Contrast sensitivity function under three light conditions in patients with type 1 diabetes mellitus without retinopathy: a cross-sectional, case–control study

Diabetes mellitus (DM) is a chronic, metabolic disease in which there is high blood-glucose levels, which could cause serious injury of the heart, blood vessels, eyes, kidneys, and nerves over time. According to the World Health Organization and International Diabetes Federation (IDF), DM is one of the fastest-growing global health emergencies in the 21st century, being considered a public health problem [1, 2]. The onset of type 1 DM (T1DM) is associated with younger people, and it is characterized by the lack of production of insulin due to an immunologic system attack [3, 4].

In the evolution of DM there could be several structures affected, such as modification in blood vessels [5] or neuropathies [6] that generally lead to diabetic foot, renal disease or eye complications, [7] mainly diabetic retinopathy. The above being said, it is known that in the retina, even with early diagnosis, at the time of DM detection, it could be observed that microangiopathy affects small retinal vessels, with a major vascular permeability, ocular haemorrhage, and lipid exudates [5]. Thus, the importance of analysing signs in preclinical diabetes aids the possibility of sooner diagnosis.

The study of visual disturbances using non-invasive tests in the DM preclinical phase leads us to consider examining visual function through analysing visual acuity (VA) or contrast sensitivity function (CSF) measurements. VA might be defined as the ability to discern between separate points and identify shapes, CS is the minor contrast detected or the possibility of differentiating two points which are close, and CSF is the contrast detection threshold measured in four or more circles with bands of varying spatial frequencies. Whilst in the standard Snellen test the only aspect considered is the letter size with a high contrast (black-on-white), observing the ability to resolve fine detail, CS measurement allows us to analyse the capacity to detect low-contrast objects of various sizes. Therefore, the qualitative information offered by CSF values concerning visual function is higher than that of VA. It is common for a person to reach the smallest letter on the card on a VA test but report visual impairment or discomfort that often corresponds to reduced CS [8]. Besides, CSF is affected by ambient lighting changes. Generally, high ambient lighting leads to better CSF. However, pupil contraction in higher light conditions must be considered. This reduces aberrations as well as diminishes foveal illumination [9]. Hence, to make a complete diagnosis of a patient's visual function, different lighting and viewing conditions must be considered [10]. CSF values can be achieved using many procedures and tests, such as measurements by variable contrast optotype (Pelli-Robson, Rabin Test or Mars Letter Test) [11] or sine-wave gratings of different spatial frequencies (CSV-1000 test or OPTIC 6500 test) [12, 13]. CSF values could be different with age and the use of several drugs [14].

The impairment of CS in people with DM has been studied by several authors. A review on psychophysical assessments of diabetic retinopathy (DR) included some studies on CS and people with DM with and without diabetic retinopathy, concluding that CS is a more sensitive test than VA for identifying early changes in visual function and signs of neurodegeneration in people with diabetes [15]. However, most of these studies were performed in patients with type 2 DM (T2DM) or do not distinguish the type of DM. Since the course of diabetes is different in patients with T1DM and T2DM, it is advisable to study them separately [16]. A systematic review on the assessment of CS in the early detection of DR has recently been published. This systematic review included 21 comparative cross-sectional studies, published in the last 11 years. It was observed that patients with DM and DR showed a deterioration of CS, however, those without signs of retinopathy showed a lower CSF than control patients, not always significantly, and without unanimity in the affected frequencies. Furthermore, there are few studies on CSF in T1DM without DR and CSF measurements were not performed in different light conditions [17].

The purpose of this study was to determine whether patients with T1DM without any sign of DR have any alteration in CSF, in relation to patients without this disease, and whether CSF assessment in three different light conditions can be an effective test in the early detection of DR.

This study analysed the CSF in three different light conditions, in people with T1DM who had no signs of DR, and in people with similar characteristics (age, sex, RE and VA) without this disease. The results showed a generalized decrease in CS in the T1DM group with respect to the control group in the three light conditions investigated, despite these patients having good VA (p > 0.05). The most affected SF was found to be the highest (18 cpd), with a statistically significant difference between groups of 0.08 log units at high, 0.10 log units at medium and 0.16 log units at low-light conditions in the mean CS values, and the lowest decrease in CS at the lowest SF (3 cpd) (p > 0.05).

Other authors have evaluated CS in people with T1DM [17, 23] with different tests [24, 25]. Georgakopoulos [26] using the CSV-1000 test, showed an impairment of CS in all spatial frequencies in patients with T1DM compared to controls, with losses between 0.10 and 0.13 log units, which were statistically significant, as in our study, although, the light conditions were not specified. Also, Krasny [23] included patients with non-proliferative diabetic retinopathy. The CSV-1000 was also used in the studies by Heravian [27] and Safi [21] however, these studies mainly involved people with T2DM, with non-proliferative diabetic retinopathy and with a higher mean age (51.0 ± 9.95 and 48 ± 6 years, respectively), without specifying the light conditions. Only Safi evaluated CS in two light conditions (500 lx and < 2 lx). Our results agree with this study in the lower CS in the diabetes group compared to the control group, but not in the SF in question, as Safi showed statistically significant differences in all SF at < 2 lx, and at 500 lx in all SF except at 3 cpd [21].

As shown in Fig. 1, the differences in CS between T1DM and non-T1DM participants are more pronounced in low-light conditions (0.11 log units in mean across SF) than in high-light conditions (0.07 log units in mean across SF), consistent with the results of Safi et al. [21]. This finding could be related, in addition to the aberrations generated by a larger pupil size in low illumination, to the illuminated area of the retina (influence of rod and cone activation) [17]. On the other hand, we did not find statistically significant differences among the three light conditions (high, medium, and low) in CS loss in people with T1DM. This is probably because the CSV-1000 test is not sensitive for detecting minimal changes [28], and the possible test-retest variability [29‐31], therefore, it would be advisable to use a more sensitive test in a future study.

In other studies, on CS in different illuminations, instead of ambient illumination, as in this study, the luminance of the test stimulus was modified by changes in the luminance of the screen and/or the use of neutral density filters to modify foveal information [32], however, the aim of this study was to observe in which room illumination it could be useful to assess CS for early detection of DR. It would be interesting to see whether changes in test luminance would cause greater effects than those observed.

Some authors indicate that in patients with T1DM there may be neuronal changes in the retina, in addition to and independently of vascular changes [25, 33, 34], Studies in patients with diabetes found a thinning of the nerve fibre layer thickness, more consistent with loss of retinal ganglion cells and their axons [35]. Moreover, several investigations have described a thinning of the central retinal thickness in patients with pre-diabetes and diabetes without any sign of DR [36‐38]. CS represents the quality of central function and is largely mediated by ganglion cells. This would imply an impairment of CSF in people with DM and the potential of its evaluation in screening for early detection of complications [15]. Future research will be necessary to establish reference threshold values for the measurement of CS in patients with T1DM [39].

Electroretinography (ERG) and optical coherence tomography (OCT) have also been investigated as clinical tests for the early detection of DR, with satisfactory results in both [36, 40‐46]. This corroborates the existence of vascular and/or neuronal changes prior to the signs visible in fundus images, which can be detected using these techniques. However, in our opinion, CSF assessment could be a good option for monitoring the patient's visual quality and follow-up. It is a less costly test than those proposed and simple to perform.

Regarding the relationship between CS and other variables, we found a statistically significant correlation with age, as has been reported in other studies [14]. No statistically significant correlation between CS and HbA1c in the T1DM group was found, coinciding with the results of Safi et al. [21]. The correlation obtained between CS and HbA1c in the total sample is not comparable, since Safi did not measure HbA1c in the control group. Geogakopoulos et al. [26], obtained a significant correlation of HbA1c and CS at 3 cpd, but without specifying the ambient light condition under which it was measured.

As for the correlation between CS and diabetes duration, an inverse correlation was obtained at 12 cpd in low-light conditions, while the other authors found no relationship. This may be due to the fact that in Georgakopoulos' study, patients were younger and had a shorter duration of diabetes (5.9 ± 3.8 years) [26], and in Heravian's study, although they had a shorter duration (5.9 ± 3.3 years), they were mostly T2DM and older [27]. Therefore, the relationship between greater CSF impairment with longer duration of T1DM diabetes is not clear, as the patient's glycaemic control also plays a role [17], nor is it clear with higher HbA1c levels. Several investigations have confirmed the influence of glucose variability on the development of diabetic microangiopathy, with a different behaviour between people with T1DM and T2DM. In the case of people with T1DM, an increased risk of retinopathy has been related, not only to high HbA1c values, but also to a high variability in these values [16].

We consider the sample size to be a limitation in this study. Indeed, a larger number of patients would have allowed us to study the behaviour of CS as a function of age and to observe whether it is the same in minors and adults. On the other hand, due to the cross-sectional design of the study, it is not possible to establish a relationship between T1D and CSF, and long-term results would be needed, thus, these results should be considered with caution. Future longitudinal studies in which CSF is followed up, would be necessary to corroborate these results and to establish the relationship with blood glucose.

This lower CS in people with T1DM, although it may have been statistically significant in some frequencies, depending on the illumination, was not clinically significant, as it was not enough to be able to discriminate when the vascular signs of DR would start, therefore, the follow-up of these patients over time is fundamental for early diagnosis.

The results of this study show that people with T1DM, without signs of diabetic retinopathy, presented a loss of CS to sine-wave gratings, with respect to people with the same characteristics without this disease, mainly in medium and high frequency, and in medium and low-light conditions. This loss of CS is related to age, however, its relationship with HbA1c values and the duration of diabetes is not so clear. Future longitudinal studies will be necessary to establish a threshold for detection of diabetic retinopathy by CSF assessment.