June 2015
Volume 56, Issue 7
ARVO Annual Meeting Abstract  |   June 2015
Conjunctival gene expression differences between aqueous-deficient and evaporative dry eye patients relative to controls
Author Affiliations & Notes
  • Roderick J Fullard
    Vision Sciences, Univ of Alabama Birmingham, Birmingham, AL
  • John L Bradley
    Naval Medical Research Unit, Wright Patterson AFB, Dayton, OH
  • Footnotes
    Commercial Relationships Roderick Fullard, None; John Bradley, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 347. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Roderick J Fullard, John L Bradley; Conjunctival gene expression differences between aqueous-deficient and evaporative dry eye patients relative to controls. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):347.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Purpose: The two main clinical types of dry eye are evaporative (EDE) and aqueous-deficient (ADDE). More severe cases can elicit signs of both dry eye types, so severity scores often combine EDE and ADDE clinical test results into a single “index”. This study tested the hypothesis that, relative to controls, conjunctival inflammatory gene expression profiles differ between ADDE and EDE patients.

Methods: From a larger dry eye cohort (OSDI>20; mean Oxford staining scores >2) patients were selected as either primary ADDE (Schirmer score<5 mm/5 min), n=13, or primary EDE (NIBUT<10s;), n=12. Controls were included based on negative results for the above tests, n=11. Conjunctival impression cytology specimens were collected from 8 sites per eye, RNA extracted, and 96 gene RT-qPCR conducted on TaqMan low density arrays. Predictive modeling weighted voting (GeneSpring) was used to classify patients. Two approaches were used: (1) clinical approach based on 26 genes whose expression correlated significantly with one or more clinical test results, and (2) SAS principal components (PCA) or canonical discriminant analysis (CDA) based on the entire 90 test-gene pool.

Results: (1) The clinical models revealed a cluster of 13 genes (CCL5, CXCL10, CXCR3, FGF2, HLA-DRA, IL10RA, IL23A, IL2RA, IL2RB, ILRG, IL7R, STAT4, TNF) that correctly predicted ADDE vs control patients with 28% ROC error. For EDE versus control, six genes (CCL2, IL2RB, IL8, MUC1, MUC4, and MUC16) correctly predicted patient groupings with a higher 32% ROC error. (2) CDA produced a 15-gene set (CCL2, CXCR3, HLA-DRA, ICAM1, IL10RA, IL18R1, IL1B, IL2RA, IL7, IL7R, MUC16, MUC4, NFKB2, STAT4, TNF) that predicted ADDE vs control patients with 29% ROC error, and EDE vs controls were best predicted by PCA with a 6-gene set (IL1B, TNF, CCL19, CCL2, CXCR3, IL8) and 38% ROC error.

Conclusions: ADDE predictive models had similar success rates for the two approaches to gene selection. Many genes overlapped models. EDE models were less robust, less genes predictive and few overlapped the two approaches. Interestingly, glycocalyx mucins were only identified by the clinical model for EDE. Results suggest a stronger predictive association between ADDE and conjunctival gene expression.


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.