The Anti-thyroid Drug Methimazole Induces Neovascularization in the Neonatal Rat Analogous to ROP

Abstract

purpose. To determine the effect of methimazole (MMI), an anti-thyroid drug known to reduce serum l-thyroxine (T4), and insulin-like growth factor (IGF)-1 concentrations, on retinal vascular development in neonatal rats.

methods. Sprague–Dawley rats (n = 175) were raised in expanded litters of 25 in room air and were exposed to MMI from birth (given as a 0.1% solution to nursing mothers for either 4 or 10 days). Experiments ended on day 4 (n = 25) or 10 (n = 50) of life. A third group was exposed to MMI for the initial 4 days of life and then allowed to recover for the next 6 days (n = 50). Fifty control rats were analyzed on day 4 (n = 25) or 10 (n = 25) of life. Left eyes were fixed, and retinas were dissected and stained with adenosine diphosphatase (ADPase). Retinas were graded for presence and severity of neovascularization (NV) in a masked manner, and retinal vascular areas were quantified. In a subsequent study, serum IGF-1 and T4 levels were measured by radioimmunoassay in an additional 200 rats exposed to treatments identical to those described.

results. Retinal NV occurred in 31% of rats exposed to 10 days of MMI and 4% (P = 0.02) of rats exposed to 4 days of MMI, followed by 6 days of recovery. None of the rats exposed to 4 days of MMI alone and none of the control animals was graded positive for NV. Retinal vascular areas were significantly reduced in rats exposed to 4 days of MMI compared with 4-day control animals (36% ± 6% vs. 50% ± 6%, P = 0.0001). Serum IGF-1 levels were markedly reduced in 4-day MMI rats compared with age-matched control animals (42 ng/mL vs. 133 ng/mL, P = 0.0001) and in 10-day MMI rats compared with 10-day control animals (133 ng/mL vs. 206.5 ng/mL, P = 0.005). Serum T4 levels were similarly suppressed in the MMI-exposed litters compared with control animals at day 10 (P = 0.008). In contrast, rats exposed to 4 days of MMI followed by 6 days of recovery had normal serum IGF-1 and T4 levels by day 10.

conclusions. The anti-thyroid drug, MMI, induces NV in neonatal rats. This may be mediated by the initial suppression of serum IGF-1. Nevertheless, the lower incidence of NV when serum IGF-1 levels are initially suppressed followed by complete recovery, is contrary to a purely permissive role for serum IGF-1, as reported previously. The relationship between the temporal course of serum IGF-1 and NV in immature retinas needs further investigation.

Retinopathy of prematurity (ROP) continues to be a significant cause of blindness worldwide. It is possible that further reductions in the prevalence and severity of blindness caused by ROP can be made only with advancements in the prevention of neovascularization (NV), rather than treatment of NV once it occurs. 

Hellstrom et al. ¹ have recently reported an association between serum insulin-like growth factor (IGF)-1 and ROP. They measured serum IGF-1 in preterm infants and found that initial low concentrations of serum IGF-1 were associated with the subsequent development of ROP. They hypothesized that retarded retinal vessel growth, in infants with low serum IGF-1 levels, results in hypoxia in the nonvascular retina, stimulating synthesis and accumulation of VEGF. As the infants mature, and when serum IGF-1 concentrations gradually increase beyond a critical threshold, this may allow VEGF-driven endothelial cell proliferation to proceed, inducing NV. 

Previous studies in neonatal rats have shown that serum IGF-1 can be suppressed by treating mothers with methimazole (MMI). ² A similar decrease in serum IGF-1 has also been shown in human hyperthyroid patients after treatment with MMI. ³ In a recent study by Berkowitz et al., ⁴ MMI treatment of neonatal rats was reported to increase the incidence and severity of NV in an oxygen-induced retinopathy (OIR) model of ROP. 

The purpose of our study was to determine the effect of MMI, a drug known to decrease serum levels of IGF-1 and thyroxine (T4), on the development of normal retinal vasculature. 

Methods

All experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee at our institution. 

Animals

Pregnant Sprague–Dawley rats were obtained from Harlan (Indianapolis, IN). Dams received a standard laboratory diet and either water ad libitum or a 0.1% solution of methimazole (MMI) in water. Light was cycled on a 12-hour light–dark schedule, and the room temperature was maintained at approximately 21°C. All animals were raised in room air and all neonates were weighed daily. 

Retinopathy Study Animals

Newborn pups from dams delivering on the same day were assigned within 24 hours of birth to expanded litters of 25. We have shown that raising neonatal rats in expanded litters results in increased incidence and severity of NV in rat models of ROP. ^{5 6} Twenty-five rat pups (n = 1 litter) were exposed to MMI (Sigma-Aldrich, St. Louis, MO) by using a 0.1% MMI solution as the drinking water of nursing dams from the time of delivery until the rats were analyzed on day 4 of life (4-day MMI rats). Fifty rats (n = 2 litters) were similarly exposed to MMI for 10 days and analyzed on day 10 (10-day MMI rats). A third group of 50 rats (n = 2 litters) were exposed to MMI for the initial 4 days of life and then allowed to recover for the next 6 days when they were analyzed (4-day MMI plus recovery rats). Control rats were never exposed to MMI and were analyzed on either day 4 (4-day control animals, n = 1 litter) or 10 (10-day control animals, n = 1 litter) of life. MMI solution fed to the mothers was changed every other day, and opaque bottles were used to prevent photodeterioration. 

Serum IGF-1 and T4 Animals

In a subsequent study to determine serum IGF-1 and serum T4 levels, 200 rats were raised in expanded litters and exposed to identical treatments as described earlier (4-day MMI, n = 50; 10-day MMI, n = 25; 4-day MMI plus recovery, n = 25; 4-day control animals, n = 50 rats; and 10-day control animals, n = 50). 

Preliminary studies in our laboratory and others ⁷ showed a possible interaction between IGF binding proteins and ketamine anesthesia (data not shown). Therefore, all animals included in the IGF-1 and T4 studies received CO₂ anesthesia. 

Analysis of Retinal Histology

On days 4 or 10 of life, rats from the retinopathy study were anesthetized with either an intramuscular injection of ketamine (80 mg/kg) and xylazine (15 mg/kg) or inhaled CO₂. To evaluate vessel morphology, left eyes were removed and fixed with 10% neutral buffered formalin for 90 minutes at 4°C. The cornea, lens, and vitreous were surgically removed, and the retina was dissected and flatmounted. Retinas were processed for magnesium-activated adenosine diphosphatase (ADPase) staining as described by Lutty and McLeod. ⁸ ADPase-stained retinas were flatmounted on microscope slides in aqueous medium (Aquamount; Lerner Laboratories, Pittsburgh, PA) with a coverslip. ADPase-stained retinas were graded for NV in a masked manner by a standard method previously validated in our laboratory. ⁹ To reduce bias toward false positives, age-matched–grading control animals were included in the masked NV evaluation. 

Each retinal quadrant was visually divided into three equal parts, or clock hours, and each clock hour was evaluated for the presence or absence of NV. Neovasularization was defined as clumps, sheets, or tufts of endothelial cells morphologically distinct from the normal vasculature, arising at the junction of the vascular and avascular retinas, as described in previous studies. ⁹ Retinas were scored for the severity of NV by counting the number of clock hours containing NV. ^{9 10} Cross-sectional histology was not performed in this study due to the high correlation of our grading method to the number of cells above the inner limiting membrane of the retina. ⁹  

For analysis of retinal vascular areas, ADPase-stained retinas were imaged with a digital camera (Spot Insight Color, model 3.20; Diagnostic Instruments Inc., Sterling Heights, MI) attached to a light microscope (Laborlux II; Ernst Leitz, Rockleigh, NJ). Vascularized and total retinal areas were traced in a masked manner using Analyze ¹¹ image analysis software (ver. 6.0.3b; available at analyzedirect.com), and the ratio of vascular to total retinal area was calculated. 

Serum IGF-1 Analysis

Carotid artery blood samples were obtained from randomly selected pups (n = 8–19 samples per group). It was not possible to collect serial blood samples across days on any single animal, since obtaining sufficient arterial blood from these neonatal animals is a terminal event, and all animals were killed immediately after blood collection. 

For the blood collection, pups were lightly anesthetized with inhaled CO₂. Under dissecting microscopy, the left carotid artery was exposed through a skin incision. The artery was transected, and 300 to 400 μL of arterial blood was collected with a 21-gauge blood collection set (Vacutainer; BD Biosciences, Franklin Lakes, NJ). Blood samples were allowed to clot on ice for 20 minutes, and then centrifuged at 3000g for 5 minutes. Serum was removed and stored at −80°C until analyzed. Acid-ethanol extraction was performed to remove IGF-1 binding proteins. ¹² Briefly, 200 μL of acid-ethanol mixture (87.5% ethanol: 12.5% [vol/vol] 2 M hydrochloric acid) was added to 50 μL of serum in 1.5 mL polypropylene tubes. The mixture was vortexed and incubated at 4°C for 30 minutes and centrifuged at 13,000g for 15 minutes. Two-hundred microliters of supernatant was removed and neutralized with 80 μL of 0.86 M Tris base, vortexed, and incubated at −20°C for 1 hour, followed by centrifugation at 13,000g for 10 minutes. Supernatant was removed and used for IGF-1 assays. Serum samples were immediately frozen at −70°C and IGF-1 radioimmunoassays were performed by the National Hormone and Peptide Program (Torrance, CA). For comparison to some other reports, values reported as nanograms per milliliter can be converted to nanomolar by dividing by 7.65, based on a molecular weight of IGF-1 of 7649. 

Serum T4 Analysis

Serum samples from the IGF-1 study just described were aliquoted and stored at −70°C for serum T4 analysis. T4 radioimmunoassays (n = 6–11 samples per group) were performed by the Yerkes Core Endocrine Laboratory (Atlanta, GA). The sensitivity of the assay was 1.0 μg/dL. 

Statistical Analysis

Incidence of NV was compared between groups by using the Fisher exact test. Severity of NV was compared using Wilcoxon tests. Rat weights were compared at each day using ANOVA and post hoc Student’s t-tests with Bonferroni corrections. Bonferroni-corrected P < 0.05 were considered statistically significant. The proportions of vascularized retina area were compared between groups of interest using t-tests. Serum IGF-1 concentrations and serum T4 concentrations were compared between groups with either Wilcoxon tests (4-day values) or Kruskal-Wallis tests and post hoc Wilcoxon tests with Bonferroni corrections (10-day values). All statistical analysis was performed using SAS (ver. 6.12 for Windows; SAS Institute, Cary, NC). 

Results

Retinopathy Study

Animal Survival and Retinas Analyzed.

As in our previous studies using expanded litters, not all rats survived. Twenty-four (96%) of the 25 4-day MMI rats, 30 (60%) of the 50 10-day MMI rats, and 28 (56%) of the 50 4-day MMI plus recovery rats survived (Table 1) . Twenty-four (96%) of the 25 4-day control animals and 23 (92%) of the 25 10-day control animals survived. All gradable retinas from surviving rats were included in the analyses. Seven (5.5%) of 128 ADPase-stained retinas were ungradable and were therefore excluded from analysis (1 from the 4-day MMI litter, 4 from the 10-day MMI litters, and 2 from the 4-day MMI plus recovery litters). 

Incidence and Severity of NV.

In the 10-day MMI litters, NV was observed in 8 (31%) of 26 retinas, with severity ranging from 1 to 3 clock hours when present (Table 1 , Fig. 1 ). The majority of affected retinas (7/8) had 1 clock hour of NV. In the 4-day MMI plus recovery litters, only 1 (4%) of 26 gradable retinas was positive for NV, with a severity of 3 clock hours. No retinas from the 4-day MMI rats or either control group were graded positive for NV. 

Retinal Vascular Areas.

The ratio of vascularized to total retinal area was reduced after 4 days of MMI treatment compared with that in 4-day control animals (36% ± 6% vs. 50% ± 6%, P = 0.0001). The retinal vascular areas were similar in the 10-day MMI rats (91% ± 3%), 4-day MMI plus recovery rats (93% ± 5%), and 10-day control animals (91% ± 4%). 

Serum IGF-1 Study

Serum IGF-1 levels were markedly reduced in the 4-day MMI rats (median, 42 ng/mL, quartiles, 21 and 49 ng/mL) compared with 4-day control animals (median, 133 ng/mL, quartiles, 112 and 168 ng/mL, P = 0.0001, Fig. 2 ). Similarly, 10-day MMI rats had significantly lower serum IGF-1 concentrations than did 10-day control animals (median, 133 ng/mL; quartiles, 126 and 143.5 ng/mL versus median, 206.5 ng/mL; quartiles, 175 and 256 ng/mL, respectively, P = 0.005). In the 4-day plus recovery rats, serum IGF-1 levels normalized to control levels by day 10 (median, 213.5 ng/mL; quartiles, 175 and 217 ng/mL; P = 0.99). 

Serum T4 Study

T4 levels in the 4-day MMI rats (median, <1 μg/dL; quartiles, <1 and 1.15 μg/dL) were not significantly different from the 4-day control animals (median, 1.16 μg/dL; quartiles, 1.14 and 1.25 μg/dL; P = 0.22). However, serum T4 concentrations in 10-day MMI rats were significantly lower than in 10-day control animals (median and quartiles, <1 μg/dL versus median, 1.89 μg/dL; quartiles, 1.44 and 2.5 μg/dL; P = 0.0084). Parallel to serum IGF-1, T4 levels in 4-day MMI plus recovery rats normalized to control levels by day 10 (median, 1.715 μg/dL; quartiles, 1.535 and 2.08 μg/dL; P = 0.99, Fig. 3 ). 

Animal Growth

Growth retardation was significant in all MMI-treated rats beginning at day 4 (P < 0.05 versus control animals for each day, Fig. 4 ). Weights in rats in the 4-day MMI plus recovery group normalized to control levels by day 10 (P = 0.99). 

Discussion

In the present study, we found that exposing neonatal rats to MMI, an anti-thyroid drug known to suppress serum IGF-1, ¹³ results in a 31% incidence of NV by day 10, which is preceded by significantly reduced vascularized retinal areas by day 4 of life. 

Previous studies have suggested an important role of IGF-1 in the normal maturation of retinal vasculature ¹⁴ and in the development of NV or ROP in cell culture, rodent models, and humans. Early studies showed that IGF-1 is a potent growth promoter of retinal endothelial cells and retinal pericytes. ¹⁵ Recent studies in IGF-1 knockout mice ¹⁶ showed that IGF-1 is critical for normal retinal vascular growth. Smith et al. ¹⁷ reported that IGF-1 plays an essential role in VEGF-induced NV in a neonatal mouse model of ROP. By systemically blocking the IGF-1 receptor, they were able to decrease retinal NV by 53%. In human premature neonates at risk for ROP, sustained low levels of serum IGF-1, followed by a subsequent increase, has been strongly associated with the development of ROP. ¹ Our finding of an MMI-induced NV, associated with early suppression of serum IGF-1 and subsequent recovery, is consistent with these previous studies. 

Our results differ from those of Berkowitz et al. ⁴ who reported no NV with MMI alone in contrast to an increased incidence and severity of retinal NV in an OIR model of ROP combined with MMI treatment. An explanation for this discrepancy may be some key differences in experimental design between our study and that of Berkowitz et al. In our study, rats were raised in expanded litters of 25. We have reported that rats raised in expanded litters are growth retarded compared with rats raised in standard litters of 10. ^{18 19} This growth retardation is probably caused by the increased competition for food, since nursing dams have only 12 nipples. In our study, most of the pups that died did so between days 4 and 6 of the 10-day experiments, allowing the surviving pups increased access to food and, therefore, improved nutrition in the second half of our 10-day studies. It has been well established that undernutrition results in reduced serum IGF-1 concentrations. ²⁰ It has also been reported in both animal ²¹ and human ^{22 23 24} studies that IGF-1 concentrations recover rapidly once nutrition is restored. It is possible, therefore, that in our study, once nutrition improved, the subsequent rise in IGF-1 (Fig. 2) , in the presence of high levels of VEGF in the avascular and presumably hypoxic retinas, may have provided the synergy necessary for the development of VEGF-mediated NV. This is consistent with the IGF-1 hypothesis proposed by Hellstrom et al. ¹ In contrast, Berkowitz et al. ⁴ used litters of 8 to 10 rats. Smaller litters, with no increased competition for food, may have less suppression of IGF-1 in early postnatal life compared with our expanded litters and thus less opportunity for a rapid relative increase in IGF-1 later on. 

Another significant difference between our study and that of Berkowitz et al. ⁴ is that we analyzed retinas at days 4 and 10 of life, whereas they analyzed the retinas at day 20. The major differences in retinal vessel growth in our study were observed at day 4, when we saw significantly retarded retinal vascular areas in MMI-treated rats versus control animals (Table 1) . In fact, our analyses showed no significant difference between vascular areas in MMI-treated retinas versus control animals by day 10 (i.e., >90% vascularized). It is likely therefore, that by 20 days, any differences in vascular development would no longer be appreciated. 

A further difference between our study and that of Berkowitz et al. ⁴ is the source of Sprague–Dawley rats. We received our rats from Harlan Laboratories and they from Hilltop Laboratories (Chatsworth, CA; Berkowitz BA, personal communication, July 2004). In oxygen-induced retinopathy, we have reported ²⁵ differences in incidence and severity of NV between neonatal Sprague-Dawley rats from difference vendors (Harlan versus Charles River, Wilmington, MA). We speculated ²⁵ that subtle genetic differences between rats from different vendors influence the predisposition to preretinal NV, despite similar insults. 

In our present study, continuous treatment with MMI for 10 days, suppressing IGF-1, resulted in retinal NV in 8 (31%) of 26 retinas and retardation of retinal vessel growth. In contrast, rats treated with MMI for 4 days, followed by a 6-day recovery period, had normalized levels of IGF-1 by day 10 and had almost no NV (only 1 of 26 retinas; 4%). This finding is intriguing, given that the total increase in serum IGF-1 from days 4 to 10 was much greater in rats receiving the short course of MMI followed by recovery than those who received continuous MMI for 10 days (Fig. 2) . These findings are contrary to the suggestions of Hellstrom et al. ¹ that IGF-1 plays a purely permissive role, because the IGF-1 increases were greater in the short-course–recovery group. One possible explanation for less NV in the short-course–recovery group is that IGF-1 levels in these pups may not have been depressed enough, and for a sufficient period, to suppress normal retinal vascular development and thus subsequently stimulate NV. However, results from our retinal vascular area studies (Table 1) suggest that 4 days of MMI treatment significantly retards retinal vessel development compared with 4-day control animals. We also found that at 4 days, IGF-1 is significantly suppressed in all rats treated with MMI (Fig. 2) . Further work on the role of serum IGF-1 in angiogenesis in immature retinas is needed. 

The thyroid hormone axis has an important role in the development of the central nervous system, including the eye ²⁶ and the retina. A recent study by Sevilla-Romero et al. ²⁷ showed substantial differences in the developing retinas of euthyroid rats compared with congenitally hypothyroid rat pups. Hypothyroid retinas were smaller, had reduced overall thickness, and had fewer dividing progenitor cells. Further, a marked delay in all main developmental parameters in the hypothyroid retinas was seen. In addition, Tilton et al. ²⁸ reported that hypothyroidism increases permeability of retinal vessels in rats, and thus may allow serum growth factors, such as IGF-1, increased access to the retina. Transient hypothyroidism is common in premature infants, and the more premature the infant, the more severe the transient hypothyroxinemia. ^{29 30} We speculate that low serum thyroxine may contribute to retardation of normal retinal vascular development, which may exacerbate the insult to the peripheral retina or developing vasculature and contribute to the subsequent development of preretinal NV (i.e., ROP in preterm infants). Although thyroid hormone supplementation in hypothyroid preterm infants remains controversial, ²⁹ further studies on the effect of thyroxine on the developing retinal vasculature are warranted. Our data support the suggestion that T4 plays an important role in abnormal angiogenesis in the immature retina, and we speculate that hypothyroidism, for a critical period in a neonate’s life, may be an additional risk factor for ROP. 

The changes we observed in serum T4 concentrations may provide an explanation for the paradox of less NV after a short course of MMI than a longer course, despite greater recovery of serum IGF-1. We found that rat pups treated continuously with MMI for 10 days had continued suppression of T4, whereas rats treated with a short course followed by recovery had normalized T4 levels by day 10. This leads to the hypothesis that suppression of T4 may be essential in the pathogenesis of NV in immature retinas. The complex interaction of the IGF-1 and thyroid hormone axis needs further investigation. 

Regarding weaknesses of our present study, the mortality rate in the MMI-treated rats after 10 days was not trivial. Nevertheless, in previous studies of neonatal rats raised in expanded litters, we observed similar mortality rates. In acidosis-induced retinopathy, rats receiving ammonium chloride ³¹ or acetazolamide, ¹⁰ had a very similar survival rate of 50% to 60% in 13-day experiments. Furthermore, it is possible that NV was actually underestimated, since the smallest and sickest rats may be more likely to have NV. Unfortunately, it is not possible to control for survival rate in these studies, and autolysis of retinal tissues precludes analysis of NV in rats that die during the course of an experiment. We did not perform postmortem examinations of neonatal rats that died before the conclusion of the study (often the mothers eat their dead pups). Therefore, we cannot completely rule out a possible toxic effect of the MMI, but, based on our findings of changes in serum T4 and serum IGF-1, we believe that our retinal findings are most likely to be a specific drug-induced NV. 

In summary, we have shown that the anti-thyroid drug, MMI, retards normal vascular development and induces NV in neonatal rats. These findings are associated with suppression of IGF-1 and T4, but the relationship is complex because complete recovery of IGF-1 is associated with less NV, and therefore serum IGF-1 must act in more than a permissive role. Further studies are warranted into the role of IGF-1 and thyroid hormone in the pathogenesis of ROP. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, April 2004.

 Supported by National Eye Institute Grant EY12798 (JMH); an unrestricted grant to the Department of Ophthalmology, Mayo Clinic College of Medicine, from Research to Prevent Blindness, Inc.; the Mayo Foundation, Rochester, Minnesota; and the philanthropy of Margaret Schroeder (JMH). JMH is a Research to Prevent Blindness Olga Keith Weiss Scholar.

 Submitted for publication June 9, 2004; revised July 23, 2004; accepted August 5, 2004.

 Disclosure: M. Mookadam, None; D.A. Leske, None; M.P. Fautsch, None; W.L. Lanier, None; J.M. Holmes, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Incidence of Neovascularization

Experimental Groups	Survival	Incidence of NV	Severity of NV in Affected Rats (clock hours)	Retinal Vascular Areas (mean % ± SD)
4-day MMI	24/25 (96)	0/23 (0)	N/A	36 ± 6
4-day MMI plus recovery	28/50 (56)	1/26 (4)	3	93 ± 5
10-day MMI	30/50 (60)	8/26 (31)	1–3	91 ± 3
4-day control	24/25 (96)	0/23 (0)	N/A	50 ± 6
10-day control	23/25 (92)	0/23 (0)	N/A	91 ± 4

 Data are the number/total group (%).

Figure 1.

 

Representative ADPase-stained retinal flatmounts from 10-day MMI rat (A) and 10-day control (B). Arrow: preretinal NV, which occurred in 31% of 10-day MMI rats. Bar, 100 μm.

Figure 2.

 

Serum IGF-1 from individual neonatal rat blood samples (n = 8–19 samples per group). In rats treated with MMI, serum IGF-1 was decreased compared with control animals at days 4 and 10. Rats treated with MMI for 4 days followed by 6 days of recovery had IGF-1 levels similar to control rats at day 10.

Figure 3.

 

Serum thyroxine (T4) concentrations from individual neonatal rat blood samples (n = 6–11 samples per group). In rats treated with MMI, serum T4 was decreased to below detectable levels at days 4 and 10. Rats treated with MMI for 4 days followed by 6 days of recovery had T4 levels similar to the control at day 10.

Figure 4.

 

Rat weights (mean ± SD) for all neonatal animals in 10-day experiments. Growth retardation was significant in all MMI treated rats beginning at day 4 (P < 0.05). Rats treated with MMI for 4 days followed by 6 days of recovery, normalized their weights to control levels by day 10. *Significantly different from room air control animals (P ≤ 0.05).