Graefe's Archive Ophthalmology for Clinical and Expedmenlal
Clinical investigations
© Springer-Verlag 1990 Graefe's Arch Clin Exp Ophthalmol (1990) 228:201-217
Differentiation of ischemic from non-ischemic central retinal vein occlusion during the early acute phase* Sohan Singh Hayreh ~, Marie R. Klugman 2, Meena Beri 1, Alan E. Kimura 1, and Patricia Podhajsky 1 1 Department of Ophthalmology and 2 Department of Preventive Medicine (Biostatistics), University of Iowa, Iowa City, IA 52242, USA Abstract. We investigated prospectively in 128 patients (140 eyes) the role of six routine clinical tests in the differentiation of ischemic central retinal vein occlusion (CRVO) from non-ischemic CRVO during its early acute phase. There were four functional tests [visual acuity, visual fields, relative afferent pupillary defect (RAPD), electroretinography (ERG)] and two morphologic tests (ophthalmoscopy and fluorescein fundus angiography). We found that none of the six tests had 100% sensitivity and specificity in such a differentiation during the early, acute phase, so that no one test can be considered a "gold standard"; however, combined information from all six is almost always reliable. Overall, the four functional tests proved far superior to the two morphologic tests in differentiating ischemic from non-ischemic CRVO: RAPD was most reliable in uniocular CRVO (with a normal fellow eye), followed closely by E R G in all cases; combined information from RAPD and E R G differentiated 97% of cases; perimetry was the next most reliable, followed by visual acuity. The two morphologic tests performed worst; fluorescein angiography provided either no information at all on retinal capillary nonperfusion (in at least one-third of the eyes during the early, acute phase) because of multiple limitations, or sometimes provided misleading information. Ophthalmoscopic appearance is the least reliable, most misleading parameter.
Hayreh's clinical studies [10 16, 18, 20, 37]. While the concept was initially dismissed by pundits, other authors have since verified Hayreh's findings [1-7, 17, 2(~29, 32-36, 38, 39]. Although the concept of two types of CRVO is now widely accepted, the criteria for their diagnosis and the differentiation between the two types still remains highly controversial. Based on our long-term, prospective, longitudinal, clinical studies on CRVO, we have previously described some criteria for differentiating ischemic from non-ischemic CRVO [16]. Through time and experience, other, more reliable criteria have emerged. To reach definite conclusions about the subject, we started a prospective study five years ago, the objective of which was to define multiple criteria and to determine their relative sensitivity and specificity for the differentiation of ischemic from non-ischemic CRVO during the early, acute stages of the disease. The diagnosis becomes much more certain after a follow-up of about a year or longer; however, at this point it may be too late, since the risk of developing anterior segment ocular neovascularization (NV) in ischemic CRVO is greatest during the first 7 months or so and minimal after that [20] (Fig. 1).
IRIS NV
'0.70 log units), we found that both R A P D and E R G provided the correct information in differentiation of isehemic from non-ischemic CRVO in 73%-76% of the eyes; R A P D alone supplied the information (with E R G within normal limits) in an additional 16%-19%; E R G alone (with R A P D within normal limits), in 5 % - 8 % more.
ERG This is also an objective functional test, very useful in the differentiation of ischemic from non-ischemic CRVO. Our study has shown that, of the various parameters investigated by us [21], the E R G parameter with the best sensitivity (80%-90%) and specificity (70%-80%) in such a differentiation was the amplitude of the b-wave (both photopic and scotopic), when the b-wave amplitude was: (a) reduced to __0.90 log unit
88% 80%
90% 97%
ERG (photopic and scotopie)
b-wave amplitude < 60% of normal or reduced by >_1 SD
80-90%
71-80%
RAPD and/or ERG
> 0.70 log unit RAPD and/or < 60% b-wave amplitude
9%100%
71%
Visual acuity
6/120 or worse Worse than 6/120
91-100% 80-100%
78-88% 87-93%
Visual fields
cannot see I2e target of Goldmann perimeter Any visual field defect present with V4e target Central scotoma and peripheral defect with I4e target Central scotoma with V4e
9zV100%
67-78%
71-82%
83-88%
92%
87%
71-81%
86-90%
81-84%
72-74%
"
a
Ophthalmoscopy > Moderate hemorrhages in posterior retina
It is evident from the above discussion that functional tests are better. It could be argued that, to classify C R V O into ischemic and non-ischemic types, our original use of information from five of the six tests under scrutiny must bias the results in favor of the initial criteria, thereby invalidating our conclusions. The final criteria which emerged (Table 10), however, indicate that this did not happen.
Fluorescein fundus angiography
Validity o f final criteria in various tests
data analysis showed that the best sensitivity and specificity were obtained with > 0.7 log units (Table 10). Moreover, we never used any one of these tests as the sole criterion to define ischemic CRVO. Thus, no test had a disproportionate influence on the final results. Also, the development o f ocular NV and follow-up information on the course of the disease - the two parameters supplying the most important information in defining ischemic C R V O are independent of and unrelated to the six tests. Our data analyses o f the various parameters in the six tests, using only ocular NV as the endpoint in the division of C R V O into two types (without any input from any of the six tests), further confirmed the validity of these test parameters for differentiating C R V O into ischemic and non-ischemic types.
As discussed in "Materials and methods", we used the following initial criteria in the five tests to define ischemic C R V O (p. 202): for visual acuity, count fingers at only 1-2 ft or worse; for visual fields, ability to see only the V-4e target or worse; for R A P D , > 1.2 log units; for fluorescein fundus angiography, marked diffuse retinal capillary obliteration; and for ophthalmoscopy, extensive cottonwool spots. ( E R G was not used as one of the defining parameters.) We combined the information from the five tests with two totally independent criteria, i.e., development of ocular N V and the course of the disease during follow-up of up to 5 years. A n eye was defined as having ischemic C R V O by the presence of no less than four of these parameters, coupled with indisputable evidence of retinal isehemia during the acute phase or on follow-up. The final criteria which showed the best sensitivity and specificity for each parameter in distinguishing ischemic from non-ischemic C R V O (Table 10) were not as extreme as the defining criteria. For example, for R A P D , we used > 1.2 log units as our defining criterion for ischemic CRVO, but the final
Retinal capillary obliteration
Not possible to give valid information for the entire sample because no such information was available in over one-third of the eyes; in eyes with available information, its sensitivity and specificity were 92-100% and 88-100%, respectively a
-
Should ocular N V be used as an end-point? Critics could argue that, to prevent results biased in favor of the initial defining criteria for ischemic and non-ischemic CRVO, we should have defined them only by the presence
216 or absence of ocular NV, and that ocular N V should be used as an endpoint. It could also be maintained that one should try to define the ability of the six tests to predict this endpoint. The numerous objections to such reasoning include the following: 1. Although ocular N V is usually regarded as the most serious complication of ischemic CRVO, it does not occur in all eyes with ischemic C R V O ; we know that there is definitely a group o f eyes with extensive retinal capillary obliteration and other evidence of considerable retinal ischemia which never develop any type o f ocular NV. The incidence of various types o f N V in ischemic C R V O is shown in Fig. 1. 2. Our experience has shown that ischemic C R V O eyes, even if they do not develop NV, still usually have a poorer prognosis for visual outcome than non-ischemic CRVO. 3. When we compared the results o f the test parameters in ischemic C R V O with and without ocular NV, we found that there were no statistically significant ( P _ 0.05) differences between the various test parameters in the two types o f ischemic CRVO. 4. To divide C R V O into two types simply on the basis of presence or absence of ocular N V would not differentiate ischemic from non-ischemic CRVO, since about one-third of the eyes with ischemic CRVO, but without any NV, would be grouped with non-ischemie C R V O ; this would vitiate the results and give misleading information about the prognosis and management of the two types of CRVO. 5. The specificities of the various test parameters, when ocular N V is used as the differentiating feature for division of C R V O into two types (Table 9), are markedly reduced as compared to those when C R V O is divided into ischemic and non-ischemic types. This is not at all surprising because, obviously, the eyes categorized as having " C R V O without ocular N V " in fact contain not only non-ischemic C R V O eyes but also about one-third o f the ischemic C R V O eyes without ocular NV. The logical sequence in evaluation o f C R V O should be, first, to differentiate ischemic from non-ischemic CRVO, and second, to try to see if one can predict which of the ischemic C R V O eyes are at risk of developing ocular NV. We investigated this but found that none of parameters evaluated in this study seem to be able to predict definitely the development o f various types of ocular N V in ischemic CRVO. F r o m this discussion, it is evident that to use ocular N V as the sole endpoint in C R V O would be totally unwarranted in any study dealing with evaluation o f prognosis and management of C R V O in ischemic and non-ischemic CRVO.
Terminology The use of a correct and universal name for a disease is absolutely essential for proper communication. The information provided by the various tests used in the present study helps to clarify the current profusion - and confusion - of eponyms used by different authors for the two types of CRVO. The terms "partial", "incomplete", "imminent", "threatened", "incipient", or "impending" C R V O describing non-ischemic C R V O are invalid [16] because our experimental [8, 9, 17, 22] and clinical [10-16, 18, 20, 37] studies have demonstrated that the C R V O is usually complete in these cases. For non-ischemic C R V O the terms "mild C R V O " [25], "hyperpermeability-response-macular-
edema-type" [29] or "hyperpermeable type" [32] are incorrect because macular edema is secondary to hyperpermeability of retinal capillaries in C R V O ; it is not only absent in a number of non-ischemic CRVOs but is also universal and significantly (P
Clinical investigations
© Springer-Verlag 1990 Graefe's Arch Clin Exp Ophthalmol (1990) 228:201-217
Differentiation of ischemic from non-ischemic central retinal vein occlusion during the early acute phase* Sohan Singh Hayreh ~, Marie R. Klugman 2, Meena Beri 1, Alan E. Kimura 1, and Patricia Podhajsky 1 1 Department of Ophthalmology and 2 Department of Preventive Medicine (Biostatistics), University of Iowa, Iowa City, IA 52242, USA Abstract. We investigated prospectively in 128 patients (140 eyes) the role of six routine clinical tests in the differentiation of ischemic central retinal vein occlusion (CRVO) from non-ischemic CRVO during its early acute phase. There were four functional tests [visual acuity, visual fields, relative afferent pupillary defect (RAPD), electroretinography (ERG)] and two morphologic tests (ophthalmoscopy and fluorescein fundus angiography). We found that none of the six tests had 100% sensitivity and specificity in such a differentiation during the early, acute phase, so that no one test can be considered a "gold standard"; however, combined information from all six is almost always reliable. Overall, the four functional tests proved far superior to the two morphologic tests in differentiating ischemic from non-ischemic CRVO: RAPD was most reliable in uniocular CRVO (with a normal fellow eye), followed closely by E R G in all cases; combined information from RAPD and E R G differentiated 97% of cases; perimetry was the next most reliable, followed by visual acuity. The two morphologic tests performed worst; fluorescein angiography provided either no information at all on retinal capillary nonperfusion (in at least one-third of the eyes during the early, acute phase) because of multiple limitations, or sometimes provided misleading information. Ophthalmoscopic appearance is the least reliable, most misleading parameter.
Hayreh's clinical studies [10 16, 18, 20, 37]. While the concept was initially dismissed by pundits, other authors have since verified Hayreh's findings [1-7, 17, 2(~29, 32-36, 38, 39]. Although the concept of two types of CRVO is now widely accepted, the criteria for their diagnosis and the differentiation between the two types still remains highly controversial. Based on our long-term, prospective, longitudinal, clinical studies on CRVO, we have previously described some criteria for differentiating ischemic from non-ischemic CRVO [16]. Through time and experience, other, more reliable criteria have emerged. To reach definite conclusions about the subject, we started a prospective study five years ago, the objective of which was to define multiple criteria and to determine their relative sensitivity and specificity for the differentiation of ischemic from non-ischemic CRVO during the early, acute stages of the disease. The diagnosis becomes much more certain after a follow-up of about a year or longer; however, at this point it may be too late, since the risk of developing anterior segment ocular neovascularization (NV) in ischemic CRVO is greatest during the first 7 months or so and minimal after that [20] (Fig. 1).
IRIS NV
'0.70 log units), we found that both R A P D and E R G provided the correct information in differentiation of isehemic from non-ischemic CRVO in 73%-76% of the eyes; R A P D alone supplied the information (with E R G within normal limits) in an additional 16%-19%; E R G alone (with R A P D within normal limits), in 5 % - 8 % more.
ERG This is also an objective functional test, very useful in the differentiation of ischemic from non-ischemic CRVO. Our study has shown that, of the various parameters investigated by us [21], the E R G parameter with the best sensitivity (80%-90%) and specificity (70%-80%) in such a differentiation was the amplitude of the b-wave (both photopic and scotopic), when the b-wave amplitude was: (a) reduced to __0.90 log unit
88% 80%
90% 97%
ERG (photopic and scotopie)
b-wave amplitude < 60% of normal or reduced by >_1 SD
80-90%
71-80%
RAPD and/or ERG
> 0.70 log unit RAPD and/or < 60% b-wave amplitude
9%100%
71%
Visual acuity
6/120 or worse Worse than 6/120
91-100% 80-100%
78-88% 87-93%
Visual fields
cannot see I2e target of Goldmann perimeter Any visual field defect present with V4e target Central scotoma and peripheral defect with I4e target Central scotoma with V4e
9zV100%
67-78%
71-82%
83-88%
92%
87%
71-81%
86-90%
81-84%
72-74%
"
a
Ophthalmoscopy > Moderate hemorrhages in posterior retina
It is evident from the above discussion that functional tests are better. It could be argued that, to classify C R V O into ischemic and non-ischemic types, our original use of information from five of the six tests under scrutiny must bias the results in favor of the initial criteria, thereby invalidating our conclusions. The final criteria which emerged (Table 10), however, indicate that this did not happen.
Fluorescein fundus angiography
Validity o f final criteria in various tests
data analysis showed that the best sensitivity and specificity were obtained with > 0.7 log units (Table 10). Moreover, we never used any one of these tests as the sole criterion to define ischemic CRVO. Thus, no test had a disproportionate influence on the final results. Also, the development o f ocular NV and follow-up information on the course of the disease - the two parameters supplying the most important information in defining ischemic C R V O are independent of and unrelated to the six tests. Our data analyses o f the various parameters in the six tests, using only ocular NV as the endpoint in the division of C R V O into two types (without any input from any of the six tests), further confirmed the validity of these test parameters for differentiating C R V O into ischemic and non-ischemic types.
As discussed in "Materials and methods", we used the following initial criteria in the five tests to define ischemic C R V O (p. 202): for visual acuity, count fingers at only 1-2 ft or worse; for visual fields, ability to see only the V-4e target or worse; for R A P D , > 1.2 log units; for fluorescein fundus angiography, marked diffuse retinal capillary obliteration; and for ophthalmoscopy, extensive cottonwool spots. ( E R G was not used as one of the defining parameters.) We combined the information from the five tests with two totally independent criteria, i.e., development of ocular N V and the course of the disease during follow-up of up to 5 years. A n eye was defined as having ischemic C R V O by the presence of no less than four of these parameters, coupled with indisputable evidence of retinal isehemia during the acute phase or on follow-up. The final criteria which showed the best sensitivity and specificity for each parameter in distinguishing ischemic from non-ischemic C R V O (Table 10) were not as extreme as the defining criteria. For example, for R A P D , we used > 1.2 log units as our defining criterion for ischemic CRVO, but the final
Retinal capillary obliteration
Not possible to give valid information for the entire sample because no such information was available in over one-third of the eyes; in eyes with available information, its sensitivity and specificity were 92-100% and 88-100%, respectively a
-
Should ocular N V be used as an end-point? Critics could argue that, to prevent results biased in favor of the initial defining criteria for ischemic and non-ischemic CRVO, we should have defined them only by the presence
216 or absence of ocular NV, and that ocular N V should be used as an endpoint. It could also be maintained that one should try to define the ability of the six tests to predict this endpoint. The numerous objections to such reasoning include the following: 1. Although ocular N V is usually regarded as the most serious complication of ischemic CRVO, it does not occur in all eyes with ischemic C R V O ; we know that there is definitely a group o f eyes with extensive retinal capillary obliteration and other evidence of considerable retinal ischemia which never develop any type o f ocular NV. The incidence of various types o f N V in ischemic C R V O is shown in Fig. 1. 2. Our experience has shown that ischemic C R V O eyes, even if they do not develop NV, still usually have a poorer prognosis for visual outcome than non-ischemic CRVO. 3. When we compared the results o f the test parameters in ischemic C R V O with and without ocular NV, we found that there were no statistically significant ( P _ 0.05) differences between the various test parameters in the two types o f ischemic CRVO. 4. To divide C R V O into two types simply on the basis of presence or absence of ocular N V would not differentiate ischemic from non-ischemic CRVO, since about one-third of the eyes with ischemic CRVO, but without any NV, would be grouped with non-ischemie C R V O ; this would vitiate the results and give misleading information about the prognosis and management of the two types of CRVO. 5. The specificities of the various test parameters, when ocular N V is used as the differentiating feature for division of C R V O into two types (Table 9), are markedly reduced as compared to those when C R V O is divided into ischemic and non-ischemic types. This is not at all surprising because, obviously, the eyes categorized as having " C R V O without ocular N V " in fact contain not only non-ischemic C R V O eyes but also about one-third o f the ischemic C R V O eyes without ocular NV. The logical sequence in evaluation o f C R V O should be, first, to differentiate ischemic from non-ischemic CRVO, and second, to try to see if one can predict which of the ischemic C R V O eyes are at risk of developing ocular NV. We investigated this but found that none of parameters evaluated in this study seem to be able to predict definitely the development o f various types of ocular N V in ischemic CRVO. F r o m this discussion, it is evident that to use ocular N V as the sole endpoint in C R V O would be totally unwarranted in any study dealing with evaluation o f prognosis and management of C R V O in ischemic and non-ischemic CRVO.
Terminology The use of a correct and universal name for a disease is absolutely essential for proper communication. The information provided by the various tests used in the present study helps to clarify the current profusion - and confusion - of eponyms used by different authors for the two types of CRVO. The terms "partial", "incomplete", "imminent", "threatened", "incipient", or "impending" C R V O describing non-ischemic C R V O are invalid [16] because our experimental [8, 9, 17, 22] and clinical [10-16, 18, 20, 37] studies have demonstrated that the C R V O is usually complete in these cases. For non-ischemic C R V O the terms "mild C R V O " [25], "hyperpermeability-response-macular-
edema-type" [29] or "hyperpermeable type" [32] are incorrect because macular edema is secondary to hyperpermeability of retinal capillaries in C R V O ; it is not only absent in a number of non-ischemic CRVOs but is also universal and significantly (P
