Scott IU, Jackson GR, Quillen DA, Larsen M, Klein R, Liao J, Holfort S, Munch IC, Gardner TW. Effect of Doxycycline vs Placebo on Retinal Function and Diabetic Retinopathy Progression in Patients With Severe Nonproliferative or Non–High-Risk Proliferative Diabetic RetinopathyA Randomized Clinical Trial. JAMA Ophthalmol. 2014;132(5):535-543. doi:10.1001/jamaophthalmol.2014.93
Inflammation may contribute to the pathogenesis of diabetic retinopathy (DR).
To investigate, in a proof-of-concept clinical trial, whether low-dose oral doxycycline monohydrate can (1) slow the deterioration of, or improve, retinal function or (2) induce regression or slow the progression of DR in patients with severe nonproliferative DR (NPDR) or non–high-risk proliferative (PDR), and to determine the potential usefulness of visual function end points to expedite the feasibility of conducting proof-of-concept clinical trials in patients with DR.
Design, Setting, and Participants
We conducted a randomized, double-masked, 24-month proof-of-concept clinical trial. Thirty patients (from hospital-based retina practices) with 1 or more eyes with severe NPDR or PDR less than Early Treatment Diabetic Retinopathy Study–defined high-risk PDR.
Patients were randomized to receive 50 mg of doxycycline monohydrate or placebo daily for 24 months.
Main Outcomes and Measures
Change at 24 months compared with baseline in functional factors (frequency doubling perimetry [FDP], Humphrey photopic Swedish Interactive Thresholding Algorithm 24-2 testing, contrast sensitivity, dark adaptation, visual acuity, and quality of life) and anatomic factors (Early Treatment Diabetic Retinopathy Study DR severity level, area of retinal thickening, central macular thickness, macular volume, and retinal vessel diameters).
From baseline to month 24, mean FDP foveal sensitivity decreased in the placebo group (−1.9 dB) and increased in the doxycycline group (+1.8 dB) (P = .02). A higher mean FDP foveal sensitivity in the doxycycline group compared with the placebo group was detected at 6 months (P = .04), and this significant difference persisted at 12 and 24 months. A difference between the groups was not detected with respect to the other visual function outcomes and all anatomic outcomes assessed.
Conclusions and Relevance
To our knowledge, this is the first observation suggesting a link between a low-dose oral anti-inflammatory agent and subclinical improvement in inner retinal function. Oral doxycycline may be a promising therapeutic strategy targeting the inflammatory component of DR. Furthermore, study results suggest that FDP, which primarily measures inner retinal function, is responsive to intervention and may be a useful clinical trial end point for proof-of-concept studies in patients with DR.
clinicaltrials.gov Identifier: NCT00511875
Diabetic retinopathy (DR) is the leading cause of visual impairment and blindness among working-aged adults around the world.1- 3 The public health burden of DR is expected to rise even further because of the increasing prevalence of diabetes mellitus and the increasing lifespan of persons with diabetes.4- 6 At present, there are no established means to slow the course of nonproliferative DR (NPDR) except for glucose, blood pressure, and cholesterol control. Therefore, new therapeutic approaches are in great demand.
The pathophysiology of DR is not completely understood, and most research has focused on vascular changes. However, more recently, involvement of neuronal and glial components in early stages of retinopathy has been observed.7 Chronic inflammatory processes that occur in experimental models of DR include increased nitric oxide production, intracellular adhesion molecule 1 upregulation, leukostasis, and increased expression of proinflammatory cytokines associated with vascular damage and neuronal cell loss.8- 10 Microglia represent the primary resident immune cell of the retina and, although normally quiescent, become activated by diabetes and have been reported11- 18 to induce inflammatory changes underlying DR. Tetracycline derivatives have been demonstrated17,19- 22 in cell culture and animal models to have anti-inflammatory and neuroprotective effects independent of their antibacterial effects. With respect to retinal disease, low-dose tetracyclines have been demonstrated17,19- 22 to inhibit microglial morphology changes, matrix metalloproteinase activity, and retinal cell apoptosis coincident with decreased caspase activity. In an open-label phase 1/2 clinical study23 including 5 participants with fovea-involving diabetic macular edema, oral minocycline, 100 mg, twice daily for 6 months was associated with improved visual acuity and reduced central macular thickness and vascular leakage, comparing favorably with historical controls from previous studies. The purpose of the present proof-of-concept clinical trial was to investigate whether oral doxycycline monohydrate, a drug capable of inhibiting microglial activation, can (1) slow the deterioration of, or improve, retinal function or (2) induce regression or slow the progression of diabetic retinopathy in patients with severe NPDR or non–high-risk proliferative DR (PDR).
Traditionally, proof-of-concept studies have been hampered by insensitive clinical trial end points, which require large sample sizes or long study durations to achieve sufficiently high power to detect potential treatment signals. To enhance the feasibility of detecting a treatment effect, we included retinal function end points. We hypothesized that the most promising of the 6 retinal function measurements included in the present study is frequency doubling perimetry (FDP) because it predominantly reflects inner retinal function, which is affected prominently in preclinical animal models of diabetes24- 26 and in patients with mild DR.27- 32 A previous study27 reported that, compared with nondiabetic individuals serving as controls, persons with NPDR had a 7.1-dB lower mean FDP foveal sensitivity (P < .001) and persons with diabetes without retinopathy had a 2.9-dB lower mean FDP foveal sensitivity (P < .001).
This randomized, double-masked, 24-month clinical trial adhered to the tenets of the Declaration of Helsinki. Approval for the study protocol was obtained from the Penn State College of Medicine Institutional Review Board, the University of Wisconsin–Madison School of Medicine Institutional Review Board, the Medical Ethics Committee of the Capital Region of Denmark, and the Danish Medicines Agency. Health Insurance Portability and Accountability Act–compliant written informed consent or an equivalent Danish consent form was obtained from all participants before eligibility screening and again before randomization into the study. Participants received financial compensation. Study oversight was provided by an independent data and safety monitoring committee. Eligible patients were recruited from the retina clinics at Penn State Hershey Eye Center and Glostrup Hospital between October 1, 2008, and May 31, 2010, and included adults with type 1 or type 2 diabetes mellitus and at least 1 eye with severe NPDR (Early Treatment Diabetic Retinopathy Study [ETDRS] level 53E) or PDR less than high-risk PDR as defined by the DRS (ETDRS level 61 or 65)33- 35 based on modified 7-standard field color stereoscopic fundus photographs graded at the University of Wisconsin–Madison Ophthalmic Epidemiology Reading Center (fundus photography procedures are described in the University of Wisconsin–Madison Ophthalmic Epidemiology Reading Center fundus photography protocol35- 39; the modified 7-standard field protocol calls for 1 or more additional fields if neovascularization is present outside of [or poorly documented within] the 7 standard fields). The rationale for studying patients with these grades of retinopathy is that they have significant retinopathy and there are no established means to modify the course of DR in this population except for glucose, blood pressure, and cholesterol control. Patients in whom panretinal photocoagulation was imminently required in the judgment of the treating retina specialist were excluded. Study inclusion and exclusion criteria are listed in Table 1. If both eyes met the eligibility criteria, the eye with the more advanced ETDRS DR level, as determined by grading of the fundus photographs at the University of Wisconsin–Madison Ophthalmic Epidemiology Reading Center, was selected to be the study eye. If both eyes were judged to have the same DR level, the eye with better visual acuity was designated the study eye. If both eyes had the same visual acuity, the right eye was designated the study eye.
The randomization scheme was a permuted block design, with eligible patients stratified by clinical center, dichotomized time since the onset of diabetes (<10 years vs ≥10 years), and dichotomized hemoglobin A1c level (<9% vs ≥9%; to convert to proportion of total hemoglobin, multiply by 0.01). Eligible patients were randomized in a 1:1 allocation ratio to receive either doxycycline monohydrate, 50 mg, or an identical placebo once daily before meals for 24 months. Study investigators, clinical coordinators, study photographers, and patients were masked to treatment assignment. Treatment adherence was assessed with monthly telephone calls from the clinical coordinator and by pill counts of the containers returned by each participant at each study visit. Study participant adherence to the treatment regimen was evaluated as the percentage of prescribed pills taken since the previous visit. If this percentage was 75% or less, the participant was considered nonadherent for the period since the previous evaluation. Participants deemed nonadherent received telephone reminders, counseling, or both, as appropriate. Nonadherent patients were not discontinued from the study because of the intent-to-treat design.
At baseline, patients underwent complete ocular examinations, including best-corrected visual acuity as assessed by the electronic E-ETDRS method,40 intraocular pressure measurement by Goldmann applanation, slitlamp and dilated funduscopic examinations, contrast sensitivity testing,41 visual field testing including FDP and Humphrey photopic Swedish Interactive Thresholding Algorithm (SITA) 24-2 visual field testing, dark adaptation testing (AdaptDx, a computer–automated dark adaptometer; MacuLogix), vision-specific quality-of-life assessment questionnaires (National Eye Institute Visual Function Questionnaire42 and Low Luminance Questionnaire43), time domain optical coherence tomography (OCT) (Stratus OCT; Carl Zeiss Meditec) imaging of each eye, modified 7-standard field color stereoscopic fundus photography, and fluorescein angiography (imaging procedures are described in the University of Wisconsin–Madison Ophthalmic Epidemiology Reading Center optical coherence tomography, fundus photography, and fluorescein angiography protocols).35- 39 Retinal vessel diameters were measured on 50° disc-centered fundus photographs in a zone around the optic nerve head and summarized as central retinal vein–equivalent diameter and central retinal artery–equivalent diameter.44
The visual function tests evaluated in the present study were selected because previous studies27,45 found that they are at least moderately impaired in patients with diabetes with or without DR. Ganglion cell function is thought to be primarily responsible for contrast sensitivity and FDP responses; therefore, contrast sensitivity and FDP were used to assess inner retinal function.27,45 Light-adapted visual sensitivity (standard white-on-white perimetry) and dark adaptation were used to assess outer retinal (photoreceptor) function. Cone photoreceptor light sensitivity is primarily responsible for light-adapted visual sensitivity, and rod photoreceptor function is responsible for dark adaptation performance.27,45
Before visual function measurements, participants underwent ETDRS refraction to determine the best optical correction for the test distance. The E-ETDRS visual acuity tester was calibrated for each visual acuity measurement so that the mean luminance of the monitor was in the specified range of 85-105 candela (cd)/m2.40 Contrast sensitivity was assessed with the Pelli-Robson contrast sensitivity chart with a luminance of 100 cd/m2 (Haag-Streit USA). The Matrix perimeter (Carl Zeiss Meditec) was used to measure the FDP 24-2 visual field. The Matrix stimulus is 0.25 cycles per degree of sinusoidal grating, which is phase reversed at 18 Hz. The grating appears to have twice as many alternating light and dark bars than are actually present. The minimum contrast threshold of the 5.0° diameter stimulus was measured at each of the 55 test locations. The frequency doubling illusion on which FDP is based is thought to arise in the magnocellular visual pathway and be ganglion cell dependent.46,47 Standard white-on-white perimetry using the SITA algorithm (Humphrey Field Analyzer 700 series; Carl Zeiss Meditec) was performed for photopic 24-2 visual field testing. For each visual field test, if the results were unreliable, the participant repeated the test. An unreliable visual field test was defined as a test with any of the following: (1) false-positives greater than 33%, (2) false-negatives greater than 33%, and (3) fixation losses greater than 33%. Study participants’ eyes were then dilated using tropicamide, 1%, and phenylephrine hydrochloride, 2.5%. Dark adaptation testing was performed after the pupils dilated to a minimum of 6 mm. Rod-mediated dark adaptation was measured for a 2° circular test spot located 5° superior to the fovea (AdaptDx; MacuLogix). Dark adaptation measures the sensitivity recovery of the rod photoreceptors in the dark following exposure to a moderate-intensity 2-millisecond flash (5.8 × 104 scotopic cd/m2 per second, equivalent to an 80% bleaching level). After bleaching, sensitivity recovery was measured for up to 20 minutes. Speed of dark adaptation was characterized by the rod intercept derived from the slope of the second component of rod-mediated dark adaptation.27,48 Thus, dark adaptation measured the speed of rod photoreceptor sensitivity recovery. Seven-field stereoscopic color fundus photographs were taken of both eyes (TRC 50-EX fundus camera; Topcon).
Follow-up ophthalmologic examinations (including E-ETDRS visual acuity, intraocular pressure measurement, and slitlamp and dilated funduscopic examinations) were performed at 3, 6, 9, 12, 15, 18, 21, and 24 months. Contrast sensitivity testing, FDP, photopic visual field testing, dark adaptation testing, quality-of-life assessment questionnaires, OCT, and modified 7-standard field color stereoscopic fundus photographs were performed at 6, 12, 18, and 24 months. Fluorescein angiography was performed at 24 months.
Hemoglobin A1c concentration, serum cholesterol and serum creatinine levels, a pregnancy test result in women with child-bearing potential, blood pressure, and body mass index were measured at baseline and every 3 months for 24 months. All adverse events were recorded. Fundus photographs were sent to the University of Wisconsin–Madison Ophthalmic Epidemiology Reading Center, where the ETDRS DR severity level was graded, the presence or absence of retinal and/or disc neovascularization was assessed, and the area of retinal thickening was measured. Side-by-side comparisons of the same eyes were performed. Fluorescein angiograms and OCT images were reviewed by the study investigators (I.U.S., G.R.J., D.A.Q., M.L., and T.W.G.).
Main outcome measures included comparison between the placebo and doxycycline groups with respect to change at 24 months compared with baseline in mean foveal sensitivity (in decibels) as measured by FDP, field sensitivity (decibels) as measured by FDP, foveal sensitivity (decibels) as measured by photopic visual field, field sensitivity (decibels) as measured by photopic visual field, log contrast sensitivity, speed of rod photoreceptor sensitivity recovery (minutes), ETDRS visual acuity (letters correct), ETDRS DR severity level, area of retinal thickening, central subfield thickness on OCT, macular volume on OCT, National Eye Institute Visual Function Questionnaire total score, and Low Luminance Questionnaire total score. In addition, the proportion of study eyes that met each of the following criteria at 24 months vs baseline were compared between the doxycycline and placebo groups: progression to high-risk PDR as defined by the DRS,33,34 a 1-step or more progression in ETDRS DR severity level, a 2-step or more progression in ETDRS DR severity level, a 3-step or more progression in ETDRS DR severity level, received panretinal photocoagulation during the study period, and received focal/grid macular photocoagulation during the study period. The trial was reviewed and monitored by an independent data safety and monitoring committee.
Summary statistics for demographic and outcome variables at baseline are provided in Table 2. The difference between the 2 randomization groups for each variable was compared using the Fisher exact test for categorical characteristics or the Welch 2-tailed, 2-sample t test for continuous measurements. No adjustment for multiple testing was made for this exploratory phase 2 study with small sample sizes. Similar analyses were performed for change in visual function at month 24 compared with baseline (Table 3), anatomic outcomes at month 24 (Table 4), and change in anatomic outcomes of the study eye at month 24 compared with baseline (Table 5).
The study enrolled 15 participants in the placebo group and 15 participants in the doxycycline group between October 2008 and May 2010. Four of the participants (27%) in the doxycycline group were lost to follow-up, and 1 participant (1%) in the doxycycline group had missing values in approximately half of the variables at month 24 resulting from fatigue during the study visit; the remaining 25 study participants (83%) completed 24 months of follow-up. Demographic and baseline characteristics of the study population are summarized in Table 2. A significant difference between study groups was not detected with respect to demographic characteristics and most of the baseline characteristics assessed. Patients in the doxycycline group had a shorter duration of diabetes mellitus compared with patients in the placebo group (19.6 vs 27.2 years; P = .03), despite no detectable difference between the groups with respect to severity of DR and degree of metabolic control.
Comparisons of change in visual function at month 24 compared with baseline in the placebo group vs the doxycycline group are reported in Table 3. From baseline to month 24, mean FDP foveal sensitivity decreased in the placebo group (−1.9 dB) and increased in the doxycycline group (+1.8 dB) (P = .02). After adjustment for the duration of diabetes, the treatment effect at 24 months remained statistically significant (P = .03). An improvement in FDP foveal sensitivity at month 24 compared with baseline was observed in 5 of 10 patients (50%) in the doxycycline group compared with none of 15 persons in the placebo group (P = .005). From baseline to month 12, mean FDP foveal sensitivity decreased in the placebo group (−1.60 dB) and increased in the doxycycline group (+1.66 dB) (P = .04); an improvement in FDP foveal sensitivity was observed in 6 of 12 patients (50%) in the doxycycline group compared with 1 of 15 patients (1%) in the placebo group (P = .02). A significantly higher mean FDP foveal sensitivity in the doxycycline group compared with the placebo group was first detected at 6 months (P = .04), and a significant difference between the two groups persisted at 12 months (P = .04) and 24 months (P = .02). The mean FDP foveal sensitivity at the 18-month time point was not significantly different between the two groups (P = .47); at the 18-month time point, 1 patient in the doxycycline arm (1%) exhibited unexplained and temporary sensitivity loss of 11 dB, which was not sustained at 24 months. Removing this outlier from the analysis resulted in a significantly higher mean FDP foveal sensitivity in the doxycycline group compared with the placebo group (P = .03). There was no significant difference between study groups with respect to the other visual function outcomes assessed.
Comparisons of anatomic outcomes in the placebo group vs the doxycycline group are displayed in Tables 4 and 5. A difference between the study groups was not detected with respect to the anatomic outcomes assessed.
Three serious adverse events were reported (motor vehicle accident, elevated serum creatinine level, and vitreous hemorrhage); none was considered related to the study drug. One mild adverse event (nausea) was considered potentially related to treatment. No drug-related ocular adverse events were reported.
In this 24-month, randomized, proof-of-concept clinical trial of patients with severe NPDR or PDR less than high-risk PDR, oral doxycycline monohydrate, 50 mg, daily was associated with significantly improved mean FDP foveal sensitivity compared with placebo. The study drug was well tolerated, with minimal drug-related adverse events and no drug-related ocular adverse events. Study limitations include a small sample size, loss to follow-up of 4 patients in the doxycycline group, missing values in approximately half of the variables at month 24 in a patient in the doxycycline group due to fatigue during the study visit, and a paucity of information related to the associations between change in FDP and (1) visual prognosis and (2) change in rate of DR progression.
Despite the small sample size, which constrains the ability to detect significant differences between study groups, a significant difference in mean FDP foveal sensitivity was detected between the doxycycline and placebo groups after 24 months of treatment, and the significant difference was present as early as 6 months following study initiation. Given the multiple outcomes assessed, it is possible that the significant difference between groups in FDP foveal sensitivity is due to chance. However, for the following reasons, we believe the observed difference in FDP foveal sensitivity is real. First, the statistical significance at 24 months reached P = .005, which is 10 times smaller than the usual significance threshold of P = .05, despite the small sample size inherent to the proof-of-concept design of the study. Second, statistical significance was observed at multiple time points and in the same direction (higher FDP foveal sensitivity in the doxycycline group compared with the placebo group) throughout the study. Third, there is consistency of the effect with previously published work27 and the fact that FDP predominantly reflects inner retinal function, which is a prominent characteristic in preclinical models of diabetes24- 26 and in patients with mild DR.27- 32 The absence of any detectable significant differences between groups with respect to visual acuity and anatomic factors, which are used most often in clinical management and clinical trials of patients with DR, suggests that more sensitive clinical end points are needed to evaluate potential treatments and preventive strategies for DR.
To put the mean FDP difference between the doxycycline and placebo groups of 3.7 dB at 24 months in context, it is useful to consider the precision of the Matrix FDP. Test-retest reliability data were collected on 30 participants with early NPDR to early PDR. The precision of the foveal sensitivity measurement was ±2.88 dB. Thus, the observed effect was larger than the precision of the foveal sensitivity measurement of the Matrix. In addition, the effect was persistent with the exception of the 18-month time point because of an outlier measurement for one participant. The effect size was modest but apparently robust.
The significant improvement in FDP foveal sensitivity in the doxycycline group compared with the placebo group could result from multiple mechanisms of inflammation reduction, including decreased vascular endothelial growth factor or interleukin 1 expression, matrix metalloproteinase activity, or microglial cell–mediated inflammation, and suggests that targeting inflammatory and neurodegenerative components of DR may be a promising therapeutic strategy. Tetracycline antibiotics, the bacterial target of which is the prokaryotic ribosome, inhibit mammalian mitochondrial protein synthesis while leaving mammalian cytoplasmic ribosomal protein synthesis unaffected.49,50 Mitochondrial function is important for inflammation, apoptosis, and energy metabolism.51 Thus, there are several potential mechanisms whereby doxycycline may modulate retinal function by altering mitochondrial function, including being involved in the conditioning effect of changes in glycemia on retinal function.52- 54 Therapeutic agents with pleiotropic effects may offer advantages to address the complex mechanisms of DR.
Findings of the present study are consistent with those of an open-label phase 1/2 clinical study23 including 5 participants with fovea-involving diabetic macular edema who received oral minocycline, 100 mg, twice daily for 6 months. In that study, minocycline was associated with improved visual acuity and reduced central macular thickness and vascular leakage, comparing favorably with historical controls from previous studies. Taken together, the results of these studies suggest that pharmacologic strategies to inhibit inflammatory or neurodegenerative changes may help slow the progression of DR. However, further studies are essential before tetracyclines should be considered as therapy for patients with DR.
Novel retinal function end points were used in this proof-of-concept study because we hypothesized that using only traditional clinical trial end points, such as visual acuity and anatomic progression of DR assessed by fundus photograph grading, would require a longer study duration or increased sample size to determine whether there was an effect of the study drug. Selection of FDP as an end point of treatment efficacy was based on its high diagnostic sensitivity for NPDR27 and neural fiber layer loss observed in human imaging studies.28- 30 Additional retinal function end points (contrast sensitivity, visual field, and dark adaptation) were evaluated to directly compare the sensitivities of these functional end points to intervention in patients with DR. Dark adaptation was responsive to tight glycemic control in a prior study,54 but did not differentiate between the study groups in the present study. Frequency doubling perimetry was the only visual function variable assessed that was responsive to intervention in the present study.
In this 24-month proof-of-concept clinical trial of patients with severe NPDR or non–high-risk PDR, daily doxycycline monohydrate, 50 mg, was associated with significantly improved FDP foveal sensitivity compared with placebo. Oral doxycycline may be a beneficial therapeutic strategy targeting the inflammatory component of DR. Frequency doubling perimetry is a promising clinical trial end point for proof-of-concept studies because of its high sensitivity to DR, responsiveness to intervention, relative rapidity (<10 minutes required to test 1 eye), and ease of use for the operator and patient. Studies evaluating the natural history of FDP responses in patients with diabetes and its clinical meaningfulness are warranted.
Submitted for Publication: June 28, 2013; final revision received November 1, 2013; accepted November 16, 2013.
Corresponding Author: Ingrid U. Scott, MD, MPH, Penn State Hershey Eye Center, Penn State College of Medicine, 500 University Dr, HU19, Hershey, PA 17033-0850 (firstname.lastname@example.org).
Published Online: March 6, 2014. doi:10.1001/jamaophthalmol.2014.93.
Author Contributions: Drs Gardner and Jackson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Scott, Jackson, Liao, Gardner.
Acquisition, analysis, or interpretation of data: Quillen, Larsen, Klein, Holfort, Munch, Gardner.
Analysis and interpretation of data: Scott, Jackson, Liao, Gardner.
Drafting of the manuscript: Scott, Jackson, Liao, Gardner.
Critical revision of the manuscript for important intellectual content: Jackson, Quillen, Larsen, Klein, Holfort, Munch, Gardner.
Statistical analysis: Jackson, Larsen, Liao.
Obtained funding: Holfort, Gardner.
Administrative, technical, or material support: Scott, Jackson, Quillen, Larsen, Holfort, Munch, Gardner.
Study supervision: Jackson, Larsen, Klein, Holfort.
Conflict of Interest Disclosures: Dr Jackson is an investor and employee of MacuLogix, Inc, Hershey, Pennsylvania, the manufacturer of the AdaptDx. No other disclosures were reported.
Funding/Support: The trial was supported in part by grant 4-2007-231 from the Juvenile Diabetes Research Foundation, a Research to Prevent Blindness Physician-Scientist Award, and the A. Alfred Taubman Medical Research Institute (Dr Gardner).
Role of the Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Maria Emanuel Ryan, DDS, PhD, of Stony Brook University, Stony Brook, New York, and Steve Levison, PhD, of Rutgers University–New Jersey Medical School, Newark, provided helpful discussions regarding the anti-inflammatory effects of tetracyclines in persons with diabetes. Steven F. Abcouwer, PhD, of the University of Michigan Kellogg Eye Center, Ann Arbor, provided helpful suggestions for the article. None of these individuals received financial compensation for their services.