Scattergram of 24 deteriorating eyes in which the rate of progression was significant with linear regression and is plotted against the progression rate obtained from the first to the last visual field (VF). Spearman rank order correlation, r=0.86, P<.001.
Scattergram of 40 deteriorating eyes with a nonsignificant linear regression. Abscissa shows the rate of visual field (VF) progression with linear regression; ordinate, the rate of VF loss as calculated from the first to the last VF (r=0.82, P<.001).
Rate of visual field (VF) loss of superior (ordinate) and inferior (abscissa) hemifields of 10 eyes of 8 patients with normal-pressure glaucoma progressing in both hemifields (rs=0.67, P=.04).
Rate of visual field (VF) loss of superior (ordinate) and inferior (abscissa) hemifields of 12 eyes of 11 patients with primary open-angle glaucoma progressing in both hemifields (rs=−0.01, P>.99).
Scattergram showing initial visual field defect (abscissa) and rate of deterioration of hemifields (ordinate) of 87 progressive hemifields in patients with normal-pressure glaucoma, primary open-angle glaucoma, and ocular hypertension (hemifields with a rate of deterioration of 0.5% or more per year only).
Rasker MTE, van den Enden A, Bakker D, Hoyng PFJ. Rate of Visual Field Loss in Progressive Glaucoma. Arch Ophthalmol. 2000;118(4):481-488. doi:10.1001/archopht.118.4.481
To investigate the rate of visual field (VF) loss in progressive glaucoma.
Outpatient department, nonreferral base.
A cohort of 34 patients with normal-pressure glaucoma (NPG), 68 patients with primary open-angle glaucoma (POAG), and 125 patients with ocular hypertension (OHT) were followed up for an average of 9 years. Visual fields were obtained annually with automated perimetry. The rate of VF loss as a percentage per year was calculated.
Twenty-three eyes with NPG, 31 with POAG, and 10 with OHT showed progression of VF loss. The mean (±SD) rates of VF deterioration were 3.7% ± 3.3% per year in NPG, 2.5% ± 1.8% in POAG, and 2.3% ± 1.3% in OHT converting to POAG, and did not differ significantly. No difference in the rate of VF loss was found between eyes with and without optic disc hemorrhages (2.7% ± 2.9% and 3.1% ± 2.1%, respectively). The rate of VF loss was not related to the initial VF status. The rate of VF loss between the superior and inferior hemifields was correlated in patients with NPG (rs=0.67, P=.04). Comparison of visual field loss with linear regression analysis showed significant slopes in only 37.5% of eyes with progression, which had a progression rate of 4.2% ± 3.0%.
The rate of VF loss did not differ between patients with NPG and POAG. The rate of deterioration was related neither to initial VF status nor to the presence of disc hemorrhages. Linear regression is applicable only in a portion of the patients who have progression of VF loss.
GLAUCOMA IS a widespread chronic optic neuropathy characterized by excavation of the optic nerve head and typical visual field (VF) defects; in approximately 60% to 70%1- 4 of cases it is accompanied by elevated intraocular pressure (IOP). Despite presumed effective IOP-lowering therapy and after a considerable period of IOP at estimated target, deterioration of VFs is known to occur.5 In previous work6 it was observed that patients with normal-pressure glaucoma (NPG) and without disc hemorrhages (DHs) showed deterioration in 3.6% of cases per year. This rate was not different from the proportion of 3.9% of cases per year observed in patients with primary open-angle glaucoma (POAG) without DHs in that study.
Not only the proportion of patients with progression, but the rate of the glaucomatous process, ie, the rate of decay of VFs during long-term follow-up, is of interest. An investigation of the rate of deterioration may answer the question of whether this rate is similar in patients with NPG and POAG. Contradictory findings have been reported7- 9 on the influence of initial VF status on the rate of progression. Therefore, a possible relationship between the defect volume at the moment the diagnosis was established and the rate of VF loss with time is of interest. Furthermore, it was observed that the presence of DHs increased the hazard rate for VF progression with factors of 5 and 4 in patients with NPG and POAG, respectively.6 The question may arise whether VFs of eyes with DHs deteriorate faster than those of eyes without DHs.
To assess the rate of VF loss, quantitative VF analysis must be performed. Various methods for quantification of VF defects have been reported.10- 14 A few studies analyzed the rate of progression by means of VF loss found at different ages15,16 retrospectively, while others used linear regression analysis of consecutive VF tests.7- 9,17
The purpose of this study was to investigate the rate of VF loss per year in patients with NPG, POAG, and ocular hypertension (OHT) who were prospectively followed up during a mean period of 9 years. The influence of the defect volume at the start of the study and the presence of DHs on the rate of VF loss were also assessed. Finally, the rate of VF loss as calculated from the first to the last field was compared with linear regression of all VF data.
At the outpatient clinic of the St Lucas Hospital in Amsterdam, the Netherlands, 227 patients with glaucoma were enrolled in a long-term study.4,6 The cohort consisted of 34 patients with NPG, 68 patients with POAG, and 125 patients with OHT. All patients had newly detected disease and were followed up for at least 3 years. Informed consent was obtained from all patients participating in the study.
At the start of the study, visual acuity, refraction, anterior segment evaluation, and gonioscopy were performed, as were funduscopy and binocular examination of the optic disc. The IOP was assessed with an applanation tonometer (Goldmann; Haag-Streit AG, Liebefeld, Switzerland). At the start of the study, in all patients, 4 IOP measurements during daytime were assessed in the absence of medication. Visual field examination was performed with computerized perimetry (Peritest; Rodenstock, Munich, Germany). Furthermore, color stereoslides of the optic nerve heads were made.
Every 3 months the patients revisited the outpatient clinic for visual acuity, anterior segment biomicroscopy, IOP measurement, and optic disc examination. In addition, every year the VFs were screened, diurnal IOP curves were performed, and stereoslides of the optic nerve head were made.
Patients were regarded as having NPG or POAG if they had an arcuate scotoma within the central 30° or a nasal step at at least 2 examinations, a glaucomatous optic disc, an open angle, and 4 measurements of IOP during daytime without medication and during the entire study of 22 mm Hg or lower for patients with NPG and exceeding 22 mm Hg without medication for patients with POAG. The optic nerve head was considered abnormal if the vertical cup-disc ratio was 0.7 or greater, if notching of the neural retinal rim area was present, or if bayoneting of retinal vessels entering the optic disc or baring of the circumlinear vessels occurred. Subjects with OHT had an average of 4 IOP measurements during daytime that exceeded 22 mm Hg, or 1 IOP measurement greater than or equal to 25 mm Hg without medication, had open angles, and were without VF defects. Patients with OHT who developed VF defects during the study were referred to as having OHT and were not included in the POAG group.
Patients with NPG received therapy if the IOP exceeded 18 mm Hg, the VF worsened, or glaucomatous change of the optic nerve head occurred. Patients with POAG were treated when diagnosed, and patients with OHT received therapy when their IOP exceeded 26 mm Hg, if they showed an increasing cup-disc ratio, or when their condition converted to POAG. The presence of a DH was not an indication for starting or changing therapy.
The perimeter used for VF testing screens 206 targets, of which 151 are located within the central 25° of the VF. The VF tests were suprathreshold gradient adapted.18,19 A VF defect was considered significant when a sensitivity reduction of at least 10 dB at a cluster of 3 or more locations occurred as an arcuate scotoma within the central 30° area or as a nasal step. This may seem a crude measure, but it should be noted that the Peritest screens more than twice the number of targets obtained with the Humphrey (Carl Zeiss, Jena, Germany) or Octopus (Interzeag, Zurich, Switzerland) perimeter test. The defects found had to be reproducible. A patient's condition was considered to be progressive when a reproducible change of at least 10 dB at 3 or more locations in a cluster or contiguous to an existing defect could be observed. If a VF indicated deterioration, it was repeated after a delay of approximately 4 months. The VFs were evaluated and rated as stable or progressing by 2 independent observers (D.B. and P.F.J.H.) in a masked fashion. Visual field changes not induced by glaucoma were screened for by evaluating changes in visual acuity and the effect of progression of cataracts 2 weeks before the VF determination. Defects of the VF induced by spectacles, eyelid, and a general loss of sensitivity owing to cataract were corrected for. The evening before the VF examination, patients receiving miotics interrupted this therapy.
Within the VFs of the eyes with progression, the following areas were discerned: in the superior hemifield, the superior central area (within 25° eccentricity) and the superior peripheral area (outside 25° eccentricity), with 70 and 26 targets, respectively; and in the inferior hemifield, the inferior central area (within 25° eccentricity) and the inferior peripheral area (outside 25° eccentricity), with 68 and 29 targets, respectively.
As maximum sensitivity loss per stimulus point was reached at a loss of 24 dB, the maximum sensitivity losses per area were 1680 dB (superior central area), 624 dB (superior peripheral area), 1632 dB (inferior central area), and 696 dB (inferior peripheral area), respectively.12 The blind spot (13 stimulus points) was not taken into account. The VF defects were quantified as a percentage of the maximum sensitivity loss of that area. The formula used for each area was as follows:
where Da indicates mean VF defect of an area as a percentage of the maximum defect of that area; x, sensitivity loss (in decibels) of a stimulus point i; and n, number of stimulus points in an area.
The differences in density and total number of stimulus points of the separate areas are equated by this formula. From the Da values of the 4 separate areas, the mean defect as a percentage of the superior hemifield, the inferior hemifield, and the total VF can be calculated. The rate of VF loss as a percentage per year was obtained by dividing the difference between the first and the last mean VF defect (percentage) by the years of follow-up. This was done for the total fields and for the superior and inferior hemifields.
To determine the effect of intertest fluctuations and a possible learning effect, the rate of VF deterioration per year was obtained not only from the first to the last but also from the second to the last VF. In addition, on all VFs, linear regression analysis was performed and the results obtained were compared with our method of evaluation of the fields with the Wilcoxon matched pairs test and Spearman rank order correlation.
Target pressure for patients with POAG was 20 mm Hg or less with a reduction in IOP of at least 20%. For patients with NPG, if treated, target pressure was 16 mm Hg, with at least a 20% reduction in IOP. For patients with OHT, if treated, target pressure was 22 mm Hg or less with a reduction in IOP of at least 20%.
Correlations between the rate of VF loss of superior and inferior hemifields were analyzed with the Wilcoxon matched pairs test and the Spearman rank order correlation. Further statistical analysis was performed with the unpaired t test and the Mann-Whitney test. Unless indicated otherwise, data are expressed as mean ± SD.
Of 227 patients included in this study, 18 patients with NPG (23 eyes), 27 with POAG (31 eyes), and 8 with OHT (10 eyes) showed deterioration of VF during follow-up. Approximately all patients had newly detected disease and were not taking glaucoma medication at the start of the study. The mean ages, mean initial VF defects in decibels (approximates mean deviation) and as a percentage of maximum defect (−1 dB is comparable with 4% loss of total VF), and the mean follow-up period of patients with and without VF deterioration are listed in Table 1. The mean age of patients with OHT without progression (62.2 ± 10.7 years) was significantly lower (P<.04) than that of patients with OHT with progressing VF defects (69.5 ± 9.1 years) and that of patients with NPG and POAG, whether they had VF progression or not. The follow-up period of approximately 9 years did not differ between the patient groups. There was no difference in mean initial VF defect between patients with NPG and POAG, being 28.9% ± 26.4% and 23.6% ± 29.2%, respectively. The period between consecutive VF tests was 1.2 ± 0.2 years.
No difference was observed in the percentage of mean defect between the superior and inferior hemifields of eyes with progression in patients with NPG (34.0% ± 35.3% and 23.9% ± 24.7%, respectively), POAG (24.7% ± 31.8% and 22.6% ± 30.8%, respectively), and OHT (1.6% ± 2.9% and 0.5% ± 1.8%, respectively).
The period from the start of the study until the occurrence of reproducible VF deterioration was 4.8 ± 2.3 years in patients with NPG, 3.6 ± 2.4 years in patients with POAG, and 7.1 ± 3.4 years in patients with OHT.
Table 2 gives the mean rates of VF loss per year for the whole VF; the rates were 3.7% ± 3.3% in 23 eyes of 18 patients with NPG, 2.5% ± 1.8% in 31 eyes of 27 patients with POAG, and 2.3% ± 1.3% in 10 eyes of 8 patients with OHT. The rates were not different between the glaucoma groups but were significantly different from the rates for the stable eyes. Table 3 indicates that the rate of VF loss between superior and inferior hemifields did not differ within and between the glaucoma groups. The mean rate of VF loss of all eyes with progression was 2.93% per year, which approximates −0.73 dB per year.
The rate of VF loss may have been influenced by a possible learning effect or intertest fluctuation. Therefore, the rate of deterioration was also calculated from the second to the last VF. The rates of VF loss obtained from the second to the last whole VF were 3.3% ± 2.7%, 2.9% ± 3.2%, and 2.8% ± 3.2% per year, respectively, for patients with NPG, POAG, and OHT and did not differ from the rates as calculated from the first to the last VFs.
Linear regression analysis was performed on all VFs. The regression was considered significant if P<.05. The results with regression analysis were compared with the results of calculated VF loss from the first to the last field of each eye. These rates did not differ within the patient groups with and without a significant regression coefficient (Table 4) (Wilcoxon matched pairs test and Spearman rank order correlation). However, the pooled data of eyes with NPG, POAG, and OHT that showed progression and had significant linear regression showed a progression rate of 4.2% ± 3.0% VF loss per year vs 3.4% ± 2.5% per year as calculated from the first to the last VF, the difference being significant (P<.03). The pooled data of eyes with progression that had a nonsignificant linear regression showed deterioration, with 2.8% ± 2.5% as calculated from the first to the last VF and 2.4% ± 2.6% with linear regression, and did not differ.
Table 5 gives the progression rate (percentage per year) of nonprogressing eyes in patient groups with either significant or nonsignificant regression coeffi cients. Of interest is the progression rate of 0.65% per year of 12 eyes with OHT with a significant regression coefficient, which may indicate that these eyes are likely to convert to POAG.
The progression rate in 24 (38%) of the 64 eyes with progression in which the rate of progression was significant with linear regression is plotted against the progression rate obtained from the first to the last VF in individual eyes in Figure 1. Analysis with Spearman rank order correlation show a high correlation between the 2 rates (r=0.86, P<.001). Also, in 40 eyes with progression (62%) without a significant linear regression, the 2 rates were highly correlated (r=0.82 P<.001) (Figure 2).
In 40 eyes without progression, the regression coefficient was not significant. This was because of a Schub-type course of the glaucomatous process (12 eyes), too small a number of VFs to perform linear regression analysis (16 eyes), and too much intertest fluctuation (12 eyes) (Table 6).
Table 7 shows that, among the patients with NPG, deterioration in 7 eyes was located only in superior hemifields, in 6 eyes in inferior hemifields, and in 10 eyes (8 patients) (43%) in both hemifields. In all cases in which both hemifields deteriorated, this occurred at a similar point in time. In the 11 of 31 eyes with POAG that had progression, deterioration occurred in superior hemifields only and in 8 eyes in inferior hemifields only. Twelve eyes (39%) of 11 patients had simultaneous deterioration in the superior and inferior hemifields. In patients with OHT, 3 of 10 eyes deteriorated only in superior hemifields, 1 eye in the inferior hemifield only, and 6 eyes (60%) (in 4 patients) had deterioration in both hemifields, of which 4 eyes showed concurrent deterioration of the superior and inferior hemifields. The other 2 eyes had a latency period between deterioration of the superior and inferior hemifields of 3 and 9 years.
Ten eyes with NPG that had progression showed deterioration in both hemifields. The rate of deterioration was correlated between superior and inferior hemifields (Spearman rank order correlation, rs=0.67; P=.04; Figure 3). In eyes with deterioration in both hemifields in patients with POAG, the rate of VF loss did not correlate between superior and inferior hemifields (rs=−0.01, P>.99; Figure 4). The proportion of eyes deteriorating in single or in both hemifields was equally divided in eyes with and without DHs of patients with NPG as well as those with POAG.
We investigated whether the rate of VF loss was related to the initial VF status. Figure 5 shows the rate of VF loss in 46 superior and 41 inferior hemifields deteriorating at a rate of more than 0.5% per year. In these hemifields, no correlation was observed between VF loss at baseline and the rate of loss per year (Spearman rank order correlation, rs=0.03). Furthermore, Figure 5 shows the rate of VF loss compared between eyes with smaller and larger initial defects in hemifields. The median initial defect of hemifields was 5.4%. Forty-three hemifields had smaller and 44 had larger defects than 5.4%. The rate of VF loss was 3.8% ± 2.6% per year in eyes with progressive hemifields with initial VF defects smaller than 5.4% and 3.7% ± 3.1% per year in eyes with a median defect larger than 5.4%, and these rates did not differ.
In 27 eyes with progressing VF loss of patients with NPG, POAG, or OHT with DHs, deterioration occurred at a rate of 2.7% ± 2.9% per year, which was not different from the rate of VF loss of eyes of patients without DHs who had no progression, which was 3.1% ± 2.1% per year. Superior and inferior hemifields deteriorated at a similar rate in patients with and without DHs (Table 8).
During the study, in addition to medical treatment, laser trabeculoplasty was performed in 6 eyes of patients with NPG, 32 eyes of patients with POAG, and 19 eyes of patients with OHT. Argon laser trabeculoplasty was performed in patients with OHT when, despite maximal conservative treatment, the IOP exceeded 26 mm Hg in the absence of a pathologic excavated optic nerve head, or exceeded 22 mm Hg in the presence of a suspect excavated optic nerve head (vertical cup-disc ratio, >0.7), or in the case of OHT converting to POAG. Trabeculectomy was performed in 6 eyes of patients with NPG, 31 eyes of patients with POAG, and 6 eyes of patients with OHT converting to POAG (3 eyes) or IOPs greater than 30 mm Hg despite maximum therapy and a vertical cup-disc ratio greater than 0.8. It should be noted that, throughout the study, patients with OHT converting to POAG were included in the OHT group.
In this prospective study, the rate of VF loss per year during a mean period of 9 years was evaluated in patients with NPG, POAG, and OHT. The rate of VF loss was similar for deteriorating eyes of all 3 patient groups, being approximately 3% per year, or a sensitivity loss of −0.7 dB of mean deviation per year. Hence, patients with NPG experience VF deterioration at a rate similar to that of patients with POAG.
Several reports have dealt with the rate of VF loss.7- 10,15- 17 However, a comparison of results may be hampered by differences in perimeters, methods of quantifying VFs, follow-up periods, and the fact that some studies are cross-sectional. The method used for obtaining a single index for the VF defect as a percentage resembles the mean deviation of Flammer et al11 and Greve and Bakker's defect volume.12 Although, during follow-up, episodes of progression may be followed by stabilization,8 we calculated the rate of VF loss as a percentage per year from the first and last VFs to have an impression of the progression rate during the whole period the patient was followed up. It should be taken into consideration that the rate of deterioration given in this study may be underestimated, since in eyes with progression the glaucomatous process may have been slowed down or stabilized by additional medical treatment or surgical procedures.
In a cross-sectional study, Jay and Murdoch15 estimated the interval between early field loss and end-stage glaucoma in patients with optimum treatment to be 38 years. In our prospective long-term follow-up study, only patients with glaucoma whose VF was deteriorating had an average rate of VF loss of about 30% of total VF per decade. This indicates that eyes with early glaucoma and under glaucoma treatment that show progression may reach end-stage glaucoma in about 33 years.
It should be noted that, in our study, only approximately one third of patients with NPG and POAG had progression.6
The estimation of progression was based on clusters of deteriorating points, the deterioration being confirmed by an additional VF examination approximately 4 months later. Visual field loss was expressed as a percentage of total VF loss. Our mean progression rate of 3% per year (−0.7 dB/year mean deviation) in eyes with progression in all 3 glaucoma groups together is in agreement with those reported in other studies,7,9,20 ranging between −0.96 and −1.39 dB of loss per year of mean deviation. These data were obtained from long-term prospective studies and assessed with linear regression.
Processing of all our VF data with linear regression revealed that only 37.5% of all eyes with progression had significant progression. It has been stated that linear regression is a simple method to estimate trends in longitudinal data such as progressive slopes of mean deviations of VFs.7 The use of linear regression is dependent on intertest variability, number of VFs available, and of the decay of the glaucomatous process.7 The assumption has to be made that the glaucomatous VF loss is gradual and linear. However, in several studies, the progression in glaucoma has been reported to be episodic.21,22 In our study, in 62.5% of the eyes with progression, VF loss was not significant with linear regression. This was due partly to the episodic nature of the glaucomatous process, partly to the large intertest fluctuation, and partly to too small a number of VF tests (<7).
These observations question the use of linear regression as the ultimate test to analyze whether significant VF loss exists. Moreover, it is debatable whether data from psychophysical tests such as VF examinations in the pathologic glaucomatous process fit in a statistical analysis with linear regression. On the other hand, Spearman rank order correlation tests between the progression rates as calculated by our method and as obtained by linear regression analysis show a high correlation, regardless of whether the regression slopes were significant. From this we conclude that regression analysis of the global VF index mean deviation is a useful statistical tool, but it should not be decisive regardless of whether a patient with glaucoma has progression. In addition, linear regression analysis of data from nonprogressive OHT showed 12 eyes with a significant linear regression coefficient and a progression rate of 0.7% ± 0.5% per year. Especially in this group of patients with long-term follow-up, linear regression may be useful to detect early trends indicating imminent VF loss.
It was observed earlier8 that the rate at which individual patients experience deterioration is variable. Also, in this study, a large range of velocities of VF loss within all 3 glaucoma groups was observed. The question arises whether this might be caused by differences in initial VF status. The large SDs of the mean initial VF defects reflect the diversity of glaucomatous damage of patients visiting an ophthalmologist for the first time. Mikelberg and associates8 stated that a greater rate of further VF loss occurred when more advanced VF loss existed at the time the diagnosis was established. In contrast, O'Brien and associates9 found a greater rate of VF loss at earlier stages of the disease. In our study, we observed that the rate of VF loss was not related to the initial VF defect, which is in accordance with the observation of Smith and associates.7 Therefore, in our opinion, it is not possible to predict from the initial VF status which patients will have deterioration in time6,7,20 or the rate at which deterioration will occur in individual patients.
Since DHs are known to increase the hazard rate for VF deterioration in glaucoma,6 it would make sense to presume that VF losses in patients with DHs might progress at a faster rate. However, the rates of VF loss per year were 2.7% ± 2.9% and 3.1% ± 2.1% in patients with and without DHs, respectively, and were not different. This supports earlier observations6,23- 26 that DHs are not likely to be the cause of VF deterioration but rather are a sign of the ongoing glaucomatous process.
Heijl and Bengtsson27 reported a substantial influence of learning in patients who underwent repeated automated VF examinations, while this was not observed by Werner et al28 and Smith et al.7 Learning effects are not likely to have influenced the results of our study, as there was no difference between the defect changes per year calculated from the first to the last and from the second to the last VFs. This might be because the 1-year intervals between subsequent VF tests are too long to induce a significant learning effect, in contrast to the 1-week intervals used by Heijl and Bengtsson.27
In earlier studies,21,22,29 it was reported that one or both hemifields are at risk for deterioration during a certain period. We observed that 56% of eyes with progression in patients with NPG and 61% of eyes with progression in patients with POAG had VF loss in one hemifield only during the entire follow-up period. In eyes with deterioration in both hemifields, a different pattern of deterioration rate was present between patients with NPG and POAG. Although in both groups deterioration of superior and inferior hemifields occurred concurrently, there was no correlation of rate of VF loss between both hemifields in patients with POAG. In contrast, in eyes of patients with NPG in which both hemifields had progression, the velocities of VF loss in the hemifields were correlated (rs=0.67; P=.04). In conclusion, despite individual differences, the eyes with NPG that show progression do not deteriorate faster than those with POAG or OHT converting to POAG. Although eyes with DHs are more at risk for progression than eyes without DHs,6 the rate of the decay of VF also did not differ between eyes with and without DHs. No correlation was observed between initial VF status and rate of VF deterioration. These results indicate a final common pathway in glaucoma whether patients suffer from POAG, NPG, or OHT and whether they do or do not have DHs.
Accepted for publication December 10, 1999.
This study was supported by grant 28-2326 from the Praeventiefonds, The Hague, the Netherlands.
Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Fort Lauderdale, Fla, May 9, 1999 (ARVO abstract 368).
Reprints: Philip F. J. Hoyng, MD, PhD, The Netherlands Ophthalmic Research Institute, PO Box 12141, 1100 AC Amsterdam ZO, the Netherlands (e-mail: firstname.lastname@example.org).