Diabetic Retinopathy Screenings Can Detect Uveal Melanoma Earlier

Photo Essay From The University Hospital Of Bordeaux. Cardiologic Hospital Of Haut Leveque. Department Of Diabetology. Photo Essay From The University Hospital Of Bordeaux. Cardiologic Hospital Of Haut Leveque. Departmnurse’s Aide ; Retinography For The Screening Of A Diabetic Retinopathy. (Photo By BSIP/UIG Via Getty Images)
Diabetic eye examinations enabled detection and treatment of posterior uveal melanomas at an earlier stage, leading to lower mortality.

Detecting malignant uveal melanoma at its earliest — and most treatable — stages may influence metastatic risk and survival. Researchers in the United Kingdom investigated the methods by which uveal melanomas are detected and found that patients with abnormalities found via diabetic screenings had a lower 5-year mortality rate due to earlier diagnosis and treatment. 

Uveal melanoma is the most common primary adult intraocular malignancy. Up to 33% of patients with the disease are asymptomatic. Many present with advanced ocular disease, with about 23% of patients reporting that their tumors were initially missed when they first developed symptoms. These delays in diagnosis can allow the condition to advance and require more aggressive treatment. 

To determine if earlier detection and treatment of smaller uveal tumors is associated with better outcomes and improved survival, investigators analyzed the anatomic stage of posterior uveal melanoma in patients referred initially to an ophthalmology service from the UK’s National Diabetic Retinopathy Screening Service, and if suspicious, referred on to an ocular oncology center. The study’s team sought to test the hypothesis that the thorough photo documentation these patients receive through this referral path would result in earlier detection and treatment of their tumor, thereby potentially improving survival. 

For the analysis, researchers used prospectively collected data of patients diagnosed with uveal melanoma. Diagnosis was determined through the presence of clinical signs of malignancy (e.g., thickness 0.2 mm, orange pigment, serous retinal detachment, and/or documented growth) or by biopsy. This patient pool was compared with randomly selected controls from the complete oncology database matched by date of birth and sex. Matching was performed by an in-house software program. To do this, the matching algorithm loaded all of the oncology database patients with posterior uveal melanoma (n = 6167) and cases detected by diabetic screening (n = 132). The order was randomized, and then the algorithm extracted five control patients for each case patient. This produced a data set of 660 controls, but 52 were excluded because the melanoma treated either originated in the conjunctiva or iris.

There were 444 men and 164 women in the control group and 96 men and 36 women in the screening group (P =.96). Age at detection of uveal melanoma was higher in the diabetic screening group (range, 27 years to 88 years; mean, 67.5 years) than in the control group (range, 22 years to 92 years; mean, 62.8 years) which is statistically significant (P =0.001). Systemic comorbidities differed across groups significantly, with much higher rates of hypertension, ischemic heart disease, hypercholesterolemia, and concurrent systemic malignancies in the group referred from screening. Of those detected via alternate methods, 6% of patients were diabetic.

The diabetic screening group of 132 cases was compared with the 608 controls using Kaplan–Meier survival probabilities and competing risk analysis. For all-cause mortality, Kaplan–Meier and cumulative incidence estimates were identical. In the control cohort, the all-cause mortality at 10, 20, and 25 years after the diagnosis was 36%, 54%, and 56%, respectively. In the diabetic screening cohort, the all-cause mortality at 10, 20, and 25 years after diagnosis was 33%, 64%, and 64%, respectively. Differences in the mortality rates between the diabetic screening group and controls were not statistically significant for all-cause mortality (P =.8). However, there was a statistically significant difference for metastatic mortality (P =.016).

“Our study demonstrates a significantly lower mortality rate from metastases in patients detected and diagnosed with uveal melanoma following an annual routine screening despite a higher rate of systemic comorbidities and consequently a higher rate of death from cardiovascular events and other concurrent malignancies,” the study’s authors write. “We may postulate that this is related to detection of the tumor before further growth and also perhaps before the progression of the tumor to a higher-risk cytological and genetic profile.”

Regarding the limitations of the study, investigators acknowledge that while they tried to ascertain the lead-time bias for their cohort of patients, it’s a challenge. They used other means of calculating the likelihood of lead-time bias by using age at diagnosis as a proxy. Another potential concern is length-time bias, which holds that some of the lesions detected at screening may never have led to symptomatology or systemic sequelae within the lifetime of the subject due to their low aggressive nature. These factors can only be addressed with prospective long-term randomized trials and impacts on all considerations of screening strategy. To adjust for this, other statistical methods have been used to determine the impact of lead- and length-time bias in the absence of a known “time from initial detectable lesion to symptomatology.”


Hussain R, Czanner G, Taktak A, et al. Mortality of Patients with Uveal Melanoma Detected By Diabetic Retinopathy Screening. Retina. 2020;40(11):2198-2206. doi: 10.1097/IAE.0000000000002763.