Treatment of Sac Fistula
Congenital
Lacrimal
To the Editor.\p=m-\TheAugust 1978 issue of the Archives contains an article entitled "Definitive Treatment of Congenital Lacrimal Sac Fistula" (96:1443-1444,1978). I have been interested in this procedure and published on it in the American Journal of Ophthalmology in 1939 (22:1023,
1939).
Joseph Laval, MD New York
Acquired Brown's Syndrome Inflammatory Origin
of
To the Editor.\p=m-\Ivery much appreciated the article titled "Acquired Brown's Syndrome of Inflammatory Origin" by John S. Hermann, MD (Archives 96:1228-1232, 1978). I have observed one case of acquired Brown's sheath syndrome, presumably of inflammatory origin. This patient was a 30-year-old woman with mild rheumatoid arthritis who experienced two separate episodes of acquired Brown's sheath syndrome occurring nearly a year apart and promptly responded to a peritrochlear injection of 0.5 mL of betamethasone (Celestone). The most recent
episode required two injections
week apart to provide complete resolution of all her symptoms, which included pain and tenderness in the trochlear area as well as the classic limitation in motility. I hope your readers will find this to be of interest in what seems to be verification of the entity of tenosynovitis in an acquired Brown's tendon sheath syndrome. Donald H. Beisner, MD one
Springfield,
Mo
Reply.\p=m-\Iwas delighted to read Dr Beisner's letter in reference to his treatment of a case of acquired Brown's syndrome with steroids injected into the trochlear area. From his description and the symptoms, his case would certainly fit into the category of inflammatory origin. Thus far, his is the only other reported case of this kind, to my knowledge, that In
has been treated with locally injected steroids. I suspect that there must be many other unreported cases of this entity, although I have spoken to many ophthalmologists who specialize in the field of strabismus, and none of these physicians had seen a similar case. I cannot understand why this entity was so common in the British Isles based on the articles of SanfordSmith1-3 and why it should be still rare here in the United States. I think it is a great service to print letters such as Dr Beisner's to alert ophthalmologists to syndromes that are rarely reported but possibly seen and puzzled over on many occasions. John S. Hermann, MD New York 1. Sanford-Smith JH: Superior oblique tendon sheath syndromes. Br J Ophthalmol 59:385-386, 1975. 2. Sanford-Smith JH: Intermittent superior oblique tendon sheath syndrome. Br J Ophthalmol 53:412-417, 1969. 3. Sanford-Smith JH: Superior oblique tendon sheath syndrome and its relationship to stenosing tenosynovitis. Br J Ophthalmol 57:859-863, 1973.
Retinal Blood Flow To the Editor.\p=m-\Ina recent article in the Archives titled "Studies on Retinal Blood Flow: I. Estimation of Human Retinal Blood Flow by Slit-
lamp Fluorophotometry" (96:893-897, 1978), Cunha-Vaz and Lima describe a technique to estimate retinal blood flow that they have applied to studies in diabetic patients.1 We believe, based on our experience with the dyedilution technique and on a critical review of this article, that this technique is not adequate to determine retinal blood flow for the following reasons:
1. In the technique of Cunha-Vaz and Lima, a quantity that is designated as the "first appearance time" of fluorescein is measured at two sites 0.9 mm apart along the superior temporal artery and is used to calculate the minimal transit time (tm) of the dye between these two points. The mean value for tm is 0.17 seconds in
Downloaded From: http://archopht.jamanetwork.com/ by a University of Michigan User on 06/18/2015
normal subjects and 0.09 seconds in anemic patients. It is well known that the first appearance time of a dye at the level of the retinal arterioles cannot be measured accurately after an intrave¬ nous injection'4—certainly not with the accuracy needed to determine tm val¬ ues in the range of 0.1 to 0.2 seconds, as reported by the authors. For such tm values to be meaningful, let us say 20% error, one would have to measure each first appearance time to an accu¬ racy of about 0.01 seconds. Signalvs-noise analysis shows that this is clearly beyond the limits of the dyedilution technique. In fact, the actual first appearance is always obscured by the background noise of the photoelec¬ tric signal, which is present even before the dye appears. The uncer¬ tainty (AtJ in the determination of ta where ta is the first appearance time is a function of the noise level (N) and of the slope (S) of the fluorescence intensity curve at the time where this intensity exceeds the noise level. It is is of the not difficult to see that order N/S. The slower the fluores¬ cence intensity rises above the noise level (small value of S), the more diffi¬ cult it is to determine accurately the first appearance time. We calculated Ata for the curves shown in Fig 2 of Cunha-Vaz and Lima's article, taking for S the aver¬ age slope during the first second of the dye passage based on the time scale of 0.1 s/division. We found that for both dye appearance curves Ata is about 0.4 seconds. This leads to an uncertainty in tm between 0.5 and 0.6 seconds. This poor accuracy in the technique of Cunha- Vaz and Lima is due not only to the slow rise in fluores¬ cence as a result of intravenous injec¬ tion, but also to the fact that the authors report that the signal-to-noise ratio in their technique is only 3. Such inherent inaccuracies would have been detected by the authors had they done repeated injections and measure¬ ments of t,„ in the same subject. 2. The authors' observation that "no significant artifacts were intro¬ duced in the recordings due to the pulsatile nature of the blood flow in
the retinal arteries" (p 896) further confirms that the fluorescence inten¬ sity emitted from the first arriving dye molecules in the central lamina was well within the noise level of the recording system and therefore was not recordable by their system. If the authors had detected this fluorescence intensity, they would have found that pulsatility critically affects the time difference tm. Variations in tn, as large as factor 3 can be expected since the maximum velocity of dye varies by as much as this factor between diastole and systole.4 In order to detect this fluorescence, however, the dye must be injected directly into the carotid artery.- Therefore, the authors' calcu¬ lations, which are based on the assumption that the fluorescein in the central lamina is detected first, are not valid. These considerations are not merely theoretical. We have verified them by applying the same technique as the authors, using a fundus camera instead of a slit lamp. We performed measurements in anesthetized owl monkeys to obtain ideal conditions of eye stability. In order to optimize the detection of the first appearance time, certain factors were varied, such as the amount of fluorescein injected intravenously, the distance between the fibers, and the noise level and response time of the electronics. The signal-to-noise ratio (defined as the height of a fluorescein curve one second after the first appearance time divided by the noise level N) was well above 20. We found, using multiple injections, that each first appearance time could not be determined with an accuracy better than ±0.2 seconds, resulting in an uncertainty in t,„ of about ± 0.3 seconds. This level of accu¬ racy was not sufficient to measure a meaningful difference in the appear¬ ance of fluorescein between two sites (1.5 mm apart) along a retinal artery. Similar conclusions were reached by van Heuven et al,5 although they injected the dye directly into the left ventricle to obtain a sharper front edge of the dye bolus. 3. A problem not discussed by the authors but that we found important to consider is the influence of choroi¬ dal fluorescence, which appears before the dye is detected in the retinal arteries. This choroidal fluorescence may easily be detected first, as one cannot guarantee that the fibers are positioned directly over the retinal artery as the dye appears, especially when the subject fixates a target with the fellow eye owing to nonconstant heterophoria. The authors say that "as 4
the recordings are done in a very short period of time, it is not difficult for the patients to maintain a steady fixa¬ tion" (p 895). This is not correct because the patients have to maintain steady fixation from the time of the injection to a time between 10 and 15 seconds after the injection in order to
make
sure
that the fibers
are
well
the artery as the dye arrives. Even under optimal condi¬ tions of fixation by the eye under
positioned
on
study, the SD of eye position is ± 15 µ a fixation period of 10 seconds or more." This means that an optical
for
fiber of the size of an artery will be off the artery by more than 15 µ for about 30% of the measuring time. The use of a low-vacuum contact lens does not decrease eye movements. The authors say that "fine adjustments may easily be made at any time" (p 895). In our experience with a camera and with a slit lamp, it is not possible to adjust the position of two fibers simultaneously to achieve the kind of accuracy needed to ensure that no one fiber collects fluorescence from the dye in the choroid. 4. Analysis of the data reported by Cunha-Vaz and Lima shows that in normal subjects (their Table 1), the average maximum speed of fluores¬ cein is 0.9 mm/0.17 s 5 mm/s. This is six to eight times lower than values found by other investigators in pigs, monkeys, and humans, using high¬ speed fluorescein cineangiography.4 :,-? This lower value for the dye maximum speed cannot be attributed to the reported elevation of the intraocular pressure (IOP) to a level of 30 mm Hg. An increase of 10 to 15 mm Hg above normal IOP would reduce the dye maximum speed by only about 20%," ie, to value around 30 mm/s. Even though the dye speeds re¬ ported in Table 1 are much slower than those found by others, the flow values obtained by Cunha-Vaz and Lima are in accordance with other published data.4 This is so because the value given for the diameter of the superior temporal artery in their normal subjects is about twice the value normally measured. Typically, in normal subjects the average diame¬ ter of large arteries adjacent to the optic disc is 100 µ," not 184 µ as given by the authors (Table 1). The only explanation we can offer for such a discrepancy is that the authors did not account for the magnification factor of their fundus angiography system, which is most probably around 2. Fail¬ ure to consider this point would result in overestimation of the flow values by a factor of about 4. In any case, =
Downloaded From: http://archopht.jamanetwork.com/ by a University of Michigan User on 06/18/2015
these flow values would still be far below the range of those published by the same authors.1" In conclusion, we have indicated various reasons that preclude the application of the technique described by Cunha-Vaz and Lima to the deter¬ mination of retinal blood flow. We believe that the blood flow values in normal and anemic subjects are not correct. The conclusions of these investigators regarding changes of blood flow in various stages of diabetes,1 even if correct, cannot be supported by their technique of mea¬ surement.
Charles E. Riva, DSc Boston Gilbert T. Feke, PhD, George T. Timberlake, PhD, Steven Sinclair, MD, and Michael Loebl, MD, provided advice and assistance and Flavia Blackwell provided editorial assistance. 1. Cunha-Vaz et al: Studies
JG, Fonseca JR, de Abreu JFR,
retinal blood flow: II. Diabetic Arch Ophthalmol 96:809-811, 1978. CT: Dynamic aspects of the retinal microcirculation. Arch Ophthalmol 79:536-539, 1968. 3. Bulpitt CJ, Kohner EM, Dollery CT: Velocity profiles in the retinal microcirculation. Bibl Anat 11:448-452, 1973. 4. Eberli B, Feke GT, Rogers F, et al: Continuous recording of the speed of red blood cells in human retinal vesels. Presented at the Association for Research in Vision and Ophthalmology. Sarasota, Fla, May 5, 1978. 5. Van Heuven WAJ, Malik AB, Schaffer CA, et al: Retinal blood flow derived from dye dilution curves. Televised fluorescein angiography. Arch Ophthalmol 95:297-301, 1977. 6. Ditchburn RW: Eye-movements and Visual Perception. London, Clarendon Press, 1973. 7. Wise GL, Dollery CT, Henkind P: The Retinal Circulation. New York, Harper & Row Publishers Inc, 1971, pp 83-118. 8. Ffytche TJ, Bulpitt CJ, Kohner EM, et al: Effect of changes in intraocular pressure on the retinal microcirculation. Br J Ophthalmol 58:514\x=req-\ 522, 1974. 9. Hogan MJ, Alvarado JA, Esperson Weddell J: Histology of the Human Eye: An Atlas and Textbook. Philadelphia, WB Saunders Co, 1971. 10. Cunha-Vaz JG, de Lima JJP: Estimation of human retinal blood flow by slitlamp fluorophotometry: II. Results and discussion. IRCS Med Sci 3:577, 1975. on
retinopathy. 2. Dollery
Reply.\p=m-\Itshould be obvious to Dr colleagues, as it is to everyone else, that a technique that permits absolute measurements of In
Riva and his
retinal blood flow in not yet available. That has never been our claim because we realize that our technique has most of the drawbacks and limitations of the two-point photometric systems, such as the one used by Riva and associates. With these facts in mind, I will try to clarify points raised in their letter. 1. We never stated in our report that we are measuring the first fluorescein molecules passing through the artery, nor was the phrase "first appearance time of fluorescein" ever
Congenital
Lacrimal
To the Editor.\p=m-\TheAugust 1978 issue of the Archives contains an article entitled "Definitive Treatment of Congenital Lacrimal Sac Fistula" (96:1443-1444,1978). I have been interested in this procedure and published on it in the American Journal of Ophthalmology in 1939 (22:1023,
1939).
Joseph Laval, MD New York
Acquired Brown's Syndrome Inflammatory Origin
of
To the Editor.\p=m-\Ivery much appreciated the article titled "Acquired Brown's Syndrome of Inflammatory Origin" by John S. Hermann, MD (Archives 96:1228-1232, 1978). I have observed one case of acquired Brown's sheath syndrome, presumably of inflammatory origin. This patient was a 30-year-old woman with mild rheumatoid arthritis who experienced two separate episodes of acquired Brown's sheath syndrome occurring nearly a year apart and promptly responded to a peritrochlear injection of 0.5 mL of betamethasone (Celestone). The most recent
episode required two injections
week apart to provide complete resolution of all her symptoms, which included pain and tenderness in the trochlear area as well as the classic limitation in motility. I hope your readers will find this to be of interest in what seems to be verification of the entity of tenosynovitis in an acquired Brown's tendon sheath syndrome. Donald H. Beisner, MD one
Springfield,
Mo
Reply.\p=m-\Iwas delighted to read Dr Beisner's letter in reference to his treatment of a case of acquired Brown's syndrome with steroids injected into the trochlear area. From his description and the symptoms, his case would certainly fit into the category of inflammatory origin. Thus far, his is the only other reported case of this kind, to my knowledge, that In
has been treated with locally injected steroids. I suspect that there must be many other unreported cases of this entity, although I have spoken to many ophthalmologists who specialize in the field of strabismus, and none of these physicians had seen a similar case. I cannot understand why this entity was so common in the British Isles based on the articles of SanfordSmith1-3 and why it should be still rare here in the United States. I think it is a great service to print letters such as Dr Beisner's to alert ophthalmologists to syndromes that are rarely reported but possibly seen and puzzled over on many occasions. John S. Hermann, MD New York 1. Sanford-Smith JH: Superior oblique tendon sheath syndromes. Br J Ophthalmol 59:385-386, 1975. 2. Sanford-Smith JH: Intermittent superior oblique tendon sheath syndrome. Br J Ophthalmol 53:412-417, 1969. 3. Sanford-Smith JH: Superior oblique tendon sheath syndrome and its relationship to stenosing tenosynovitis. Br J Ophthalmol 57:859-863, 1973.
Retinal Blood Flow To the Editor.\p=m-\Ina recent article in the Archives titled "Studies on Retinal Blood Flow: I. Estimation of Human Retinal Blood Flow by Slit-
lamp Fluorophotometry" (96:893-897, 1978), Cunha-Vaz and Lima describe a technique to estimate retinal blood flow that they have applied to studies in diabetic patients.1 We believe, based on our experience with the dyedilution technique and on a critical review of this article, that this technique is not adequate to determine retinal blood flow for the following reasons:
1. In the technique of Cunha-Vaz and Lima, a quantity that is designated as the "first appearance time" of fluorescein is measured at two sites 0.9 mm apart along the superior temporal artery and is used to calculate the minimal transit time (tm) of the dye between these two points. The mean value for tm is 0.17 seconds in
Downloaded From: http://archopht.jamanetwork.com/ by a University of Michigan User on 06/18/2015
normal subjects and 0.09 seconds in anemic patients. It is well known that the first appearance time of a dye at the level of the retinal arterioles cannot be measured accurately after an intrave¬ nous injection'4—certainly not with the accuracy needed to determine tm val¬ ues in the range of 0.1 to 0.2 seconds, as reported by the authors. For such tm values to be meaningful, let us say 20% error, one would have to measure each first appearance time to an accu¬ racy of about 0.01 seconds. Signalvs-noise analysis shows that this is clearly beyond the limits of the dyedilution technique. In fact, the actual first appearance is always obscured by the background noise of the photoelec¬ tric signal, which is present even before the dye appears. The uncer¬ tainty (AtJ in the determination of ta where ta is the first appearance time is a function of the noise level (N) and of the slope (S) of the fluorescence intensity curve at the time where this intensity exceeds the noise level. It is is of the not difficult to see that order N/S. The slower the fluores¬ cence intensity rises above the noise level (small value of S), the more diffi¬ cult it is to determine accurately the first appearance time. We calculated Ata for the curves shown in Fig 2 of Cunha-Vaz and Lima's article, taking for S the aver¬ age slope during the first second of the dye passage based on the time scale of 0.1 s/division. We found that for both dye appearance curves Ata is about 0.4 seconds. This leads to an uncertainty in tm between 0.5 and 0.6 seconds. This poor accuracy in the technique of Cunha- Vaz and Lima is due not only to the slow rise in fluores¬ cence as a result of intravenous injec¬ tion, but also to the fact that the authors report that the signal-to-noise ratio in their technique is only 3. Such inherent inaccuracies would have been detected by the authors had they done repeated injections and measure¬ ments of t,„ in the same subject. 2. The authors' observation that "no significant artifacts were intro¬ duced in the recordings due to the pulsatile nature of the blood flow in
the retinal arteries" (p 896) further confirms that the fluorescence inten¬ sity emitted from the first arriving dye molecules in the central lamina was well within the noise level of the recording system and therefore was not recordable by their system. If the authors had detected this fluorescence intensity, they would have found that pulsatility critically affects the time difference tm. Variations in tn, as large as factor 3 can be expected since the maximum velocity of dye varies by as much as this factor between diastole and systole.4 In order to detect this fluorescence, however, the dye must be injected directly into the carotid artery.- Therefore, the authors' calcu¬ lations, which are based on the assumption that the fluorescein in the central lamina is detected first, are not valid. These considerations are not merely theoretical. We have verified them by applying the same technique as the authors, using a fundus camera instead of a slit lamp. We performed measurements in anesthetized owl monkeys to obtain ideal conditions of eye stability. In order to optimize the detection of the first appearance time, certain factors were varied, such as the amount of fluorescein injected intravenously, the distance between the fibers, and the noise level and response time of the electronics. The signal-to-noise ratio (defined as the height of a fluorescein curve one second after the first appearance time divided by the noise level N) was well above 20. We found, using multiple injections, that each first appearance time could not be determined with an accuracy better than ±0.2 seconds, resulting in an uncertainty in t,„ of about ± 0.3 seconds. This level of accu¬ racy was not sufficient to measure a meaningful difference in the appear¬ ance of fluorescein between two sites (1.5 mm apart) along a retinal artery. Similar conclusions were reached by van Heuven et al,5 although they injected the dye directly into the left ventricle to obtain a sharper front edge of the dye bolus. 3. A problem not discussed by the authors but that we found important to consider is the influence of choroi¬ dal fluorescence, which appears before the dye is detected in the retinal arteries. This choroidal fluorescence may easily be detected first, as one cannot guarantee that the fibers are positioned directly over the retinal artery as the dye appears, especially when the subject fixates a target with the fellow eye owing to nonconstant heterophoria. The authors say that "as 4
the recordings are done in a very short period of time, it is not difficult for the patients to maintain a steady fixa¬ tion" (p 895). This is not correct because the patients have to maintain steady fixation from the time of the injection to a time between 10 and 15 seconds after the injection in order to
make
sure
that the fibers
are
well
the artery as the dye arrives. Even under optimal condi¬ tions of fixation by the eye under
positioned
on
study, the SD of eye position is ± 15 µ a fixation period of 10 seconds or more." This means that an optical
for
fiber of the size of an artery will be off the artery by more than 15 µ for about 30% of the measuring time. The use of a low-vacuum contact lens does not decrease eye movements. The authors say that "fine adjustments may easily be made at any time" (p 895). In our experience with a camera and with a slit lamp, it is not possible to adjust the position of two fibers simultaneously to achieve the kind of accuracy needed to ensure that no one fiber collects fluorescence from the dye in the choroid. 4. Analysis of the data reported by Cunha-Vaz and Lima shows that in normal subjects (their Table 1), the average maximum speed of fluores¬ cein is 0.9 mm/0.17 s 5 mm/s. This is six to eight times lower than values found by other investigators in pigs, monkeys, and humans, using high¬ speed fluorescein cineangiography.4 :,-? This lower value for the dye maximum speed cannot be attributed to the reported elevation of the intraocular pressure (IOP) to a level of 30 mm Hg. An increase of 10 to 15 mm Hg above normal IOP would reduce the dye maximum speed by only about 20%," ie, to value around 30 mm/s. Even though the dye speeds re¬ ported in Table 1 are much slower than those found by others, the flow values obtained by Cunha-Vaz and Lima are in accordance with other published data.4 This is so because the value given for the diameter of the superior temporal artery in their normal subjects is about twice the value normally measured. Typically, in normal subjects the average diame¬ ter of large arteries adjacent to the optic disc is 100 µ," not 184 µ as given by the authors (Table 1). The only explanation we can offer for such a discrepancy is that the authors did not account for the magnification factor of their fundus angiography system, which is most probably around 2. Fail¬ ure to consider this point would result in overestimation of the flow values by a factor of about 4. In any case, =
Downloaded From: http://archopht.jamanetwork.com/ by a University of Michigan User on 06/18/2015
these flow values would still be far below the range of those published by the same authors.1" In conclusion, we have indicated various reasons that preclude the application of the technique described by Cunha-Vaz and Lima to the deter¬ mination of retinal blood flow. We believe that the blood flow values in normal and anemic subjects are not correct. The conclusions of these investigators regarding changes of blood flow in various stages of diabetes,1 even if correct, cannot be supported by their technique of mea¬ surement.
Charles E. Riva, DSc Boston Gilbert T. Feke, PhD, George T. Timberlake, PhD, Steven Sinclair, MD, and Michael Loebl, MD, provided advice and assistance and Flavia Blackwell provided editorial assistance. 1. Cunha-Vaz et al: Studies
JG, Fonseca JR, de Abreu JFR,
retinal blood flow: II. Diabetic Arch Ophthalmol 96:809-811, 1978. CT: Dynamic aspects of the retinal microcirculation. Arch Ophthalmol 79:536-539, 1968. 3. Bulpitt CJ, Kohner EM, Dollery CT: Velocity profiles in the retinal microcirculation. Bibl Anat 11:448-452, 1973. 4. Eberli B, Feke GT, Rogers F, et al: Continuous recording of the speed of red blood cells in human retinal vesels. Presented at the Association for Research in Vision and Ophthalmology. Sarasota, Fla, May 5, 1978. 5. Van Heuven WAJ, Malik AB, Schaffer CA, et al: Retinal blood flow derived from dye dilution curves. Televised fluorescein angiography. Arch Ophthalmol 95:297-301, 1977. 6. Ditchburn RW: Eye-movements and Visual Perception. London, Clarendon Press, 1973. 7. Wise GL, Dollery CT, Henkind P: The Retinal Circulation. New York, Harper & Row Publishers Inc, 1971, pp 83-118. 8. Ffytche TJ, Bulpitt CJ, Kohner EM, et al: Effect of changes in intraocular pressure on the retinal microcirculation. Br J Ophthalmol 58:514\x=req-\ 522, 1974. 9. Hogan MJ, Alvarado JA, Esperson Weddell J: Histology of the Human Eye: An Atlas and Textbook. Philadelphia, WB Saunders Co, 1971. 10. Cunha-Vaz JG, de Lima JJP: Estimation of human retinal blood flow by slitlamp fluorophotometry: II. Results and discussion. IRCS Med Sci 3:577, 1975. on
retinopathy. 2. Dollery
Reply.\p=m-\Itshould be obvious to Dr colleagues, as it is to everyone else, that a technique that permits absolute measurements of In
Riva and his
retinal blood flow in not yet available. That has never been our claim because we realize that our technique has most of the drawbacks and limitations of the two-point photometric systems, such as the one used by Riva and associates. With these facts in mind, I will try to clarify points raised in their letter. 1. We never stated in our report that we are measuring the first fluorescein molecules passing through the artery, nor was the phrase "first appearance time of fluorescein" ever
