Discriminant Function to Distinguish Age-Related from Sjogren's Keratoconjunctivitis Sicca

Tests of tear function:

(1) lactoferrin concentration, as determined by immunoprecipitation: Lactoferrin is secreted in the tear gland acini but not in the tubular portions of the gland.

(2) wetting of a filter paper strip for 5 minutes (Schirmer test): measures tear fluid production

(3) tear film breakup time: measures tear film instability and is related to the protein concentration in the tear fluid

(4) Rose bengal score: reflects secondary damage to the superficial epithelium of the cornea and conjunctiva

(5) lysozyme: Lysozyme is secreted by both the acini and the tubular structures of the tear gland.

The first 3 were selected for developing a discriminant score.

(Table 1, page 115)

discriminant function =

= (-0.083 * (diameter of lactoferrin precipitation in mm)) - (0.0219 * (wetting of filter paper strip at 5 minutes in mm)) - (0.1414 * (tear break-up in seconds))

Interpretation:

• Only patients who clinically have keratoconjunctivitis sicca should be evaluated.

• If the discriminant function is greater than or equal to the breakpoint selected, then the patient is said to have Sjogren's keratoconjunctivitis.

• If the discriminant function is less than the breakpoint, then the patient is said to have age-related, non-Sjogren's keratoconjunctivitis.

(Table 2, page 115)

Equations from JMP to describe the plot of breakpoint  versus probability of  misclassifying a patient:

percent misclassification of a patient with Sjogren's KCS at a given breakpoint (from 0 to 100) =

= (-4.58 * ((discriminant function)^2)) - (51.375 * (discriminant function)) - 18.499

for breakpoints >= -1.99, the percent misclassification of a patient with non-Sjogren's KCS at a given breakpoint (from 0 to 100) =

= (21.93 * ((discriminant function)^2)) + (114.44 * (discriminant function)) + 144.11