Detecting sarcoidosis-associated uveitis aided with new approach

Statistical modeling approach called latent class analysis used to identify SAU

Steve Bryson, PhD avatar

by Steve Bryson, PhD |

Share this article:

Share article via email
The eye of a person using a telescope to look at stars is shown on its lens.

A new modeling approach that incorporates standard tests and clinical signs can identify people who have sarcoidosis-associated uveitis (SAU), or inflammation of the eye, a new study shows.

Based on this analysis, data showed a combination of eye tests could distinguish SAU from non-SAU patients with a high degree of accuracy. The results “could be suggestive of future improvement in SAU diagnostic criteria,” the researchers wrote in “Ocular Signs and Testing Most Compatible with Sarcoidosis-Associated Uveitis: A Latent Class Analysis,” which was published in Ophthalmology Science.

Sarcoidosis, a condition wherein small clumps of immune cells called granulomas form, is a significant cause of uveitis, an inflammation of the eye’s middle layer. Up to 80% of cases have uveitis as the presenting feature.

Diagnosing uveitis is painless and straightforward. It involves looking for inflammation in the eye through a microscope. Confirming if a person with uveitis also has sarcoidosis is more challenging, however, because a sarcoidosis diagnosis usually involves a highly invasive tissue biopsy to find granulomas.

Recommended Reading
lymphopaenia biomarker

Inflammation Limited to Front of Eye Can Aid Recovery in Ocular Sarcoidosis

A new approach to identifying SAU in uveitis

Scientists in the U.S. and Japan conducted a latent class analysis (LCA) to help identify people with SAU among uveitis patients. LCA is a statistical modeling approach that identifies people who share characteristics, allowing distinct “clusters” to be detected. The study included 826 people with uveitis seen at a six-month follow-up who had laboratory results. The LCA incorporated recommended tests and clinical signs from the revised International Workshop on Ocular Sarcoidosis (IWOS), which are guidelines for diagnosing ocular sarcoidosis.

LCA modeling generated two classes that best fit the data. Class 1 consisted of 548 people and represented those without SAU, while class 2 was made up of 278 participants and was most representative of SAU. Compared with class 1, class 2 had a higher proportion of people of Asian descent, female, and who received a sarcoidosis diagnosis and/or had uveitis the clinician suspected was actually SAU.

On imaging scans, more participants in class 2 tested positive for bilateral hilar lymphadenopathy (BHL), an enlargement of lung lymph nodes, than in class 1 (47.8% vs. 3.6%). Similar results were seen for granuloma-positive biopsies from the lung, lymph node, skin, or conjunctiva (30.2% vs. 4.6%).

Consistent with these findings, class 2 participants were more likely to show SAU signs, including mutton-fat keratic precipitates, that is, clumps of immune cells that form on the clear front part of the eye, and iris nodules, or opaque patches known as “snowballs/string of pearls” in the vitreous, the liquid-filled space inside the eye.

The most significant predictors of a class assignment included snowballs/string of pearls vitreous opacities, vascular inflammation, called periphlebitis, and/or macroaneurysm, that is, dilation of eye blood vessels, bilaterality, which means both eyes are affected, and BHL.

The combination of these four tests yielded a diagnostic sensitivity of 84.8% and a specificity of 95.4%. Sensitivity refers to a test’s ability to correctly identify patients with SAU (class 2) and specificity refers to the likelihood of correctly identifying people without it (class 1).

A subgroup analysis found the separation between the classes was best when only Japanese participants were included. In comparison, the model didn’t show a good separation between classes using participants from other regions.

A third classification

The researchers also conducted a sensitivity analysis for a three-class model because its fit to the data was comparable to that of the two-class model.

Similarly, class 1 included 468 participants and showed the lowest proportion of eye signs or abnormal lab results, matching the non-SAU class. Class 2 consisted of 142 people who were more likely to present with ocular signs or abnormal lab tests, most represented by the SAU class.

Class 3, which had 216 participants, had eye signs and lab test results between classes 1 and 2. It had the highest number of participants with multiple chorioretinal peripheral lesions, or several areas of damage to the retina, the photoreceptor cells at the back of the eye, and the choroid, which is a thin layer of tissue in the middle layer of the eye wall.

The sensitivity of other tests, such as antimicrobial lysozyme, alanine transaminase (a marker for liver injury), and lactate dehydrogenase (a tissue damage marker), was higher in class 3 than class 2.

The researchers proposed, based on these findings, that class 3 may represent three patient groups — one without SAU, an SAU subtype with lung involvement, and another SAU subtype with less lung involvement.

“Latent class modeling, incorporating tests and clinical signs from the revised IWOS criteria, effectively identified a subset of participants with clinical features indicative of SAU,” the researchers said. “Using a combination of tests provided a satisfactory performance in classifying the SAU subclasses identified by the 2-class LCA model.”

“The classes identified by the 3-class LCA model … may have potential implication for clinical practice, and hence should be validated in further research,” they wrote.