Abstract
Aims:
Explore potential of 31 tear biomarkers involved in screening for diabetic peripheral neuropathy (DPN). Assess the utility of aesthesiometry for measuring corneal damage in DPN and determine optimal cutoff point for detecting DPN.
Methods:
Screening test pilot study recruited 90 participants from a tertiary hospital in Lima, Peru. Participants were grouped by diabetes and neuropathy status. Tears collected on Schirmer strips, and proteins measured by both ELISA and multiplex-bead assay. Corneal sensitivity was measured by aesthesiometry, and DPN through vibration perception threshold testing.
Results:
There were 89 participants included in the analysis. The mean age was 55.7±1.46, and 58.4% were female. MMP-9 and TGF-alpha concentrations were higher in participants with DPN versus diabetes alone, though not significant. Aesthesiometry was decreased in individuals with DPN when compared to participants with diabetes alone (p<0.01) and normal controls (p<0.01). Optimal cutoff point for aesthesiometry was found to be 5.8mm, with 79% sensitivity and 75% specificity.
Conclusions:
Tears are an insufficient standalone tool for detecting DPN based on the biomarkers analyzed. Aesthesiometry is a simple, inexpensive, and accurate method to assess corneal damage associated with moderate-severe DPN, and its integration into screening practices has potential to improve detection of DPN in poor-resource settings.
Keywords: diabetes, type 2 diabetes, diabetic neuropathy, tear biomarkers, aesthesiometry, corneal neuropathy
1. INTRODUCTION
Diabetic peripheral neuropathy (DPN) is a debilitating, progressive complication of type 2 diabetes. In low and middle-income countries, limited budget and poor access to resources are frequent barriers to the implementation of high technology devices to alleviate disease burden (1).
In Peru, a study in a national public hospital found the prevalence of DPN to be 57% (2). The high cost of management leads to amputations in approximately 6% of individuals with DPN in poor-resource settings due to medical noncompliance or lack of finances (2–4). Considering 1/3 of Peruvians live in poverty, low-cost, point-of-care tests are needed (2). Having an effective means of early detection of DPN is crucial for early intervention, which would have a major impact in alleviating its social, economic, and medical burden.
Nerve conduction studies and skin/nerve biopsies are gold standards for detecting DPN (5). In vivo corneal confocal microscopy (IVCM), which assesses the morphology of corneal nerve fibres, also provides accurate information regarding early stages of DPN (6, 7). However, these tests require expensive equipment and well-trained staff.
Tears are an underutilized reservoir of biomarkers that show great potential for medical research. Qin et al. found 514 tear proteins, of which 365 were present in other tissues and organs throughout the body (8). Previous studies have correlated tear biomarkers involved in corneal health, such as substance P, to IVCM findings in individuals with diabetes (9). Other potential biomarkers include nerve growth factor (NGF) and insulin-like growth factor binding protein 3 (IGFBP3) (10–14).
Inflammatory pathways are also potential sources of biomarkers since they have a well-established role in the development of diabetes. Matrix metalloproteinases (MMPs) are involved in wound healing primarily by degeneration of damaged tissue and breakdown of the capillary basement membrane for angiogenesis, and neutrophil gelatinase-associated lipocalins (NGALs) stimulate MMPs, thus increasing the pro-inflammatory state in diabetes (15).
Corneal nerve damage can act as the bridge between tear biomarkers and DPN findings (16), as IVCM findings have been correlated with early DPN as well as changes in tear biomarker concentrations (5, 9). However, as IVCM equipment is not available in resource-constrained settings such as Peru, a surrogate for detecting corneal damage is needed. Non-contact corneal aesthesiometry (NCCA) and Cochet-Bonnet aesthesiometry are lower-cost mechanisms that have been used to detect corneal sensitivity, with decreased sensitivity implying corneal nerve damage (16–19).
In this pilot study, we explored the screening potential of 31 tear biomarkers involved in both corneal health and inflammatory pathways in detecting DPN. Additionally, we assessed the utility of aesthesiometry for measuring corneal damage in DPN and evaluated the screening capability of these markers for diabetes and diabetic retinopathy. We hypothesize that our findings will be reflective of our understanding of the pathophysiology of nerve damage in diabetes.
2. SUBJECTS, MATERIALS AND METHODS
2.1. Ethics
The study received approval from the Institutional Review Board at Hospital Nacional Cayetano Heredia (HNCH), the Universidad Peruana Cayetano Heredia (UPCH), the Asociacion Benefica PRISMA in Lima, Peru, and Johns Hopkins University in Baltimore, USA.
2.2. Study design and setting
We conducted a screening test pilot study in Hospital Nacional Cayetano Heredia, a national public, tertiary-care hospital located in Lima, Peru with a wide catchment area and large network of specialty care.
Data relevant for sample size calculation was limited but led to a total of nine in each group in order to achieve >80% power to detect a sensitivity (or specificity) of ≥75%, assuming a precision of 15%. To aim for a normal distribution of our data, we enrolled 30 participants per group.
2.3. Study population
Participants included were those aged 18 and older, non-habitual contact lens wearers (not having worn contacts for one month preceding the study), and those able to provide informed consent. Those with a history of alcoholism, vitamin B12 deficiency, autoimmune disease (i.e. Sjogren’s syndrome, psoriasis), HIV infection, thyroid disease, or chronic liver or kidney disease were excluded from the study. People on corticosteroids, doxycycline, or prostaglandin analogues were also excluded, as well as those with eye injuries or infections (within the last two weeks), dry eye disease, or corneal damage. In addition, pregnant women were also excluded.
There were three groups: (1) individuals with diabetes and neuropathy (DN), (2) individuals with diabetes but without neuropathy (DNN), and (3) non-diabetic, non-neuropathic individuals as controls.
2.4. Variable definitions
Type 2 diabetes was defined based on clinical records plus an HbA1c ≥6.5% (48mmol/mol), thereby confirming the diagnosis. DPN was assessed by vibration perception threshold (VPT) testing. VPT was measured with a biothesiometer (Diabetik Foot Care India Pvt Ltd, Chennai, India), a validated method for assessing neuropathy, particularly useful in low-resource settings (20). The maximum value for this machine is 50V. Participants were asked to lie supine, and the stylus of the biothesiometer was applied perpendicular to the pulp on the plantar surface of the hallux of both feet. The amplitude of the vibration was gradually increased until the participant could detect the vibration. This process was performed in triplicate for each foot. An average value of ≥24.5mV in either foot was considered neuropathy (21). Thus, our three groups were defined using a combination of these definitions: (1) diabetes with neuropathy (DN), defined by clinical records, HbA1c ≥6.5% (48mmol/mol), and an average biothesiometer reading of ≥24.5mV in either foot; (2) diabetes without neuropathy (DNN), defined by clinical records, HbA1c ≥6.5% (48mmol/mol), and an average biothesiometer reading of <24.5mV in either foot; and (3) nondiabetic, non-neuropathy controls, defined by clinical records, HbA1c <6.5% (48mmol/mol), and biothesiometry reading <24.5mV.
2.5. Index tests
Variables used as index tests were corneal sensitivity, assessed by aesthesiometry, and tear protein concentrations of substance P, matrix metalloproteinase 9(MMP-9), tissue inhibitor of metalloproteinase 1(TIMP-1), insulin-like growth factor binding protein 3(IGFBP3), lipocalin-2(LCN2), and 26 cytokines: interleukin 1a (IL-1a), IL-1b, −2, −3, −4, −5, −6, −7, −9, −10, −12, −15, −17, epidermal growth factor (EGF), tumor growth factor-alpha (TGF-alpha), granulocyte colony stimulating factor (GCSF), granulocyte-monocyte colony stimulating factor (GM-CSF), interferon gamma (IFNg), GRO-alpha (also called chemokine ligand 1, or CXCL1), monocyte chemotactic protein-3 (MCP3), MCP1, macrophage-derived chemokine (MDC), interferon gamma-induced protein 10 (IP-10), tumor necrosis factor-alpha (TNF-alpha), and vascular endothelial growth factor (VEGF) as analyzed by multiplex bead assay. Corneal sensitivity was measured in each eye by an ophthalmologist using a Luneau Cochet-Bonnet aesthesiometer, and results were averaged from both eyes for analysis. Tears were collected on Schirmer strips and proteins quantified by ELISA and multiplex-bead assays.
2.6. Procedures
A trained nurse recruited participants with known diabetes status from the database of endocrinology clinic patients at Hospital Nacional Cayetano Heredia. The control group was comprised of acquaintances, friends, and family of these clinic patients. An eligibility criteria questionnaire was administered and oral consent obtained to conduct procedures to verify selection criteria. Procedures, performed by an ophthalmologist, consisted of a visual acuity exam, slit-lamp biomicroscopy, intraocular pressure measurement (Goldmann aplannation tonometer, UK), and tear film breakup time assessing for dry eye.
Eligible participants provided written informed consent prior to undergoing study procedures. Study procedures were completed in one or two visits, no more than two weeks apart.
Trained staff recorded socio-demographic and medical information using REDCap (Research Electronic Data Capture) tools hosted at Johns Hopkins University (22). Weight, height, body fat percent (TANITA BF-350 Body Composition Analyzer, Tokyo, Japan) were measured. Blood pressure was also measured in triplicate (OMRON HEM 780, Tokyo, Japan). A 5mL blood sample was collected to measure HbA1c level.
2.6.1. Peripheral Neuropathy Assessment
Vibration perception threshold (VPT) testing was conducted using a biothesiometer as detailed earlier. Additional information was gathered to calculate a neuropathy symptom score (NSS) and neuropathy disability score (NDS) for each participant to more comprehensively characterize neuropathy. These scales are validated, reliable screening tools for assessing the severity of neuropathy (23). A combined NSS+NDS of >10 has been determined to be a good cutoff for neuropathy (24). Distal pedal pulses were checked to rule out non-diabetic etiologies. Two trained personnel performed these procedures without knowledge of the participants’ medical history or diabetes status.
2.6.2. Ophthalmic assessment
A certified ophthalmologist conducted or supervised all ophthalmic assessments. Tears were collected using Schirmer test strips (Gulden Opthalmics, Pennsylvania, USA). A strip was placed in the inferior fornix of the participant’s left eye without anesthetic. After five minutes, the strip was removed with gloves, and tear volume, in millimeters as per Schirmer strip markings, was recorded. This procedure was repeated in the same eye ten minutes later. The Schirmer strips were air dried, sealed in small plastic bags, and stored at room temperature for analysis (8).
Next, corneal sensitivity was measured in each eye using a Luneau Cochet-Bonnet aesthesiometer (Western Ophthalmics, Washington, USA), and fundoscopic exam was performed to assess for diabetic retinopathy. Pupil dilation with tropicamide eye drops (0.5%) was performed prior to indirect ophthalmoscopy as per clinical guidelines. Severity of retinopathy was classified according to the International Clinical Diabetes Retinopathy Scale (25).
2.6.3. Sample Processing
Proteins were extracted from the Schirmer strips using a protocol derived from Qin et. al (8). Briefly, when tear samples were ready for analysis, dried strips were cut into small pieces and transferred into a 0.6mL tube with 200uL (100mM NH4HCO3 50mM NaCl) buffer. Next, the tube was shaken softly for 30 minutes. The liquid was then transferred to another 0.6mL tube, centrifuged at 12000 rpm for five minutes, and the supernatant was used for analysis. Protein extraction was done within three months of storage.
2.6.4. Tear Analysis
Total protein was quantified using the Bradford method using a 1:40 dilution. Five biomarkers of interest were quantified using commercially bought enzyme-linked immunosorbent assay (ELISA) kits (Substance P: Abcam human substance P kit, 1:4 dilution; MMP-9: Thermofisher Quantikine ELISA kit, 1:4 dilution; TIMP-1: Thermofisher Quantikine ELISA kit, 1:10 dilution; IGFBP3: Thermofisher Quantikine ELISA kit, 1:2 dilution; Lipocalin-2: Thermofisher NGAL Human ELISA kit, 1:10 dilution). According to the kit manuals, the lowest limits of detection for Substance P, MMP-9, TIMP-1, IGFBP3, and Lipocalin-2 were 8.04pg/mL, 0.05ng/mL, 0.08ng/mL, 0.08ng/mL, and 0.0065ng/mL respectively.
The remaining 26 cytokines were measured in each sample in duplicate using multiplex bead analysis (Milliplex Human Cytokine/Chemokine Magnetic Bead Panel, MilliporeSigma, Massachusetts, USA). The tear samples were diluted four-fold, and samples were randomized onto the assay plates. Personnel conducting the laboratory procedures were unaware of the participants’ group status. Standard curves were generated per kit instructions and used to calculate concentrations in the tear samples.
2.6.5. Determination of concentration in tears obtained via Schirmer strips
The amount of each biomarker measured in tears extracted from Schirmer strips was expressed as total recovery in either ng/mL or pg/mL. Normalization calculations were based on a previously cited procedure (26). To adjust for varying volumes of sample collected on Schirmer strips per participant, we multiplied the raw concentration obtained from the assay by the total extraction sample volume (0.2mL) to get a total mass of extracted sample. Next, the total mass was divided by the Schirmer strip volume and multiplied by 1000 to get final protein concentration, either in pg/mL or ng/mL.
2.7. Statistical analysis
Tables summarizing continuous variables were presented as mean and standard deviations (SD). Shapiro-Wilk was used to test normality. We used standardized normal probability plots to analyze the dispersion and assess non-normality. Statistical differences between group means were studied using the Mann-Whitney and Kruskal-Wallis tests for variables not normally distributed. Chi-squared tests assessed differences between categorical variables. A stepwise linear regression was also performed. P-values were adjusted for multiple hypothesis testing using the Benjamini-Hochberg correction and filtered using 0.05 as significance cut-off (27). Spearman’s correlation was used to calculate correlation between proteins and aesthesiometry. The receiver-operator characteristic (ROC) curves for all proteins were examined, and optimal cutoff points were determined using the Youden index. We used the following cutoffs for AUC as previously reported: excellent (0.9–1.0), good (0.8–0.9), regular (0.7–0.8), poor (0.6–0.7), and bad (0.5–0.6) (28). All the statistical analyses described were performed using STATA 15.0 (Stata Corp, College Station, TX, USA).
In a post-hoc analysis, we re-classified individuals with diabetes into 3 groups: no, mild, and moderate-severe neuropathy. This was done to distinguish the ability of index tests in predicting varying grades of neuropathy. No neuropathy was defined as a biothesiometer reading of <15V, mild neuropathy as ≥15V and <24.5V, and moderate-severe as ≥24.5V. (24) ROC curves were created to illustrate differences in predictive ability regarding mild and moderate-severe neuropathy.
3. RESULTS
3.1. Population Characteristics
Ninety participants were recruited for the study, but only 89 were included in the analysis due to missing clinical data (30 controls, 29 DNN, and 30 DN). All 89 samples were run through the ELISA assays (Substance P, MMP-9, TIMP1, IGFBP3, and Lipocalin2). Due to volume limitations, 69 were run through the multiplex assay (21 controls, 24 DNN and 24 DN).
Among the population characteristics including all 89 participants, 58.4% were female, and the control group was significantly younger than the DNN and DN group (p<0.01).
Our assessment for dry eye by tear film breakup time was normal for all study participants and did not significantly differ between groups (p=0.49).
Comprehensive neuropathy examination was consistent with biothesiometer results, with a mean NSS+NDS of 11.1(±3.5) in DN individuals, indicative of neuropathy. This score was significantly higher compared to 5.2(±4.2) in the DNN group (p<0.01), but scores did not differ between the DNN group and controls (p=0.13).
Among those with diabetes (n=59), duration of diabetes did not differ between groups (p = 0.76). Additionally, retinopathy was present in 18/59 individuals with diabetes, and 14 of them had concurrent neuropathy (p<0.01). Detailed characterization of the population is shown in Table 1.
Table 1:
Population Characteristics by Study Group
| Study Group | ||||
|---|---|---|---|---|
| Characteristic | Diabetes with neuropathy (n=30) | Diabetes without neuropathy (n=29) | Non-diabetic, non-neuropathic controls (n=30) | p-value (chi-square) |
| Sex: female, n(%) | 14 (46.7%) | 17 (58.6%) | 21 (70.0%) | 0.19 |
| Mean (SD) Age | 62.1 (9.0) | 58.0 (12.0) | 47.4 (15.5) | |
| Age, n(%) | ||||
| <55 years | 5 (16.7%) | 10 (34.5%) | 22 (73.3%) | <0.01 |
| ≥55 years | 25 (83.3%) | 19 (65.5%) | 8 (26.7%) | |
| Education Level | ||||
| Primary | 5 (16.7%) | 5 (17.2%) | 1 (3.3%) | |
| Secondary | 19 (63.3%) | 13 (44.8%) | 12 (40.0%) | 0.04 |
| Superior | 6 (20.0%) | 11 (37.9%) | 17 (56.7%) | |
| SES based on assets | ||||
| Low | 12 (40.0%) | 8 (27.6%) | 10 (33.3%) | |
| Middle | 8 (26.7%) | 9 (31.0%) | 12 (40.0%) | 0.65 |
| High | 10 (33.3%) | 12 (41.4%) | 8 (26.7%) | |
| Health Insurance | ||||
| No | 1 (3.3%) | 4 (13.8%) | 5 (16.7%) | 0.23 |
| Years of diabetes, n(%) | 0.52 | |||
| ≥10 years | 22 (73.3) | 19 (65.5) | -- | |
| Mean (SD) HbA1c | 10.0 (2.2) | 8.1 (1.8) | 5.6 (0.4) | <0.01 |
| HbA1c ≥7% | 26 (86.7) | 19 (65.5) | -- | 0.06 |
| (53mmol/mol), n(%) | ||||
| Current Smoker, n(%) | 5 (16.7) | 0 (0.0) | 3 (10.3) | 0.08 |
| Medication use n(%) | ||||
| None | 0 (0) | 6 (20.7) | -- | |
| Oral Agent | 25 (83.3) | 21 (72.4) | -- | 0.02 |
| Insulin | 14 (46.7) | 7 (24.1) | -- | |
| Previous foot ulcer, n(%) | 8 (26.7) | 2 (6.9) | -- | 0.01 |
| Mean (SD) NSS + NDS * | 11.1 (3.5) | 5.2 (4.2) | 4.2 (3.3) | <0.01 |
| Hypertension, n(%) | 5 (16.7) | 4 (13.8) | 4 (13.3) | 0.92 |
| Mean (SD) BMI | 29.0 (4.4) | 28.8 (4.9) | 29.9 (4.7) | 0.62 |
| BMI | ||||
| <25 kg/m2 | 6 (20.0%) | 6 (20.7%) | 4 (13.3%) | |
| 25–30 kg/m2 | 14 (46.7%) | 12 (41.4%) | 12 (40.0%) | 0.84 |
| >30 kg/m2 | 10 (33.3%) | 11 (37.9%) | 14 (46.7%) | |
| Mean (SD) Body fat %, n(%) | 30.1 (8.9) | 32.1 (7.4) | 34.1 (8.4) | |
| Body fat % | ||||
| Underfat+Healthy | 13 (43.3%) | 15 (51.7%) | 11 (36.7%) | |
| Overweight | 10 (33.3%) | 5 (17.2%) | 7 (23.3%) | 0.42 |
| Obese | 7 (23.3%) | 9 (31.0%) | 12 (40.0%) | |
| Retinopathy, n(%) | 14 (46.7%) | 4 (13.8%) | -- | <0.01 |
| Average aethesiometry (Corneal sensitivity), n(%) | 5.0 (±1.06) | 5.8 (±0.37) | 5.7 (± 0.37) | <0.01 |
NSS = neuropathy symptom score; NDS = neuropathy disability score
3.2. Tear Biomarkers
There was no significant difference in total protein concentration among the three groups (p = 0.66)
All five biomarkers analyzed by ELISA were detectable. Of the 26 cytokines analyzed by multiplex assay, 14 were reliably detectable within the parameters of the assay.
3.3. Tears as a screening tool for DPN
Of 19 proteins analyzed, inflammatory markers MMP-9 and TGF-alpha were nearest significance (adjusted p=0.06 and p=0.07, respectively), and their concentrations strongly trended upward as individuals progressed to DPN. Mean concentrations for MMP-9 were 166.1 (95%CI [49.4 – 282.8]) ng/mL, 476.8 (95%CI [161.1 – 792.5]) ng/mL and 996.6 (95%CI [232.3 – 1760.8]) ng/mL in controls, DNN, and DN groups respectively. Mean concentrations for TGF-alpha were 47.9 (95%CI [21.7 – 74.0]) pg/mL, 147.8 (95%CI [41.8 – 253.8]) pg/mL and 289.6 (95%CI [143.0 – 436.3]) pg/mL in controls, DNN, and DN groups respectively. The complete list of proteins is shown in Table 2.
Table 2:
Tear protein concentration by group
| Study Group | ||||
|---|---|---|---|---|
| Diabetes with neuropathy (n=30) | Diabetes without neuropathy (n=29) | Non-diabetic, non-neuropathic controls (n=30) | ||
| Mean [95% CI] | Mean [95% CI] | Mean [95% CI] | p-value | |
| Total Protein (mg/mL) | 1.1 (±1.51) | 1.0 (±0.79) | 1.2 (±0.21) | 0.54 |
| ELISA proteins | adjusted p-value | |||
| Substance P (pg/mL) | 7710.9 [4920.010501.8] | 6135.0 [4331.1–7939.0] | 7075.1 [5479.8–8670.6] | 0.24 |
| MMP-9 (ng/mL) | 996.6 [232.3–1760.8] | 476.8 [161.1–792.5] | 166.1 [49.4–282.8] | 0.06* |
| TIMP-1 (ng/mL) | 1554.3 [819.5–2289.0] | 1236.4 [691.9–1780.9] | 983.9 [714.8–1253.0] | 0.24 |
| IGFBP3 (ng/mL) | 2473.2 [516.3–4430.0] | 1987.2 [913.9–3060.6] | 782.3 [408.9–1155.6] | 0.69 |
| Lipocalin-2(ng/mL) | 363946.8 [246514.8–481378.9] | 262500.1 [149745.5-375254.8] | 183281.5 [91071.6-275491.4] | 0.20 |
| 26 Cytokine Multiplex Assay (pg/mL) | corrected p-value | |||
| TGF-alpha | 289.6 [143.0–436.3] | 147.8 [41.8–253.8] | 47.9 [21.7–74.0] | 0.07* |
| EGF | 3224.8 [2371.5–4078.2] | 2856.6 [1695.7–4017.5] | 2059.0 [1244.4–2873.5] | 0.15 |
| GCSF | 2765.8 [1551.0–3980.6] | 2155.0 [1433.4–2876.6] | 1072.8 [675.0–1470.6] | 0.72 |
| GRO | 15526.2 [10385.020667.5] | 11619.0 [7653.815584.1] | 9172.6 [5919.012426.1] | 0.33 |
| MCP-3 | 181.2 [95.5–266.8] | 201.1 [82.5–319.8] | 279.5 [159.2–399.8] | 0.54 |
| MDC | 374.8 [159.8–589.8] | 332.7 [163.5–501.8] | 333.1 [188.2–478.0] | 0.46 |
| IL-1a | 196.9 [121.8–272.0] | 146.4 [83.1–209.7] | 115.7 [64.5–166.8] | 0.29 |
| IL-4 | 849.6 [385.2–1314.0] | 681.3 [306.8–1055.7] | 976.7 [447.4–1506.0] | 0.56 |
| IL-7 | 302.5 [140.9–464.2] | 226.3 [108.7–343.9] | 242.5 [140.5–344.4] | 0.68 |
| IP10 | 46421.6 [26783.266060.1] | 54546.8 [31400.277693.4] | 50812.2 [28218.173406.4] | 0.54 |
| MCP-1 | 3348.2 [1384.2–5312.3] | 2311.4 [664.1–3958.6] | 1121.4 [568.6–1674.3] | 0.29 |
| VEGF | 1765.2 [1125.5–2405.0] | 1561.4 [267.0–2113.7] | 1215.9 [777.3–1654.5] | 0.66 |
| IL-15 | 42.4 [20.6–64.1] | 32.2 [18.5–45.9] | 41.8 [18.2–65.3] | 0.52 |
| IL-10 | 47.4 [14.4–80.5] | 36.5 [10.0–63.0] | 59.7 [14.6–104.8] | 0.27 |
=near significance
The ROC analyses identified 3 proteins with AUCs above 0.60 (AUC[95%CI]: MMP-9 0.61[0.44 – 0.77], NEGF 0.62[0.45–0.79] and TGF-alpha 0.66[0.50 – 0.81]). The AUCs of 9 proteins (Lipocalin-2, TIMP-1, Substance P, GRO-alpha, MCP1, MDC, IL-1a, MCP3 and IL-10) fell between 0.55 and 0.60. The AUCs for the other proteins fell below 0.55. The cutoff points calculated were 167.4ng/mL, 1505.9pg/mL and 81.7pg/mL for MMP-9, NEGF and TGF-alpha, respectively.
3.4. Tears as a screening tool for diabetes
Only MMP-9 (adjusted p = 0.04) and growth stem cell factor GSCF (adjusted p = 0.03) remained statistically significant between the DNN and control group.
3.5. Tears as a screening tool for diabetic retinopathy
When individuals with diabetes were grouped by retinopathy, IL-15, IL-7, and IL-10 concentrations were non-significantly increased in individuals with retinopathy (adjusted p=0.30).
3.6. Corneal sensitivity as related to DPN
Corneal sensitivity, assessed by aesthesiometry, was significantly reduced among DN individuals compared to the DNN (adjusted p<0.01) and control group (adjusted p<0.01). Corneal sensitivity did not differ between the DNN and control group (adjusted p = 0.13). Even after adjusting for potential confounders (i.e. age, sex, and smoking) in a stepwise regression involving tear proteins and aesthesiometry, aesthesiometry remained significantly decreased in the DN group compared to the DNN group (adjusted p<0.01).
3.7. Corneal sensitivity as related to tear biomarkers
Among DN individuals, decreased aesthesiometer results were significantly correlated with increased levels of certain inflammatory markers, namely MMP-9(p<0.01), GRO-alpha(p<0.01), TIMP-1(p=0.03), GCSF(p=0.01), IL-1a(p=0.03), IL-4(p=0.03), and IL-7(p<0.01). No significant correlation was found among participants in the DNN group.
3.8. Corneal sensitivity as related to retinopathy
Corneal sensitivity was decreased in individuals with retinopathy versus those without, though not statistically significant (p=0.07).
3.9. Other relevant findings among individuals with diabetes
MMP-9+TGF-alpha combined had an AUC of 0.65, and, using this variable’s information for optimal cutoff point, had a sensitivity of 58% and specificity of 71%. The AUC for corneal sensitivity testing by aesthesiometry was 0.82 (CI 95%: 0.70 – 0.94) with an optimal cutoff point of ≤ 5.8 and sensitivity and specificity of 79% and 75%, respectively. Monofilament testing had a lower AUC of 0.70 (CI 0.56– 0.84) compared with aesthesiometry. Combining tests did not significantly improve the AUC. See Table 3 and Figure 1 for complete details. In our post-hoc analysis, the AUC of aesthesiometry in predicting mild neuropathy was 0.53. See Figure 2 for complete details.
Table 3:
Variables to assess moderate-severe neuropathy among individuals with diabetes. Receiver-operative curve analysis and area under the curve results
| Variable | AUC | Best threshold | Sensitivity | Specificity | CI− | CI+ |
|---|---|---|---|---|---|---|
| MMP-9 | 61% | 167.4 ng/mL | 67% | 58% | 44.28 | 77.24 |
| TGF-alpha | 65% | 81.7 pg/mL | 71% | 58% | 49.68 | 81.22 |
| Aethesiometry | 82% | 5.8 | 79% | 75% | 70 | 94.06 |
| Tears (MMP-9+TGF-alpha) | 65% | - | 58% | 71% | 48.44 | 80.38 |
| Tears + Aethesiometry | 84% | - | 79% | 88% | 71.41 | 95.95 |
Figure 1:
ROC curves comparing (a) individual methods and (b) combinations of methods to predict moderate-severe neuropathy (VPT ≥ 24.5V) among participants with diabetes AUC=area under the curve, T=tears (TGF-alpha+MMP-9), M=monofilament, A=aesthesiometry
Figure 2:
ROC curves comparing (a) individual methods and (b) combinations of methods to predict mild neuropathy (VPT= 15–24.5V) among participants with diabetes AUC=area under the curve, T=tears (TGF-alpha+MMP-9), M=monofilament, A=aesthesiometry
4. DISCUSSION
The present study is the first to analyze the utility of a large spectrum of biomarkers in the tear film for assessing DPN. Our data also contributes to characterizing the protein makeup of the tear film. Although tears did not prove to be an optimal test for detecting DPN, MMP-9 and TGF-alpha strongly trended upward in DPN. As MMP-9 also correlated with corneal aesthesiometry, it has potential to be a useful predictor. Larger sample-sized studies may be able to further evaluate its utility.
More notably, our study is the first to demonstrate by objective measures that aesthesiometry results are decreased in individuals with DPN. We also believe this study is the first to present an optimal cutoff point for aesthesiometry useful for screening for DPN. When considering these results in conjunction with our post-hoc analysis, aesthesiometry may have more utility in predicting moderate-severe DPN and may be more beneficial for determining risk for foot ulceration. Nevertheless, our findings raise the prospect of using aesthesiometry to detect DPN, particularly in resource-poor settings where nerve conduction studies and electromyography may not be available.
The rationale of this study was to identify a simple, objective method to screen for DPN. The Semmes Weinstein monofilament test already exists as a common method to screen for DPN in the primary care setting. However, a recent meta-analysis in 2017 demonstrated that the pooled sensitivity of this tool to detect DPN was 53%, indicating that a more sensitive tool is warranted (29). Biothesiometry is also a validated, objective measure of DPN(20) but is relatively more expensive. As aesthesiometry was demonstrably more accurate than monofilament testing, had a significant correlation with biothesiometry results, and is less costly, it has potential to be a low-cost alternative for DPN screening.
Our finding, associating aesthesiometry and DPN, matches what was reported by Tavakoli et al, with a difference being we used objective measures to define neuropathy. By also proposing an optimal cutoff point for this tool to predict DPN, our results both substantiate and add to this claim (17).
Prior studies have suggested that altered concentrations of tear biomarkers in diabetes could be an indicator of corneal neuropathy and subsequently an early indicator of DPN (9, 30). However, a recent study by Tummanapalli et. al confirmed our result that substance P levels did not differ in type 2 diabetes compared to controls (31). Nevertheless, the study interestingly found that substance P and calcitonin gene-related peptide levels were significantly associated with corneal nerve damage in participants with type 1 diabetes, thus providing another avenue for research of the tear film in diabetes.
The utility of inflammatory markers in tears can further be explored. Several inflammatory markers, notably MMP-9, were elevated in type 2 diabetes in our study. Inflammatory cytokines have been strongly associated with the development of type 2 diabetes (32). Matrix metalloproteinases play an integral role in wound healing. Hyperglycemia can affect the enzyme activity that controls expression of MMPs and their counterparts (TIMPs), creating a prolonged inflammatory response that leads to impaired wound healing and vascular complications (33). Often, both MMPs and TIMPs will be elevated in this pro-inflammatory state due to an overexpression of these biomarkers, a phenomenon seen in our study (34).
Given the complexities of cytokine pathways, authors have suggested using cytokine ratios to allow for better comparisons between data samples because ratios reflect the balance effect of cytokine response better than absolute values (35). Given the low detectability of our assay, we were not able to create ratios for analysis, but this should be considered in future studies.
In vivo confocal microscopy has demonstrated early signs of nerve damage in individuals who develop diabetic neuropathy (30, 36). The present study confirmed the use of aesthesiometry as a surrogate measure for corneal neuropathy in low-resource settings. Studies correlating aesthesiometry with in vivo confocal microscopy will provide us with data on whether aesthesiometry can be an early detector of neuropathy. Additionally, these studies can evaluate the feasibility of performing corneal sensitivity testing during annual diabetes eye examinations to see if it increases screening rates of diabetic neuropathy.
Several studies have looked at the relationship between retinopathy and corneal sensitivity, with a greater percentage showing a non-association (37, 38). Our study did not find an association between aesthesiometry and diabetic retinopathy, nor did the tear proteins analyzed distinguish retinopathy among individuals with diabetes. Both of these findings suggest different mechanisms are responsible for the development of these diabetic complications.
Certain biomarker concentrations in our control group matched those in other studies, but normal levels vary dramatically (35, 39, 40). This variation could arise due to differing age and ethnicity of populations studied as well as different tear collection methods and assay protocols. Using validated, commercial kits whose manufacturers openly publish quality metrics regarding intra and inter-assay precision values increased our confidence in the measured biomarkers. However, in order to allow for comparisons between studies and determine screening and diagnostic capabilities on a larger scale, standardization of a tear collection protocol, storage method, and processing is necessary.
Strengths of this study include the large set of biomarkers we assessed compared to existing studies. Additionally, we used objective and comprehensive measures to characterize neuropathy in study participants. The study also has some limitations. Due to budget limitations, we used HbA1c levels and medical records to group participants in our study. IVCM was not available in Peru, and aesthesiometry was used as an alternative measure for corneal neuropathy. It is important to recognize that aesthesiometry is not a surrogate measure for IVCM, as the two assess different measures. Nevertheless, by using aesthesiometry, we were able to have a marker of corneal pathology to be able to more confidently associate tear biomarker levels with DPN. Additionally, because we only analyzed diabetes cases with an HbA1c>6.5% (48mmol/mol), the probability of finding differences in tear biomarker concentrations was higher than what it probably would be in a community setting when compared to non-diabetic controls. The small sample size of our exploratory study limits the interpretation of results for both the post-hoc analysis and retinopathy analysis, and significance could have been masked. Limited by tear volume and budget due to the amount of proteins analyzed, we ran each sample only once in the ELISAs. Lastly, though both capillary and Schirmer methods are appropriate for analyzing proteins, basal tears collected via the capillary method more accurately represent the natural tear film. We elected to use Schirmer strips for the purposes of our study because their use requires less expertise, does not involve the risk of using glass in the eye, and yields higher fluid volumes for analysis compared to the capillary method.
In conclusion, although tears are easily obtainable and have potential clinical utility in detecting diabetic peripheral neuropathy (DPN), it is an insufficient standalone screening tool for DPN based on the present study. On the other hand, inflammatory tear biomarkers have potential to screen for diabetes and should be further explored. Corneal aesthesiometry proved to be the most sensitive predictor for moderate-severe DPN, and its integration can potentially improve screening rates in poor-resource settings. The present study confirms the use of corneal aesthesiometry as a surrogate measure for corneal damage in low-resource settings, which is subsequently linked to DPN, and proposes an optimal cutoff value for aesthesiometry in DPN screening.
Highlights.
Tear MMP-9 & TGF-alpha levels trend upward in diabetic peripheral neuropathy (DPN)
Tear inflammatory markers (MMPs) have potential to detect diabetes
Aesthesiometry correlates with DPN; better predictor than monofilament testing
Optimal cutoff for aesthesiometry 5.8mm, 79% sensitivity and 75% specificity
5. ACKNOWLEDGMENTS:
The authors acknowledge Dr. Miguel Pinto and Dr. Jorge Reyes Diaz for providing the space to conduct the study, and CRONICAS for providing clinical testing equipment. We acknowledge Dr. Manuela Verastegui Pimentel for providing laboratory space and equipment, and Edith Malaga, Jessy Condori Añazco, Sussana Oad, Grace Trompeter, and Hannah Steinburg for assisting with study procedures.
FUNDING:
This work was supported by the NIH Fogarty International Center Grant #D43TW009340. The funder had no role in the design or conduct of this research.
Footnotes
DECLARATIONS OF INTEREST:
None
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
6. REFERENCES
- 1.Guariguata L, Whiting D, Weil C, Unwin N. The International Diabetes Federation diabetes atlas methodology for estimating global and national prevalence of diabetes in adults. Diabetes research and clinical practice. 2011;94(3):322–32. [DOI] [PubMed] [Google Scholar]
- 2.Villena JE. Diabetes Mellitus in Peru. Annals of global health. 2015;81(6):765–75. [DOI] [PubMed] [Google Scholar]
- 3.Hicks CW, Selvarajah S, Mathioudakis N, Perler BA, Freischlag JA, Black JH 3rd, et al. Trends and determinants of costs associated with the inpatient care of diabetic foot ulcers. Journal of vascular surgery. 2014;60(5):1247–54.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Slovenkai MP. Foot problems in diabetes. The Medical clinics of North America. 1998;82(4):949–71. [DOI] [PubMed] [Google Scholar]
- 5.Tesfaye S, Boulton AJ, Dyck PJ, Freeman R, Horowitz M, Kempler P, et al. Diabetic neuropathies: update on definitions, diagnostic criteria, estimation of severity, and treatments. Diabetes care. 2010;33(10):2285–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Papanas N, Ziegler D. Corneal confocal microscopy: a new technique for early detection of diabetic neuropathy. Curr Diab Rep. 2013;13(4):488–99. [DOI] [PubMed] [Google Scholar]
- 7.Lovblom LE, Halpern EM, Wu T, Kelly D, Ahmed A, Boulet G, et al. In vivo corneal confocal microscopy and prediction of future-incident neuropathy in type 1 diabetes: a preliminary longitudinal analysis. Canadian journal of diabetes. 2015;39(5):390–7. [DOI] [PubMed] [Google Scholar]
- 8.Qin W, Zhao C, Zhang L, Wang T, Gao Y. A Dry Method for Preserving Tear Protein Samples. Biopreservation and biobanking. 2017;15(5):417–21. [DOI] [PubMed] [Google Scholar]
- 9.Markoulli M, You J, Kim J, Duong CL, Tolentino JB, Karras J, et al. Corneal Nerve Morphology and Tear Film Substance P in Diabetes. Optometry and vision science : official publication of the American Academy of Optometry. 2017;94(7):726–31. [DOI] [PubMed] [Google Scholar]
- 10.Lambiase A, Micera A, Sacchetti M, Cortes M, Mantelli F, Bonini S. Alterations of tear neuromediators in dry eye disease. Archives of ophthalmology (Chicago, Ill : 1960). 2011;129(8):981–6. [DOI] [PubMed] [Google Scholar]
- 11.McNamara NA, Ge S, Lee SM, Enghauser AM, Kuehl L, Chen FY, et al. Reduced Levels of Tear Lacritin Are Associated With Corneal Neuropathy in Patients With the Ocular Component of Sjogren’s Syndrome. Investigative ophthalmology & visual science. 2016;57(13):5237–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wu YC, Buckner BR, Zhu M, Cavanagh HD, Robertson DM. Elevated IGFBP3 levels in diabetic tears: a negative regulator of IGF-1 signaling in the corneal epithelium. The ocular surface. 2012;10(2):100–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Symeonidis C, Papakonstantinou E, Galli A, Tsinopoulos I, Mataftsi A, Batzios S, et al. Matrix metalloproteinase (MMP-2, −9) and tissue inhibitor (TIMP-1, −2) activity in tear samples of pediatric type 1 diabetic patients: MMPs in tear samples from type 1 diabetes. Graefe’s archive for clinical and experimental ophthalmology = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. 2013;251(3):741–9. [DOI] [PubMed] [Google Scholar]
- 14.Grus FH, Sabuncuo P, Dick HB, Augustin AJ, Pfeiffer N. Changes in the tear proteins of diabetic patients. BMC ophthalmology. 2002;2:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Yan L, Borregaard N, Kjeldsen L, Moses MA. The high molecular weight urinary matrix metalloproteinase (MMP) activity is a complex of gelatinase B/MMP-9 and neutrophil gelatinase-associated lipocalin (NGAL). Modulation of MMP-9 activity by NGAL. The Journal of biological chemistry. 2001;276(40):37258–65. [DOI] [PubMed] [Google Scholar]
- 16.Dash SK. Corneal Sensitivity Is Reduced and Relates to the Severity of Neuropathy in Patients with Diabetes. Response to Tavakoli et al. 2007;30(12):e142-e. [DOI] [PubMed] [Google Scholar]
- 17.Tavakoli M, Kallinikos PA, Efron N, Boulton AJM, Malik RA. Corneal Sensitivity Is Reduced and Relates to the Severity of Neuropathy in Patients With Diabetes. Diabetes care. 2007;30(7):1895–7. [DOI] [PubMed] [Google Scholar]
- 18.Nielsen NV. Corneal sensitivity and vibratory perception in diabetes mellitus. Acta ophthalmologica. 1978;56(3):406–11. [DOI] [PubMed] [Google Scholar]
- 19.Murphy PJ, Lawrenson JG, Patel S, Marshall J. Reliability of the Non-Contact Corneal Aesthesiometer and its comparison with the Cochet–Bonnet aesthesiometer. Ophthalmic and Physiological Optics. 1998;18(6):532–9. [PubMed] [Google Scholar]
- 20.Jayaprakash P, Bhansali A, Bhansali S, Dutta P, Anantharaman R, Shanmugasundar G, et al. Validation of bedside methods in evaluation of diabetic peripheral neuropathy. Indian J Med Res. 2011;133(6):645–9. [PMC free article] [PubMed] [Google Scholar]
- 21.Pourhamidi K, Dahlin LB, Englund E, Rolandsson O. Evaluation of clinical tools and their diagnostic use in distal symmetric polyneuropathy. Primary care diabetes. 2014;8(1):77–84. [DOI] [PubMed] [Google Scholar]
- 22.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. Journal of biomedical informatics. 2009;42(2):377–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Asad A, Hameed MA, Khan UA, Ahmed N, Butt MU. Reliability of the neurological scores for assessment of sensorimotor neuropathy in type 2 diabetics. JPMA The Journal of the Pakistan Medical Association. 2010;60(3):166–70. [PubMed] [Google Scholar]
- 24.Young MJ, Breddy JL, Veves A, Boulton AJ. The prediction of diabetic neuropathic foot ulceration using vibration perception thresholds. A prospective study. Diabetes care. 1994;17(6):557–60. [DOI] [PubMed] [Google Scholar]
- 25.Wu L, Fernandez-Loaiza P, Sauma J, Hernandez-Bogantes E, Masis M. Classification of diabetic retinopathy and diabetic macular edema. World J Diabetes. 2013;4(6):290–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.VanDerMeid KR, Su SP, Krenzer KL, Ward KW, Zhang JZ. A method to extract cytokines and matrix metalloproteinases from Schirmer strips and analyze using Luminex. Mol Vis. 2011;17:1056–63. [PMC free article] [PubMed] [Google Scholar]
- 27.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289–300. [Google Scholar]
- 28.Safari S, Baratloo A, Elfil M, Negida A. Evidence Based Emergency Medicine; Part 5 Receiver Operating Curve and Area under the Curve. Emerg (Tehran). 2016;4(2):111–3. [PMC free article] [PubMed] [Google Scholar]
- 29.Wang F, Zhang J, Yu J, Liu S, Zhang R, Ma X, et al. Diagnostic Accuracy of Monofilament Tests for Detecting Diabetic Peripheral Neuropathy: A Systematic Review and Meta-Analysis. J Diabetes Res. 2017;2017:8787261-. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Stuard WL, Titone R, Robertson DM. Tear Levels of Insulin-Like Growth Factor Binding Protein 3 Correlate With Subbasal Nerve Plexus Changes in Patients With Type 2 Diabetes Mellitus. Investigative ophthalmology & visual science. 2017;58(14):6105–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Tummanapalli SS, Willcox MDP, Issar T, Yan A, Pisarcikova J, Kwai N, et al. Tear film substance P: A potential biomarker for diabetic peripheral neuropathy. The ocular surface. 2019. [DOI] [PubMed] [Google Scholar]
- 32.Liu C, Feng X, Li Q, Wang Y, Li Q, Hua M. Adiponectin, TNF-alpha and inflammatory cytokines and risk of type 2 diabetes: A systematic review and meta-analysis. Cytokine. 2016;86:100–9. [DOI] [PubMed] [Google Scholar]
- 33.Ayuk SM, Abrahamse H, Houreld NN. The Role of Matrix Metalloproteinases in Diabetic Wound Healing in relation to Photobiomodulation. J Diabetes Res. 2016;2016:2897656-. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lobmann R, Ambrosch A, Schultz G, Waldmann K, Schiweck S, Lehnert H. Expression of matrix-metalloproteinases and their inhibitors in the wounds of diabetic and non-diabetic patients. Diabetologia. 2002;45(7):1011–6. [DOI] [PubMed] [Google Scholar]
- 35.Cook EB, Stahl JL, Lowe L, Chen R, Morgan E, Wilson J, et al. Simultaneous measurement of six cytokines in a single sample of human tears using microparticle-based flow cytometry: allergics vs. non-allergics. Journal of immunological methods. 2001;254(1–2):109–18. [DOI] [PubMed] [Google Scholar]
- 36.Tavakoli M, Quattrini C, Abbott C, Kallinikos P, Marshall A, Finnigan J, et al. Corneal confocal microscopy: a novel noninvasive test to diagnose and stratify the severity of human diabetic neuropathy. Diabetes care. 2010;33(8):1792–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Dogru M, Katakami C, Inoue M. Tear function and ocular surface changes in noninsulin-dependent diabetes mellitus. Ophthalmology. 2001;108(3):586–92. [DOI] [PubMed] [Google Scholar]
- 38.Alvarenga LS, Martins EN, Grottone GT, Morales PH, Paranhos A Jr., Freitas D, et al. Usefulness of corneal esthesiometry for screening diabetic retinopathy. Revista de saude publica. 2003;37(5):609–15. [DOI] [PubMed] [Google Scholar]
- 39.Liu J, Shi B, He S, Yao X, Willcox MDP, Zhao Z. Changes to tear cytokines of type 2 diabetic patients with or without retinopathy. Mol Vis. 2010;16:2931–8. [PMC free article] [PubMed] [Google Scholar]
- 40.LaFrance MW, Kehinde LE, Fullard RJ. Multiple cytokine analysis in human tears: an optimized procedure for cytometric bead-based assay. Current eye research. 2008;33(7):525–44. [DOI] [PubMed] [Google Scholar]