Colour vision in AIDS patients without HIV retinopathy
Author links open overlay panelJoergSommerhalderaEdoardoBaglivoabChristineBarbeyaBernardHirschelbAndreRothaMarcoPelizzoneaShow morehttps://doi.org/10.1016/S0042-6989(98)00011-XGet rights and contentUnder an Elsevier user licenseopen archive
Abstract
Patients suffering from AIDS develop ocular complications, the most frequent being HIV retinopathy. It is however not clear, if functional visual impairments can be observed as early indicators of ocular complications, before clinical diagnosis of HIV retinopathy is made at fundus examination. To address this issue, we measured colour vision in a group of 49 AIDS subjects with normal clinical fundi using the `two equation method'. This method, combining red-green Rayleigh and the blue-green Moreland metameric matches, enables more complete and quantitative assessments of colour vision than those based on pigmentary tests. Data were collected on our computer controlled colorimeter and compared to those of normal subjects. While most AIDS subjects without HIV retinopathy demonstrated normal colour vision, a significant portion of them had wider matches than normal subjects (11% for the Rayleigh equation and 16% for the Moreland equation). Furthermore, matching ranges of the Moreland equation were significantly correlated with CD4 lymphocyte counts. Patients with low CD4 values tended to produce larger matching ranges than the patients with high CD4 values. A within subject study on 17 patients confirmed this trend and showed that the patients who increased/decreased their CD4 blood counts generally improved/impaired their colour discrimination in the Moreland match. No such correlation was found between the matching ranges of the Rayleigh equation and the CD4 counts. These results show that colour discrimination is slightly reduced in some AIDS subjects, although there are no detectable ocular complications. They also suggest two different types of colour vision impairments in AIDS patients without retinopathy: one reversible process affecting colour discrimination in the blue-green range; and another irreversible process affecting colour discrimination in the red-green range.
Keywords
AIDSColour discriminationColour matchingColour vision testing
1. Introduction
Since 1982, it has become obvious that 90% of the patients suffering from Acquired Immunodeficiency Syndrome (AIDS) develop ocular complications [1]. Among ocular manifestations, HIV retinopathy represents the most common manifestation observed in these patients and is clinically detected in approximately two thirds of them [2]. Some studies have demonstrated that these anomalies are present in all patients when tested by fluorescein angiography or at post-mortem examination 3, 4. It remains unclear, however, if HIV retinopathy is the consequence of an infection of the vascular endothelium by HIV, the consequence of vascular damage resulting from circulating immune complexes, the consequence of an abnormal retinal blood flow, or a mixture of these.
Recently various authors have reported deficits in colour vision in AIDS patients. Quiceno et al. [5]found significantly higher error scores for the Farnsworth 100-Hue test in AIDS patients than in patients with AIDS related complex, HIV-positive patients or normal controls. They did not identify a specific axis for these colour vision impairments but correlated them with HIV retinopathy. Geier et al. [6]confirmed this observation when they found a clear correlation between colour contrast sensitivity in all three axis and the severity of HIV retinopathy (number of cotton wool spots). They also found (weaker) correlations between Walter Reed staging and colour contrast sensitivity, or between CD4 lymphocyte counts and colour contrast sensitivity. Finally, Kozak and Bullimore [7]found that even HIV-positive patients without HIV retinopathy can have S-cone/tritanopic abnormalities. Possible explanations for the mechanisms leading to such colour vision impairment were largely discussed in these papers but no certain etiologic factor has been determined up to now. However, colour vision impairments in patients with normal clinical fundi may be early indicators for upcoming ocular involvement in AIDS disease.
The aim of this study was to find out whether measurable modifications of colour vision in AIDS patients without clinical HIV retinopathy could be observed using the `two equation method'. This method, which combines the red-green Rayleigh equation and the blue-green Moreland equation, is based on metameric colour matches 8, 9. It was demonstrated 10, 11, 12, 13that metameric matches enable more powerful quantitative assessments of normal and defective colour vision than pseudo-isochromatic or arrangement tests (e.g. Farnsworth 100-Hue, Ishihara plates or tests on computer graphics systems).
2. Methods
2.1. Subjects
49 Patients (13 females and 36 males, 94 eyes) with AIDS attending the HIV/AIDS Clinic of the Geneva University Hospital participated in this study. Age ranged from 29 to 49 years (mean and median age 37 years). CD4 lymphocyte counts ranged from 4 cells/ml to 600 cells/ml (median 70 cells/ml). All patients were Caucasians except two of Afro-Caribbean origin.
The patients were selected on the basis of two complete ocular examinations (visual acuity, slit-lamp biomicroscopy, tonometry, and direct and indirect ophthalmoscopy) performed by one of us (EB). The first examination enabled us to preselect patients showing no ocular complication. The second examination, performed one month later, served as a control and only patients showing no ocular complication at both examinations were included in the study. At the time of participation in this study, none of them had an acute opportunistic infection anywhere and all of them had normal clinical fundi. Inclusion criteria were: (1) age of 18 years or more; (2) visual acuity of 0.8 or better; and (3) normal clinical fundi. Exclusion criteria were: (1) a history of congenital or acquired eye disease; (2) diabetes; (3) drug abuse; (4) myopia demanding a correction of more than −3D; and (5) other diseases influencing colour vision (e.g. vascular eye diseases related to arterial hypertension or sudden hearing loss). Informed consent was obtained for each patient and a good motivation for the test was demanded. When tested all patients were on customized individual anti-retroviral therapy such as Bactrim® (trimethoprim-sulfamethoxazole), Retrovir® (AZT) (Geier et al. [14]reported in a case report on two patients a temporary alteration of tritan colour contrast sensitivity at the beginning of Zidovudine (AZT) treatment; none of our patients was at the beginning of an AZT treatment), 3TC® (Lamivudine), Zerit® (Stavudine) or Hivid® (Zalcitabine) and some of them received protease inhibitors such as Invirase® (Saquinavir), Norvir® (Ritonavir) or Crixivan® (Indinavir).
All patients had one full colour vision examination. Furthermore, we asked the patients to come back for a second test to investigate possible individual changes in their colour vision. Only 17 patients (34 eyes) out of 49 actually came back for a second examination. The time between the first and second test varied from 2 to 20 months.
2.2. Colour vision assessment and statistical analysis
All colour vision tests were conducted by the same person (JS) on our laboratory computer-controlled anomaloscope (a commercial version of this instrument is now available from Interzeag AG (Schlieren, Switzerland) but the 7° observation field size is however not yet implemented in this miniaturised instrument) with four independent light channels [15]. The Rayleigh equation (670+545≡589 nm) on a 2° circular observation field was always tested first. Afterwards the Moreland equation (490+436≡desaturated 480 nm) was tested on a 7° observation field. After each colour vision test the optical density of the crystalline lens was measured with the Lens Opacitometer 701™ (the Lens Opacitometer 701™ is available from Interzeag, Switzerland; for more technical details see [16]) to exclude colour vision impairment due to abnormal optical lens density.
The colour vision testing procedure was exactly identical to the one used for a previous normal population study [17](a more detailed description of the testing procedure itself and the statistical analysis of the subject's responses can be found in 18, 19). The result of the colour vision test are two values expressed on a 0–100 scale: the match-mid-point (MMP), assessing the perceptual relationship between the colours defined by the mixture field primaries (used to equalize the reference primary), and the matching range (MR) assessing colour discrimination around the reference primary.
The data of the previous normal population study (100 subjects, mean age 29 years) were used to establish reference distributions and limits for normal subjects. Student's t-test was used to assess the statistical significance of differences between group means, while the F-test was used to compare variances of group distributions. The significance of correlations between CD4 blood counts and colour vision results was tested using correlation coefficient testing [20].
3. Results
The results of our colour vision measurements on the AIDS subjects are summarized in Fig. 1 (the MMP values of our reference normal population were slightly shifted towards the lower wavelength primary, primarily to compensate for the mean age difference of 8 years between the AIDS patients and the normal subjects; for the influence of age on colour vision results see 21, 22, 23). While most AIDS patients without retinopathy had normal colour vision, 10 (11%) out of 92 eyes exceeded the normal limits in the Rayleigh equation and 15 (16%) out of 92 eyes exceed the normal limits in the Moreland equation. Surprisingly, only 3 eyes showed abnormal results in both equations. Since the normal limits were computed to be exceeded in only 5% (by definition) of the cases in a normal population, abnormal results appear to be slightly more frequent in our AIDS patient data. More detailed statistical analysis were done to see if this trend was significant.
Fig. 1. Colour vision in 49 AIDS patients. MR vs. MMP. The triangle indicates the region of possible results. The rectangle at the bottom indicates the region of normal results. Results lying outside the rectangle are considered as abnormal colour vision results.
3.1. MMP
Histograms of all the MMP values measured on the AIDS subjects are presented in Fig. 218, 19. These data were fitted to a normal-distribution and compared to those obtained (by exactly the same procedure) on normal subjects. There is no statistically significant difference between the MMP distributions, either for the Rayleigh equation (unpaired t-test P=0.36) or for the Moreland equation (unpaired t-test P=0.69).
Fig. 2. Histograms for the MMP of 49 AIDS patients. A normal distribution is fitted to the data and compared with the distribution for normal subjects.
3.2. MR
Histograms of the MR values measured on the AIDS subjects are presented in Fig. 3. These data were fitted to a log-normal distribution (for more details concerning our data analysis using log-normal distributions see [17]). Comparisons with the distributions obtained (with exactly the same procedure) on normal subjects reveal that the AIDS patients had significantly larger MRs on both equations (unpaired t-test P=0.028 for the Rayleigh equation; P=0.002 for the Moreland equation). A close look at the fitted curves shows that the log MR distribution for the Moreland equation is simply shifted towards higher values, while the log MR distribution of the Rayleigh equation is not only slightly shifted towards higher values but also significantly larger than the log MR distribution for the normal subjects (F-test P=0.006).
Fig. 3. Histograms for the log MR of 49 AIDS patients (the finite size of the smallest steps in the iterative algorithm determining the MR results in artefactual clustering of data for extremely small matching ranges, therefore we ignored these data before fitting log MR with a normal distribution which explains the empty lower parts under the fitted normal distributions). A normal distribution is fitted to the data and compared with the corresponding distribution for normal subjects.
3.3. Correlation with the CD4 blood count
The only parameter which was found to be significantly correlated with the CD4 lymphocyte counts is the MR of the Moreland equation (Fig. 4a, P<0.05) (we could not find any significant correlation between colour vision results and other parameters, such as best corrected visual acuity or lens opacity). Patients with low CD4 values have a tendency to produce larger MRs than patients with high CD4 values. Moreover, for the 17 patients tested twice, we found a significant correlation between the ΔMR of the Moreland equation and ΔCD4 counts (Fig. 4b, P=0.01), confirming the correlation between the MR of the Moreland equation and the CD4 blood counts. CD4 count improvement results generally in MR improvement. It is interesting to note that in the case of these 17 patients, we have a within subject study where individual variability of colour vision does not smear out the results.
Fig. 4. (a) Log MR for the Moreland equation vs. CD4 lymphocyte counts (artificially low MR values were uniformly fixed at a MR value of 0.4 units; there is no change in significance if we ignore these data when computing the regression). CD4 values are indicated on a logarithmic scale. (b) Log MR change in the Moreland equation vs. CD4 count change for 17 AIDS patients. Positive values indicate higher values at the second examination (positive ΔCD4 values≡improvement of the CD4 lymphocyte counts; negative Δ log MR values≡improvement of colour discrimination).
4. Discussion
This study shows a slight impairment of colour vision in some AIDS patients with normal clinical fundi. In these cases, the effect of the HIV disease is essentially reflected by a reduced colour discrimination in the Moreland and/or the Rayleigh equations. Our results for the Moreland equation confirm earlier results by Kozak and Bullimore [7]who found reduced S-cone contrast sensitivity for HIV patients without clinical signs and a very recent study of Muller et al. [24]who have reported an elevated error scores for the Farnsworth–Munshell 100-Hue test in HIV patients without HIV retinopathy.
We found significant correlations between the CD4 blood count values and colour discrimination in the Moreland equation: (1) in all tested patients, higher CD4 blood counts were significantly correlated with better colour discrimination; and, (2) in the 17 patients tested twice, improvements/impairments in CD4 blood counts were significantly correlated to improvements/impairments of colour discrimination. Geier et al. [6]had found a significant correlation between CD4 blood counts and tritan/protan colour contrast sensitivity in a group of subjects including patients with HIV retinopathy, while Mueller et al. [24]did not find any significant correlation between Farnsworth 100-Hue error scores and CD4 blood counts in their subjects without HIV retinopathy. The discrepancy of our results with the latter study is probably due to the lesser sensitivity and specificity of the Farnsworth 100-Hue test.
The observed colour vision impairment in AIDS patients without HIV retinopathy and the correlation between their MR results for the Moreland equation and CD4 lymphocyte counts can not indicate the exact mechanisms leading to HIV retinopathy. Some medicaments are known to affect colour vision [25]. Although the medication might cause the observed impairments in colour discrimination, this is very unlikely, because increases in CD4 blood counts were correlated with improved colour discrimination. Furthermore, these improvements suggest that the pathological processes affecting colour vision could be (at least partially) reversible.
While colour vision was impaired in a similar number of patients in the Rayleigh and in the Moreland equation, it is interesting to note that only 3 cases had abnormal colour vision results in both equations. Moreover, in contrast to the Moreland equation, colour discrimination in the Rayleigh equation was not significantly correlated with the CD4 counts. As a consequence, one may postulate two types of colour vision impairment in AIDS patients without HIV retinopathy: one process, affecting colour discrimination in the Moreland equation, which is reversible; and a second process, affecting colour discrimination in the Rayleigh equation, which is apparently irreversible (a learning process would not only affect the Moreland equation but also the Rayleigh equation and it would not be correlated to CD4 counts, thus, a learn effect can therefore be excluded). This hypothesis needs to be confirmed by more measurements and the long term follow up of AIDS patients.
Impairment of colour vision can be an indicator for subsequent ocular complications. Early detection of such ocular complications is becoming more and more important since successful research in AIDS treatment leads to longer life expectation for these patients.
References
1G.N Holland, A Tufail, M.C JordanCytomegalovirus diseasesOcul Infect Immun, 81 (1996), pp. 1088-1129View Record in ScopusGoogle Scholar2D.A Jabs, W.R Green, R Fox, et al.Ocular manifestations of acquired immunodeficiency syndromeOphthalmology, 96 (1989), pp. 1092-1099ArticleDownload PDFView Record in ScopusGoogle Scholar3B.J Glasgow, A.K WeisbergerA quantitative and cartographic study of retinal microvasculopathy in acquired immunodeficiency syndromeAm J Ophthalmol, 118 (1994), pp. 46-56ArticleDownload PDFView Record in ScopusGoogle Scholar4D.A Newsome, W.R Green, E.D Miller, et al.Microvascular aspects of acquired immune deficiency syndrome retinopathyAm J Ophthalmol, 98 (1984), pp. 590-601ArticleDownload PDFView Record in ScopusGoogle Scholar5J.I Quiceno, E Capparelli, A.A Sadun, et al.Visual dysfunction without retinitis in patients with acquired immunodeficiency syndromeAm J Ophthalmol, 113 (1992), pp. 8-13ArticleDownload PDFView Record in ScopusGoogle Scholar6S.A Geier, G Hammel, J.R Bogner, et al.HIV related ocular microangiopathic syndrome and color contrast sensitivityInvest Ophthalmol Vis Sci, 35 (1994), pp. 3011-3021View Record in ScopusGoogle Scholar7L.C Kozak, M.A BullimoreVisual changes in human immuno-deficiency virus infectionOptom Vis Sci, 71 (1994), pp. 557-561CrossRefView Record in ScopusGoogle Scholar8A Roth, M Pelizzone, D Hermès, J SommerhalderNeuere Ueberlegungen und Entwicklungen zur klinischen Untersuchung des Farbensehens-die zwei-GleichungsmethodeFortschr Ophthalmol, 86 (1989), pp. 374-379View Record in ScopusGoogle Scholar9Roth A, Pelizzone M, Hermès D, Sommerhalder J. L'examen de la vision colorée par la méthode des deux équations métamériques. Ophtalmologie 1990;4:197–205.Google Scholar10Roth A. The power of metameric colour equations in testing colour vision. In: Ohta Y, editor. Colour Vision Deficiencies. Amsterdam: Kugler and Ghedini, 1990:181–190.Google Scholar11Pokorny J, Smith VC. Metameric matches for assessment of colour vision. I. Theoretical considerations. In: Verriest G, editor. Colour Vision Deficiencies VII, Documenta Ophthalmologica Proc Ser 39. The Hague: Dr W. Junk, 1984:83–94.Google Scholar12Pokorny J, Smith VC. Colour matching as a clinical tool. Theory of modification by disease. In: Ohta Y, editor. Colour Vision Deficiencies. Amsterdam: Kugler and Ghedini, 1990:255–267.Google Scholar13Moreland, JD. The clinical utility of anomaloscopy. In: Ohta Y, editor. Colour Vision Deficiencies. Amsterdam: Kugler and Ghedini, 1990:127–143.Google Scholar14S.A Geier, M Held, J.R Bogner, et al.Impairment of tritan vision after initiation of treatment with zidovudine in patients with HIV disease or AIDSBr J Ophthalmol, 77 (1993), pp. 315-316CrossRefView Record in ScopusGoogle Scholar15Pelizzone M, Sommerhalder J, Roth A, Hermès D. Automated Rayleigh and Moreland matches on a computer controlled anomaloscope. In: Drum B, Moreland JD, Serra A, editors. Colour Vision Deficiencies X, Documenta Ophthalmologica Proc Ser 54. Dordrecht: Kluwer, 1991:151–159.Google Scholar16R De Natale, J Flammer, M Zulauf, T BebieInfluence of age on the transparency of the lens in normals: a population study with help of the Lens Opacity Meter 701Ophthlamologica, 197 (1988), pp. 14-18CrossRefView Record in ScopusGoogle Scholar17Rossillion B, Pelizzone M, Sommerhalder J, Roth A. Automated Moreland equations on 7° and 2° fields. In: Drum B, editor. Colour Vision Deficiencies XII, Documenta Ophthalmologica Proc Ser 57. Dordrecht: Kluwer, 1995:481–488.Google Scholar18Pelizzone M, Sommerhalder J, Roth A, Hermès D. Automated Rayleigh and Moreland matches: optimization of stimulation parameters for normal observers. In: Drum B, editor. Colour Vision Deficiencies XI, Documenta Ophthalmologica Proc Ser 56. Dordrecht: Kluwer, 1993:345–355.Google Scholar19M Pelizzone, J Sommerhalder, A RothDétermination des seuils de discrimination dans les épreuves d'égalisation coloréeOphtalmologie, 9 (1995), pp. 229-234Google Scholar20Geigy Scientific Tables, vol. 2. Basel: Ciba Geigy, 1982:63.Google Scholar21Moreland JD. Matching range and age in a blue green equation. In: Drum B, editor. Colour Vision Deficiencies XI, Documenta Ophthalmologica Proc Ser 56. Dordrecht: Kluwer, 1993:129–134.Google Scholar22Roth A, Pelizzone M, Sommerhalder J, Hermès D, Simona F. The two equation method: III. Results in normal subjects above 50 years of age. Correlation with lens opacity. In: Drum B, Moreland JD, Serra A, editors. Colour Vison Deficiencies X, Documenta Ophthalmologica Proc Ser 54. Dordrecht: Kluwer, 1991:353–359.Google Scholar23Roth A, Sommerhalder J, Hermès D, Pelizzone M. Vision colorée et vieillissement maculaire. Ophtalmologie 1998;12:5–8.Google Scholar24A.J Mueller, D.J Plummer, R Dua, et al.Analysis of visual dysfunction in HIV-positive patients without retinitisAm J Ophthalmol, 124 (1997), pp. 158-167ArticleDownload PDFView Record in ScopusGoogle Scholar25E Zrenner, J NowickiMedikamentös induzierte Funktionsstörungen der Zapfenfunktion und ZapfeninteraktionFortschr Ophthalmol, 82 (1985), pp. 589-594View Record in ScopusGoogle ScholarCopyright © 1998 Elsevier Science Ltd. All rights reserved.