Methods

This was a prospective study of 52 consecutive patients with tuberculosis (104 eyes) treated with anti-tubercular therapy, which included ethambutol given at a dose of 15–20 mg/kg per day for 2 months. The other drugs in the treatment regimen included isoniazid, rifampicin and pyrazinamide. Patients were recruited from the Directly Observed Treatment Strategy (DOTS) Centre (Dr R P Centre for Opthalmic Sciences, New Delhi, India) and outpatient department of our hospital.

The patients had to take the drugs from a single source, the DOTS centre, which provided free anti-tubercular drugs under the revised National Tuberculosis Control Programme of the Government of India. The patients took the medicines in front of the DOTS centre personnel, which ensured 100% compliance with the therapy.

All the patients were examined at our centre. The exclusion criterion were:

• Presence of any other disease causing optic neuropathy.

• Intake of any other drug known to cause optic neuropathy except the first-line anti-tubercular drugs.

• Ocular/central nervous system tuberculosis.

• Any ocular disease affecting the parameters that were being evaluated.

• Pre-existing colour vision defects.

Patient evaluation

A detailed clinical history was taken, including the history of present illness, occupation and history of smoking. A complete general physical and systemic examination was carried out. A thorough ophthalmic examination was done, including cycloplegic refraction, slit-lamp examination for anterior segment and pupillary reactions. Specific ophthalmic tests were done with the appropriate spectacle correction. Visual acuity was assessed on ETDRS charts. Colour vision was noted using Ishihara pseudoisochromatic plates and with the Oculus Heidelberger anomaloscope (Oculus, Wetzlar, Germany), as it specifically picks up red-green defects, which are known to be more common with ethambutol toxicity.11 Contrast sensitivity was measured using the Pelli–Robson chart.

Fundus examination was done with the direct and indirect ophthalmoscopes to rule out ocular tuberculosis and other ocular pathology. Intraocular pressure was measured using Goldmann’s applanation tonometer. The Amsler’s grid chart was used to look for any central scotoma and the Goldmann kinetic perimeter was used to evaluate the peripheral visual fields. Stereoacuity was measured using a standard test booklet (The Netherlands Organization for Applied Scientific Research (TNO)).

Pattern-reversal VER were noted on a Nicolet Bravo EP with 1015 visual stimulator and monitor (Nicolet Biomedical, Madison, Wisconsin, USA). Monocular, whole-field stimulation with a checkerboard pattern (reversal time of 500 ms) was used. All the patients were tested with the same machine, from a distance of 1 m, in standard ambient conditions.

Retinal nerve fibre layer (RNFL) thickness was assessed using optical coherence tomography (Zeiss Optical Coherence Tomographer, Model 3000 OCT-3; Carl Zeiss Meditec, Dublin, California, USA). Pupils of both eyes were dilated with tropicamide 1% eye drops. When the pupil was at least 5 mm in diameter, OCT scans were taken using the “Fast RNFL thickness” protocol of the machine. Peripapillary RNFL thickness was measured in each of the four quadrants. The quadratic and average RNFL thickness was noted for both eyes separately.

The first examination was done as a baseline just before starting the drug. Thereafter, examinations were done each month, concluding 1 month after stopping the ethambutol therapy. OCT examination was done before the start of therapy and at 1 month after stopping the drug.

Statistical analysis was performed using the Student’s t test. A p value of <0.05 was considered significant.

Results

There were 29 men and 23 women with ages ranging from 11 to 56 years (mean age 28.1 years). Pulmonary tuberculosis was present in 39 (75%) of patients; the remainder had extra-pulmonary disease, including seven (13%) patients with uterine tuberculosis, five (10%) with lymph node tuberculosis and one (2%) with skeletal tuberculosis. None had renal tuberculosis, known to be a risk factor for optic neuropathy.

No patient complained of diminution or blurring of vision or any other ocular problem at any time during the study. No change was seen in visual acuity of any of the patients on ETDRS charts.

Colour vision and contrast sensitivity were normal in all the patients. No change in the fundus or the intraocular pressure was observed at any point of time during the study. Amsler’s grid charting was normal at all the visits. Assessment of stereoacuity with the TNO test did not show any significant change in any of the patients.

Visual field defects were found in 7.69% (eight out of 104) of eyes of four patients (all were bilaterally affected) in the study at 2 months of therapy. These were in the form of peripheral isopter contraction. No abnormality in central fields was detected. Four eyes in three patients still demonstrated persistent visual field defects even after 1 month of stopping the therapy.

Pattern-VER showed increase in the mean latency of the P₁₀₀ wave component at 1 and 2 months of therapy (p<0.001 for both) (table 1). There was marginal improvement in the mean latency at 1 month after stoppage of the drug, but this was still significantly increased compared with the baseline mean (p = 0.002). No significant change was found in the mean or individual patient’s amplitude of the VER curve at any of the follow-up examinations.

VER latency in each individual eye at 1- and 2-month follow-up visits and at 1 month after stopping the drug was compared with the baseline latency of that eye. A greater than 10 ms increase in latency was considered a significant increase in that patient. Ten ms corresponded to 2 SD from the mean in baseline VER latency (108.75 (SD 5.02) ms). At the 2 month visit, ten eyes showed an increase in VER latency of 10–15 ms, while five eyes showed an increase in latency of >15 ms. Of these 15 eyes of 11 patients, four patients had increased latency in both of their eyes, and seven patients had increased latency in only one eye. In the unilaterally affected cases, the other eye also showed an increased latency in six out of seven cases, but this increase was not significant (ie <10 ms).

VER examination 1 month after stopping the drug showed that out of the 15 eyes that had increased latency on previous examination, 12 had recovered the latency to within 10 ms of their baseline values. Of the remaining three eyes, one of them still had significantly increased latency (delay of 14 ms) and two eyes had very significant increase (ie >15 ms).

The mean of average and quadratic RNFL thickness at 1 month after stopping ethambutol were compared with the baseline mean (table 2). No significant change in average thickness was seen. RNFL thickness in the temporal quadrant was significantly lower at 1 month follow-up compared with baseline (p = 0.011), while no significant change was observed in the superior, inferior and nasal quadrants.

OCT values of RNFL thickness in each eye were compared with the baseline value of that eye; a >20 μm decrease in the thickness was taken as significant. A thickness of 20 μm corresponded to a 2 SD difference from the mean of average thickness at baseline. Three eyes (2.88% of 104) of two patients showed significant decrease (>20 μm decrease in OCT values) in temporal quadrant RNFL thickness, while the same was not observed in the other quadrants or in the average thickness in any of the eyes. Visual function in all these three eyes was also affected at 1 month after stopping ethambutol, with all three eyes demonstrating increase in latency on VER (>10 ms increase) and residual visual field defects.

In our study, no patient complained of any clinical symptoms, which is an incidence of 0% to <2%. Subclinical toxicity was found in 19.23% (20/104) of the total eyes. This toxicity has been detected in the form of increased latency of pattern-VER, peripheral defects on visual field examination and decreased temporal RNFL thickness.

Reversal of the observed subclinical defects was seen in 80% of eyes after 1 month of stoppage of the drug.

Discussion

The incidence of toxicity is variably reported in literature and it appears to be <1% at the presently used dose of 15–20 mg/kg per day. In our study none of the patients developed clinical symptoms, which is an incidence of 0 to <2%. Subclinical toxicity was seen in 19.23% of the total eyes, which was in the form of delayed latency of pattern-VER, peripheral defects on visual field examination and temporal RNFL thickness loss.

There are no clear risk factors for irreversible visual damage due to the drug, but old age, renal insufficiency and chronic smoking are said to increase the risk of toxicity. None of these risk factors were found in the patients with the observed subclinical defects.

All the patients recruited obtained the drug from a single source, the DOTS Centre, which provides free anti-tubercular drugs under the revised National Tuberculosis Control Programme of the Government of India. This ensured that the drug given to all the patients was of the same potency, thereby avoiding the manufacturer-related bias. As per the DOTS requirement, all the patients had to take the medicines in front of the DOTS Centre personnel, which ensured 100% compliance with the therapy.

Contrast sensitivity as measured on Pelli–Robson chart was not affected in any of the patients, as has been demonstrated earlier.3 Arden plates are affected by the ambient lighting conditions, are observer-dependent and are also known to show a high false-positive rate.12 The Pelli–Robson chart on other hand is relatively unaffected by the ambient lighting conditions, and also has high test–retest reliability.13

The incidence of visual field defects is highly variable among the various studies and these were found to be central, peripheral or both. In general, visual field defects tend to appear with the use of higher dosage of the drug especially in cases with obvious visual deficit.8 10 Visual field defects in our study were in the form of peripheral isopter contraction and point defects. No central field defects was detected.

Though VER is an objective tool for assessing optic nerve function, there are very few reports of its use for early detection of ethambutol-induced optic neuropathy with inconsistent results. The most important finding in these studies was the increased latency of the pattern-VER curve, although with extremely variable incidence (0% to 42.8%).1 14 15 The results of our study have shown an increase in latency in 14.4% of total eyes with 2 months of therapy with reversal of the VER findings in 80% of eyes after 1 month of stoppage of the drug.

In general, visual functional defects with ethambutol are considered to be reversible in most of the cases as seen in our study. Recovery may occur as late as 1 year after discontinuation of ethambutol, although numerous reports of permanent visual damage have been reported.2 6 8 9 Although the exact mechanism of this ocular neurotoxic effect is largely unknown, animal studies have shown ethambutol toxicity in the retinal ganglion neurons of rodents.16 The peripapillary RNFL thickness was assessed using OCT, which is a direct measure the peripapillary RNFL thickness.17 Although the average thickness of the four quadrants has not changed significantly, a significant decrease in the mean peripapillary RNFL thickness in the temporal quadrant was noted while no change was observed in any of the other quadrants. On individual comparison of RNFL thickness of each eye, 2.88% (three out of 104) of total eyes have been found to have a significant RNFL thickness loss in the temporal quadrant.

Recent studies on RNFL thickness on OCT in diagnosed cases of ethambutol-induced optic neuropathy have observed significant RNFL thickness loss in almost all the quadrants with maximum involvement of the temporal quadrant.18 19 Moreover, the amount of RNFL thickness-loss correlated with the severity of clinically measured visual function deficit.18 In our study, although only subclinical involvement was observed, significant RNFL changes were seen only in the temporal quadrant. The anatomical changes on OCT coincided with visual functional defect and delayed conduction on VER, suggesting that these findings were not just by chance. Our study suggests that the macular fibres may be most sensitive to toxic damage and may be the only long-term visible sign of a toxic insult. This appears to be in agreement with the previous studies that have also suggested that ethambutol has predilection for smaller papillomacular bundle, similar to other mitochondrial optic neuropathies.3 Although none of the cases demonstrated classical central visual field defects, it could be due to the early detection of these cases. Intake of the drug over a longer time could result in greater damage to the papillomacular bundle and present as visual field defects.

From our study we recommend that apart from visual acuity, colour vision, visual fields and contrast sensitivity, VER and OCT should be added as important tools in detecting early ethambutol toxicity. This is particularly important when dosages >15–20 mg/kg per day are used or when ethambutol is used for periods longer than 2 months.