ROP infants.[20,21] The current study aims to evaluate the status of tear fluid inflammatory factors in predicting ROP in infants as a non-invasive novel method by studying the association of inflammatory factors in the tear fluid of infants with and without ROP.

Methods

Study cohort

This cross-sectional pilot study was approved by Institutional Ethics Committee. The study was conducted between August 2016 to January 2017, wherein 29 Asian Indian infants who presented to our KIDROP tele-ROP screening service were included.[22,23] Subject recruitment and sample collection procedure were conducted as per Institutional Ethics Committee guidelines and in accordance with the tenets of the Declaration of Helsinki. Before recruitment and tear fluid sample collection from pre-term infants, written informed consent was obtained from either the parents or legal guardians. The pre-term infants enrolled in the study were based on the national screening guideline cut-off criteria, that is, birth weight of <2000 g and/or a gestational age of <34 weeks.[24] The first screening visit was performed within 30 days of birth according to the national guidelines. The follow-up period was defined based on the national ROP guidelines.[24] All infants underwent photo documentation during each screening visit on the RetCam Shuttle (Natus, California, USA) or the Neo Camera (Forus Health, Bengaluru, India). The staging of the disease was performed according to the International Classification of Retinopathy of Prematurity 2 (ICROP).[25] Treatment was performed on eyes that had fulfilled the ETROP (Early Treatment for Retinopathy of Prematurity) guidelines.[26] Disease staging, categorization, follow-up, and management were performed by the same ROP specialist.

The infants were categorized based on the presence of ROP during the first visit. “No ROP” group included pre-term infants who did not have ROP during the first ophthalmic examination visit after birth and did not develop ROP thereafter (n = 14). The “ROP” group included pre-term infants who presented with ROP during the first visit of the study (n = 15). As it was more relevant to determine the inflammatory factor profile variation between pre-term infants who did and did not develop ROP, the “No ROP” group qualified as an appropriate control group rather than the term infants. This approach would yield molecular information that could be used in stratifying pre-term infants with a higher risk of developing ROP.

Tear fluid collection and extraction

Tear fluid or conjunctival secretion was collected using sterile Schirmer's strips (5 × 35 mm²; Contacare Ophthalmics, and Diagnostics, India) during the first ROP screening visit. The procedure was performed before the use of any topical medication, including topical anesthetics and those required for pupillary dilatation. Briefly, the strips were placed one in each of the conjunctival fornices, simultaneously in both eyes. The strips were removed after a sufficient amount of tear fluid (wetting of strips until the 20 mm mark or more) was collected and stored in a sterile microcentrifuge tube at −80°C. Tear fluid was extracted as previously described.[27] Briefly, Schirmer's strips were cut into small pieces, agitated in 300 µl of 1x sterile phosphate-buffered solution (PBS) for 2 h at 4°C at 300 rpm, and centrifuged to elute the tear fluid. The eluted tear fluid was used for further processing to measure pro-inflammatory factors as described below.

Measurement of inflammatory factors

The levels of pro-inflammatory factors – IL-6, IL-8, MCP1-CCL2, RANTES-CCL5, and soluble L-selectin in the eluted tear fluid were measured by multiplex ELISA–Cytometric Bead Array (BD^TM CBA Human Soluble Protein Flex Set System, BD Biosciences, USA) using a flow cytometer (BD FACSCantoII, BD Biosciences, USA) as previously described.[20,27] BD FACSDiva software (BD Biosciences, USA) was used to acquire the beads and record signal intensities. FCAP array Version 3.0 (BD Biosciences, USA) was used to determine the absolute concentration of the analytes using respective standards. The absolute concentrations were then normalized to the wetting length (tear fluid volume) and dilution (extraction buffer volume) as described earlier.[20,27]

Statistical analysis

The distribution of the data was determined by the Shapiro–Wilk normality test. The differences in the level of pro-inflammatory factors between the groups were statistically analyzed by the Mann–Whitney test. Spearman's rank correlation coefficient was determined to infer the association between the pro-inflammatory factors and birth weight or gestation age of pre-term infants. Receiver operating characteristic (ROC) curve analysis to determine the area under the ROC curve (AUC), sensitivity, specificity, and odds ratio was performed to determine the relevance of the various factors studied to differentiate pre-term infants with and without ROP. The optimal cut-off value with the maximum possible sensitivity and specificity to be used for odds ratio calculation was determined using the ROC curve analysis. A P value less than 0.05 was considered statistically significant. Statistical analyses were performed with either GraphPad Prism 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) or MedCalc® Version 12.5 (MedCalc Software bvba, Belgium).

Results

The birth weight and gestation age were significantly lower in pre-term infants with ROP compared with pre-term infants without ROP in the current cohort [Supplementary Fig. 1 ^{(650.4KB, tif)} ]. No significant difference in the PMA was observed between the groups [Supplementary Fig. 1 ^{(650.4KB, tif)} ]. Marked changes in the tear fluid inflammatory factors were observed in the pre-term infants with ROP [Figs. 1 and 2]. A discernible decrease in tear fluid IL-6, though not statistically significant, was observed in pre-term infants with ROP compared with those without ROP [Fig. 1a]. The lower levels of tear fluid IL-6 were observed across the stages of ROP in pre-term infants [Fig. 2a]. Tear fluid IL-8 levels were higher in pre-term infants with ROP [Fig. 1b], particularly in pre-term infants with stage 3 ROP compared with pre-term infants without ROP or lower stages of ROP [Fig. 2b]. MCP1-CCL2 levels in the tear fluid were observed to be similar in pre-term infants with ROP compared to pre-term infants without ROP [Fig. 1c]. No significant difference was observed in the tear fluid MCP1-CCL2 levels among the different stages of the ROP [Fig. 2c]. RANTES-CCL5 was observed to be significantly increased in the tear fluid of infants with ROP compared with those without ROP [Fig. 1d]. Pre-term infants with stage 3 ROP exhibited higher levels of RANTES-CCL5 in their tear fluid compared with pre-term infants without ROP or with lower stages of ROP [Fig. 2d]. Soluble L-selectin was observed to be lower (statistically insignificant) in the pre-term infants with ROP [Fig. 1e], including different stages of ROP [Fig. 2e].

Figure 1.

Tear fluid inflammatory factor levels in pre-term infants with ROP. The graphs indicate the concentration of IL-6 (a), IL-8 (b), MCP1-CCL2 (c), RANTES-CCL5 (d), and sL-selectin (e) in the tear fluid of pre-term infants with and without ROP at the first ophthalmology screening visit. No ROP (n = 14); ROP (n = 15). Bar graphs indicate mean ± SEM; *P < 0.05, Mann–Whitney test; ROP – retinopathy of prematurity; IL – interleukin; MCP1 – Monocyte Chemoattractant Protein-1; RANTES – Regulated on Activation, Normal T cell Expressed and Secreted; sL-selectin – soluble L-selectin; SEM – Standard error mean

Figure 2.

Tear fluid inflammatory factor levels in pre-term infants with different stages of ROP. The graphs indicate the concentration of IL-6 (a), IL-8 (b), MCP1-CCL2 (c), RANTES-CCL5 (d), and sL-selectin (e) in the tear fluid of pre-term infants with and without ROP at the first ophthalmology screening visit. No-ROP (n = 14); ROP infants were subcategorized as stage 1 (n = 2), stage 2 (n = 5), and stage 3 (n = 8) as per the ICROP classification. No ROP (n = 14). Bar graphs indicate mean ± SEM; *P < 0.05, **P < 0.01, Mann–Whitney test; ROP – retinopathy of prematurity; IL – interleukin; MCP1 – Monocyte Chemoattractant Protein-1; RANTES – Regulated on Activation, Normal T cell Expressed and Secreted; sL-selectin – soluble L-selectin; SEM – Standard error mean

Unlike other tear fluid inflammatory factors measured in the current study, RANTES-CCL5 levels presented an inverse correlation with birthweight (r − 0.469, P = 0.07) and gestation age (r − 0.653, P = 0.011), only in pre-term infants with ROP [Table 1]. Furthermore, the levels of RANTES-CCL5 normalized to gestation age (RANTES: GA ratio) was significantly higher in the pre-term infants with ROP compared with pre-term infants without ROP [Fig. 3a]. Similarly, tear fluid levels of RANTES-CCL5 normalized to birth weight (RANTES: BW ratio) was significantly higher in the pre-term infants with ROP compared with pre-term infants without ROP [Fig. 3b]. In addition, the status of tear fluid RANTES-CCL5 levels with reference to another chemokine, MCP1-CCL2 (that was not significantly different among the study groups) as determined by their ratio (CCL5:CCL2 ratio) revealed that the RANTES-CCL5 was higher with reference to MCP1-CCL2 levels in pre-term infants with ROP compared with those without ROP [Fig. 3c].

Table 1.

Correlation of tear fluid inflammatory factors level with birth weight and gestation age of pre-term infants with and without ROP

Tear analytes (pg/ml)	Birthweight (grams)				Gestation (weeks)
	No-ROP		ROP		No-ROP		ROP
	r	P	r	P	r	P	r	P
IL-6	-0.051	0.863	-0.018	0.950	-0.363	0.202	0.188	0.520
IL-8	0.110	0.708	-0.343	0.210	-0.086	0.771	-0.461	0.097
MCP1	-0.126	0.669	-0.238	0.393	0.234	0.422	0.076	0.796
RANTES	0.267	0.356	-0.469	0.078	0.172	0.556	-0.653	0.011
sL-selectin	0.037	0.899	-0.011	0.972	-0.405	0.151	-0.280	0.378

r – Spearman rank correlation coefficient. P<0.05 is statistically significant. ROP – retinopathy of prematurity; IL – interleukin; MCP1 – Monocyte Chemoattractant Protein-1; RANTES – Regulated on Activation, Normal T cell Expressed and Secreted; sL-selectin – soluble L-selectin

Figure 3.

Status of tear fluid RANTES levels with reference to birth weight, gestation age, and MCP1 levels in ROP. The graphs indicate the ratio of (a) RANTES and gestation age levels (RANTES: GA), (b) RANTES and birth weight (RANTES: BW), and (c) RANTES and MCP1 (CCL5:CCL2) in pre-term infants with and without ROP. No ROP (n = 14); ROP (n = 15). Bar graphs indicate mean ± SEM. *P < 0.05, **P < 0.01, Mann–Whitney test; ROP – retinopathy of prematurity; MCP1 – Monocyte Chemoattractant Protein-1; RANTES – Regulated on Activation, Normal T cell Expressed and Secreted; GA – Gestation age; BW – Birth weight; SEM – Standard error mean

To study the relevance of RANTES-CCL5 in ROP pathobiology and risk stratification in pre-term infants, the status of RANTES-CCL5 with reference to gestation age and birth weight was analyzed using the ROC curve analysis to determine the AUC. The area under the curve (AUC = 0.786; P = 0.011) was observed for “RANTES-CCL5:gestation age ratio” [Fig. 4a]. The cut-off value of >0.3 for “RANTES-CCL5:gestation age ratio” determined using Youden's index criterion in the ROC curve analysis showed 85.7% sensitivity, 64.3% specificity, and an odds ratio of 2.7 (95% CI 0.6–12.1, P = 0.195) as shown in Fig. 4d. The area under the curve (AUC = 0.798; P = 0.0003) was observed for “RANTES-CCL5:birth weight ratio” [Fig. 4b]. The cut-off value of >0.006 for “RANTES-CCL5:birth weight ratio” showed 86.7% sensitivity, 64.3% specificity, and an odds ratio of 7.2 (95% CI 1.35–38.3, P = 0.02) as shown in Fig. 4e. The largest area under the curve (AUC = 0.824; P < 0.0001) was observed for “RANTES-CCL5:MCP1-CCL2 ratio” [Fig. 4c]. The cut-off value of >0.24 for “RANTES-CCL5:MCP1-CCL2 ratio” showed 60% sensitivity, 92.8% specificity, and an odds ratio of 6.8 (95% CI 1.12–41.8, P = 0.03) as shown in Fig. 4f. This observation suggests the relevance of tear fluid RANTES-CCL5 levels in the ROP screening and management.

Figure 4.

Relevance of tear fluid RANTES levels in ROP screening. Graphs indicate receiver operating characteristic (ROC) curve and odds ratio analyses of tear fluid RANTES levels adjusted to birth weight, gestation age, or MCP1 levels to distinguish between pre-term infants with and without ROP. The area under the receiver operating characteristic curve (AUC) for RANTES level and gestation age ratio – CCL5:GA (a), RANTES level and birth weight age ratio – CCL5:BW (b), and RANTES level and MCP1 level ratio – CCL5:CCL2 (c) are represented. Odds ratio plots based on optimal cut-off (as mentioned in y-axis) determined by the ROC analysis for CCL5:GA (d), CCL5:BW (e), and CCL5:CCL2 (f) are represented as well. No ROP (n = 14), ROP (n = 15). MCP1 – Monocyte Chemoattractant Protein-1; RANTES – Regulated on Activation, Normal T cell Expressed and Secreted

Discussion

The role of aberrant inflammatory response and altered inflammatory mediators have evolved as one of the critical determinants in ROP onset and progression.[3,6,7,8] Docosahexaenoic acid, a key precursor for specialized pro-resolution mediators (SPMs), which facilitates endogenous inflammation resolution mechanism was significantly reduced in ROP.[28] Following up on this observation, omega-3 supplementation that increases SPM levels was observed to reduce the severity of ROP.[29] Further, genetic polymorphism in inflammatory genes has also been reported be associated with an increased risk of ROP onset and progression.[21,30,31] A preclinical study has shown that the use of cyclosporin A, an anti-inflammatory agent, prevents ROP.[32] These observations studies strongly suggest the relevance of inflammatory factors in ROP pathogenesis. Angiogenesis is a balance between proangiogenic and antiangiogenic factors, and inflammatory factors aid in both normal angiogenesis and pathological angiogenesis as well. Hence, it is beneficial to identify inflammatory factors that are associated with pathological angiogenesis in preterm infants. Factors that are altered in ROP infants compared to preterm term infants with developing retinas would aid in the identification of factors that may be unique to pathological angiogenesis. Despite the overwhelming knowledge regarding the role of inflammatory factors in ROP pathogenesis, the spectrum of inflammatory factors investigated remains limited.[2] Hence, the addition of knowledge regarding the status of unexplored factors and validation of the status of studied factors in different sample types from ROP infants are essential. Further, the use of non-invasive samples such as tear fluid for molecular biomarker studies in ROP would be beneficial. However, to date, there are only two studies that have used tear fluid for biomarker discovery in ROP infants.[20,21] Of these, one study has reported altered inflammatory factors in the tear fluid of ROP infants.[21] The current study is one of the very early studies to report tear fluid inflammatory factors in preterm infants as potential biomarkers for ROP.

The factors investigated in the tear fluid of ROP infants in the current study include inflammatory factors that are known for their role in regulating pathological angiogenesis. IL-6, an inflammatory cytokine, is known to contribute toward pathological angiogenesis including retinal angiogenesis, and is increased in diabetic retinopathy.[33,34,35,36] Similarly, chemokine IL-8 (CXCL8) is reported to be important in facilitating angiogenesis and increased in retinal angiogenic conditions.[33,34,37,38] MCP1-CCL2, an inflammatory chemokine, is elevated in pathological angiogenic conditions including retinal conditions such as diabetic retinopathy with a potential contribution to the pathogenic process.[33,34,37,39,40,41] RANTES-CCL5 is a chemokine commonly associated with chronic inflammation and evidence is increasing regarding its relevance in pathological angiogenesis.[42,43,44] Elevated levels of RANTES-CCL5 have been observed in diabetic retinopathy.[45,46,47] sL-selectin, a soluble version of the cell adhesion molecule, was studied as it is known to be present in immune cells and in the context of its migration in developed blood vessels.

Tear fluid RANTES-CCL5 and IL-8 levels were observed to be higher in ROP infants, whereas, IL-6 and sL-selectin levels were lower in ROP infants with MCP1-CCL2 levels unaltered in ROP infants in the current study [Figs. 1 and 2]. RANTES-CCL5 was shown to be relevant in the clinical discrimination of pre-term infants with and without ROP [Fig. 4]. A study reported a higher level of RANTES-CCL5 in the vitreous humor of ROP infants,[9] which is supportive of the current study's findings. However, other studies have reported no significant differences in the level of the vitreous humor of RANTES-CCL5 between controls and ROP infants,[21] decreased levels in serum,[48] and blood[49] of ROP infants. An increased IL-8 level in the tear fluid of ROP infants in the current is supported by reports that have also shown an increase in the IL-8 levels in the vitreous humor,[21] aqueous humor,[50] tear fluid,[21] serum,[51] and blood[52] of ROP infants. A study reported that the levels of IL-8 in the urine were not significantly different between control pre-term infants and ROP infants.[51] Studies have also reported conflicting observations with reference to the levels of IL-8 in amniotic fluid, with one study demonstrating a significant increase[53] and the other with no significant difference[54] between controls and ROP infants. Various reports have also shown differences in the observation with reference to IL-6 and MCP1-CCL2 levels in ROP infants. Tear fluid IL-6 levels that were lower albeit insignificant in the current study were in line with other studies that reported no significant difference in its level in the vitreous humor,[21] tears,[21] serum,[51] urine,[51] and amniotic fluid[54] of ROP infants compared to controls. However, there were other studies that have reported IL-6 levels to be significantly higher in the vitreous humor,[9] aqueous humor,[50] serum,[55] blood,[49] cord blood plasma,[56] maternal plasma,[57] and amniotic fluid[53] of ROP infants. Tear fluid MCP1-CCL2 levels, which were similar between the ROP infants and controls in the current study, were similar to other studies that measured the levels of MCP1-CCL2 in aqueous humor[50] and tear fluid.[21] However, a study reported an increase in MCP1-CCL2 levels in cord blood serum samples of ROP infants.[58] Tear fluid sL-selectin levels, observed to lower in the current study, have not been investigated and reported earlier in the context of ROP. However, plasma sE-selectin levels have been studied and were observed to be significantly higher in ROP infants.[59,60] Observations from the current study suggest the importance of studying the relevance of RANTES-CCL5 and IL-8 in ROP pathogenesis.

The contrasting results of certain inflammatory factors in the maturing retina can be due to normal angiogenesis in maturing retina present at various other anatomic sites in the early life of a neonate, the type of controls (pre-term infants versus full-term infants), and stage or severity of ROP in the respective study cohort. We found a similar correlation in our previous study, wherein angiogenic factors were also elevated in the maturing retina. To establish a temporal causal relationship, a larger sample size is needed along with pre-term infant controls and different stages of ROP. As there have been no studies in ROP with tears, the relationship between these factors could not be determined with the actual intraocular milieu. Other confounding risk factors such as sepsis and potentially pro-inflammatory systemic conditions were not excluded during the analysis. The tear levels may not represent the actual intraocular levels but the association is important to determine babies at risk of developing or worsening ROP. The estimation of angiogenic factors in our previous report[20] and inflammatory markers in this report can be assessed at the first visit to determine babies at risk of developing or progressing ROP. This can aid us in determining the frequency of visits of these infant patients and early intervention to inhibit the progression of the disease process. This novel study demonstrates the feasibility of using tear fluid of preterm infants undergoing routine ROP screening using non-invasive and easy-to-employ methods to obtain vital information about their tear analytes, which can be used for risk categorization, prognostication, and follow-up. This has provided us with a new tool in the management of these tiny and precious babies.

Financial support and sponsorship

This work was funded by Narayana Nethralaya Foundation, Bengaluru, India. The funders had no role in study design, data collection, and analysis.

Conflicts of interest

There are no conflicts of interest.

Supplementary Figure 1

Birth weight, gestation age and post menstrual age of pre-term infants in the study cohort. The graphs indicate birth weight – BW in grams (a), gestation age – GA in weeks (b), and post menstrual age – PMA in weeks (c) of the study cohort at first visit. No ROP (n = 14); ROP (n = 15). Bar graphs indicate Mean ± SEM. **P < 0.01, ***P < 0.001, Mann–Whitney test; ROP – retinopathy of prematurity; SEM – standard deviation