Ocular Adverse Effects of Systemic Treatment With Isotretinoin

Objective To examine whether isotretinoin therapy could result in deleterious ocular effects, as previously described in case report studies.

Design Retrospective cohort study.

Setting The study was conducted using the electronic medical databases of a large health maintenance organization in Israel.

Patients The study population consisted of 14 682 adolescents and young adults who were new users of isotretinoin for acne and 2 age- and sex-matched comparison groups (isotretinoin-naive patients with acne and acne-free patients).

Main Outcome Measures Ocular adverse effects (AEs) or purchases of ophthalmic medications within 1 year after the first dispensed isotretinoin prescription.

Results In total, 13.8% of the isotretinoin group experienced ocular AEs vs 9.6% of the isotretinoin-naive group and 7.1% of the acne-free group. During a 1-year follow-up period, the isotretinoin group had significantly higher risk for any ocular AEs (hazard ratio, 1.70; P < .001) compared with the acne-free group. No such increased risk was observed for the isotretinoin-naive group. The isotretinoin group had higher relative risks for inflammatory and structural AEs.

Conclusion Isotretinoin use may be associated with short-term ocular events, especially conjunctivitis, underscoring the importance of educating patients and caregivers about these potentially important AEs of the therapy.

Isotretinoin is the drug of choice to treat severe acne vulgaris.¹ The drug is thought to improve all known etiologies of this common folliculo-occlusive phenomenon that affects mainly adolescents and manifests clinically in various cutaneous lesions. Ocular adverse effects (AEs) associated with isotretinoin use have been reported, resulting mostly from changes to the eyelids and the surface of the cornea or lacrimal abnormality that leads to dry eye. The association is documented in the literature, with case reports^2-4 describing blepharoconjunctivitis, keratoconjunctivitis sicca, cutaneous photosensitivity, contact lens intolerance, refractive changes, papilledema, pseudotumor cerebri, and abnormal retinal function.

In a comprehensive review of 2449 documented reports of ocular AEs associated with isotretinoin use and 38 different signs and symptoms, Fraunfelder et al⁵ and Fraunfelder⁶ classified the following AEs as certain (according to the World Health Organization classification of isotretinoin-induced ocular AEs): abnormal meibomian gland secretion, blepharoconjunctivitis, corneal opacities, decreased dark adaptation, decreased tolerance to contact lenses, decreased vision, increased tear osmolarity, meibomian gland atrophy, myopia, ocular discomfort, ocular sicca, photophobia, pseudotumor cerebri, and keratitis. All reported AEs were found to be reversible on discontinuation of isotretinoin therapy.

The few prospective studies^7-10 on AEs of isotretinoin use in patients with acne were based on small cohorts (<55). The most frequent AEs in these studies were blepharitis and conjunctivitis, and the primary associated finding was marked pathologic decrease in the breakup time. All abnormal findings in these studies were reversible shortly after cessation of the isotretinoin treatment.

While Fraunfelder et al⁵ surveyed various signs and symptoms from investigations of many isotretinoin users, none of the studies included a control group, introducing the possibility of confounding factors, such as acne-related general dermal effects. In addition, these investigations relied on spontaneous random reports rather than systematic follow-up analysis of recorded data. Furthermore, voluntary postmarketing spontaneous reporting systems are not reliable sources of information, limited by underreporting, incomplete data, and lack of follow-up analysis. The present retrospective cohort study was undertaken to assess the risk of ocular AEs in a large and unselected population of isotretinoin users, to describe the most common AEs, and to evaluate the period of highest risk.

The cohort was drawn from the databases of the Maccabi Healthcare Services (MHS), a health maintenance organization with 2 million members in Tel Aviv, Israel. The MHS computerized databases, described previously,¹¹ contain information on all dispensed community prescriptions, hospital discharge data, and results of laboratory investigations, recorded under a unique number for each member. The databases from January 1, 2000, through December 31, 2007, were surveyed to identify isotretinoin users and 2 groups of unexposed control subjects with and without reported acne. Institutional ethics committee approval was obtained. Patients with diagnoses of diabetes mellitus, cardiovascular diseases, cancer, or thyroid disorders before the index date (defined as the date of the first dispensed isotretinoin prescription) were excluded, as were patients who visited an ophthalmologist before the index date, to rule out preexisting known ocular disease or complaint.

Incident cases of ocular AEs were characterized by the onset date, defined as the date of the first diagnostic code associated with an ophthalmologic diagnosis occurring within 1 year after the index date. New users of isotretinoin were identified among all MHS members aged 14 to 20 years who between January 1, 2000, and December 31, 2007, had 2 or more dispensed prescriptions of isotretinoin at least 28 days apart. The recommended dosage of isotretinoin in Israel is similar to that in other countries (0.5-1.0 mg/kg/d), to a cumulative dose of 120 to 150 mg/kg. Isotretinoin was approved for use at a minimal copayment in the MHS in September 1997 (Roaccutane; Hoffman-LaRoche) and in January 2003 (Curatane; Douglas Pharmaceuticals Ltd).

Demographic variables at the index date included baseline values of age and sex. Socioeconomic status was defined by 1995 national census data according to the poverty index of the member's enumeration area (range, 0-20). The poverty index was based on several variables, including household income, education, crowding, material conditions, and car ownership. Patients with diabetes mellitus were identified by the MHS computerized registry of patients with diabetes mellitus.¹¹ Cancer history was provided by the Israel National Cancer Registry. Health services use was assessed by the number of visits to the primary care physician, which was based on data collected in the year before the index date. Information on study outcomes (ocular diseases) was based on data collected within 1 year after the index date. Ocular diagnoses by ophthalmologists were classified into the following 3 categories: inflammatory (eg, conjunctivitis, chalazion, hordeolum, blepharitis, etc), structural (eg, corneal opacities, subconjunctival hemorrhage, cataract, etc), and other (tear disorder and less likely AEs, such as glaucoma).

A χ² test for categorical variables and an analysis of variance for continuous variables were used to determine significant differences in baseline characteristics between the groups. A Cox proportional hazards model was used to estimate hazard ratios (95% CIs) for isotretinoin use, adjusting for acne and the following baseline values: age (in 1-year intervals), sex, socioeconomic status, and the number of visits to the primary care physician. This project received approval from the health maintenance organization ethical Helsinki committee for research involving humans.

After applying the inclusion and exclusion criteria, 14 682 new users of isotretinoin for acne were eligible for analysis. They were compared with 14 682 age- and sex-matched isotretinoin-naive patients having acne and with 14 682 age- and sex-matched acne-free patients.

Baseline characteristics of the 3 study groups are given in Table 1. The mean (SD) age of the study population was 16.55 (1.89) years. Similar age distributions were observed among the groups. Patients in the isotretinoin-naive group were more likely to be of lower socioeconomic status. Patients in the acne-free group had fewer visits to their primary care physician in the year before the index date. These differences in baseline characteristics were adjusted for in a multivariate model.

The sex-specific and total incidence rates of inflammatory, structural, or other ocular diagnoses are given in Table 2. Among the study population, inflammatory ocular diseases were diagnosed in 1791 patients (991 in the isotretinoin group, 446 in the isotretinoin-naive group, and 354 in the acne-free group). Structural ocular diseases were diagnosed in 268 patients (144 in the isotretinoin group, 70 in the isotretinoin-naive group, and 54 in the acne-free group). The observed attributable risk increases for the isotretinoin group vs the isotretinoin-naive group were 3.7%, 0.1%, and 4.2% for inflammatory, structural, and other ocular events, respectively. The observed relative risks for the isotretinoin group vs the isotretinoin-naive group were 2.23, 2.06, and 1.44 for inflammatory, structural, and other ocular events, respectively.

The most common ocular diagnoses were conjunctivitis, hordeolum, chalazion, and blepharitis. A subanalysis of the most frequent ocular disorders is summarized in Table 3. Other ocular disorders that did not reach statistical significance, possibly because of few cases, included corneal opacities (5 cases in the isotretinoin group, 4 cases in the isotretinoin-naive group, and 3 cases in the acne-free group), subconjunctival hemorrhage (0 cases in the isotretinoin group, 5 cases in the isotretinoin-naive group, and 2 cases in the acne-free group), cataract (1 case in the isotretinoin group, 5 cases in the isotretinoin-naive group, and 3 cases in the acne-free group), and tear disorder, which was a nonspecific diagnosis (9 cases in the isotretinoin group, 7 cases in the isotretinoin-naive group, and 1 case in the acne-free group).

Table 4 compares the ophthalmologic medication use rates within 1 year after the index date among the study groups, demonstrating an increased incidence in the isotretinoin group compared with the 2 other groups. A similar pattern was found for the frequency of ophthalmologic surgical operations.

The Figure shows the cumulative survival times to the first ocular event in the 3 study groups. The increased risk of the isotretinoin group vs the isotretinoin-naive and acne-free groups starts immediately from the first month and continues throughout the year, with a peak hazard at approximately 4 months after the index date.

Table 5 summarizes the associations between isotretinoin exposure and ocular AEs within 1 year after the index date, adjusted for several confounders. Isotretinoin use was associated with increased risk for the development of any ocular disorders (P < .001), with hazard ratios of 2.33 (95% CI, 2.06-2.64) for inflammatory AEs and 2.10 (95% CI, 1.52-2.91) for structural AEs. Socioeconomic status was not significantly associated with outcomes in any of the groups (data not shown).

The results of the present study demonstrate a strong association between isotretinoin therapy and the development of clinically meaningful ocular AEs, with the peak increased risk at 4 months after the first dispensed isotretinoin prescription. This is in agreement with previous much smaller investigations that studied various signs and symptoms of ocular abnormalities associated with isotretinoin use.⁵

The most common AEs attributed to isotretinoin use in our study were conjunctivitis, hordeolum, chalazion, blepharitis, eye pain, and dry eye, which is in line with other studies^3,7-9 that evaluated isotretinoin-induced ocular AEs qualitatively and quantitatively. Acute conjunctivitis, a common condition,¹² was the most frequent diagnosis in our cohort, occurring 1.7 times more frequently in the isotretinoin group compared with the isotretinoin-naive group. Isotretinoin therapy was associated with a strong increase in the risk of 3 other frequent diagnoses, namely, hordeolum, chalazion, and blepharitis.

These elevated risks can be attributed to the known biological effect of the drug to induce meibomian gland dysfunction (MGD). Mathers et al¹³ observed in patients that during isotretinoin use meibomian glands appeared significantly less dense and atrophic. Excretion thickness and tear osmolarity increased significantly, while Schirmer test scores did not change significantly.¹⁴

Further support for the hypothesis that isotretinoin therapy increases the risk of blepharitis due to MGD comes from the animal models by Lambert and Smith,¹⁵ who presented the first experimental evidence that systemic isotretinoin treatment affects the morphologic structure and shrinks the meibomian gland. Isotretinoin seems to inhibit the ability of the meibomian acinar cell to differentiate while stimulating the epithelium lining the ducts and acini to proliferate. These findings are compatible with those by Landthaler and colleagues,¹⁶ who demonstrated an inhibitory effect of isotretinoin on human skin sebaceous glands in which the differentiation of acinar cells in these glands is compromised. Rismondo and Ubels¹⁷ found in a rabbit model that isotretinoin did not affect lacrimal gland function, ruling out toxic effects of isotretinoin on the gland; Schirmer test scores increased significantly (in contrast tothe findings by Mathers et al¹³), suggesting the involvement of another mechanism promoting eye irritation. Meibomian gland function depression induces increased evaporation and tear film instability, causing drying and secondary irritation of the conjunctiva and cornea. In addition, the presence of isotretinoin and its metabolites in the tear film may directly irritate the ocular surface.

Meibomian gland dysfunction is a pathologic condition that has been associated with various ocular surface abnormalities, including blepharitis, conjunctivitis, rosacea, Sjögren syndrome, and contact lens intolerance.^13,18 Meibomian glands secrete lipids into the tear film, forming a superficial lipid layer that stabilizes it. Abnormalities in these glands cause instability in the tear film, resulting in chronic irritation of or damage to the ocular surface epithelium.¹⁹ Hyposecretion of meibomian lipids resulting from obstruction of the gland orifices is the most common abnormality in MGD²⁰; it causes increased evaporation of tears and has been identified as one of the major causes of keratoconjunctivitis sicca.²¹

In 2011, MGD was defined by the International Workshop on Meibomian Gland Dysfunction²² as a chronic diffuse abnormality of the meibomian glands, commonly characterized by terminal duct obstruction or changes in the glandular secretion that may result in alteration of the tear film, symptoms of eye irritation, clinically apparent inflammation, and ocular surface disease. Their new classification system divides MGD into 2 major categories, namely, low-delivery states and high-delivery states. Low-delivery states are further classified as hyposecretory or obstructive. On the basis of the proposed classification, MGD secondary to medication use (isotretinoin in this case) would be considered low-delivery hyposecretory. Therefore, MGD induced by isotretinoin use may well explain the high incidence of conjunctivitis, blepharitis, and dry eye among these patients. The high incidence of blepharitis may in turn explain the high incidences of hordeolum and chalazion, which are related to blepharitis.²³

Our finding that patients in the isotretinoin group were more likely to require ophthalmic medications (19.4% of patients) than those in the isotretinoin-naive group (9.8% of patients) or the acne-free group (5.3% of patients) further supports the contribution of isotretinoin to ocular AEs. The burning, itching, lid inflammation, and conjunctival and corneal changes associated with blepharitis, hordeolum, and conjunctivitis result in discomfort and pain that drive the patient to seek treatment.

Capturing AEs associated with the use of a drug during its postmarketing phase is complicated by the lack of funding allocated to the task, the high costs involved, the absence of proper follow-up analysis, the incompleteness of information, and the possibility of underreporting. The present study, which was designed to overcome these obstacles, has several advantages over previous attempts to quantify AEs, including a large and unselected study population, a long follow-up period, a comparison with 2 matched reference groups to reduce potential biases, and the reliance on systematically collected data as opposed to anecdotal case reporting. These advantages may explain the higher incidence of ocular AEs associated with isotretinoin use in our study compared with the aforementioned studies. Moreover, we believe that the incidence of isotretinoin AEs in our study was actually higher and may include more severe diagnoses such as papilledema, pseudotumor cerebri, and abnormal retinal function. These were not captured in our study, probably because the patients had been referred for ophthalmic or neurological investigations by specialists other than ophthalmologists.

Our finding of an increased incidence of ocular AEs among female patients is in line with the known greater use of health care services by women.²⁴ The calculated elevated risk of ocular AEs related to isotretinoin exposure was adjusted for several important variables and is unlikely to be confounded by a methodological effect.

Some limitations of the study should be addressed. It is known that individuals with acne experience more ocular diseases than the general population due to the disease itself, and it is reasonable to assume that the ocular disturbance is directly associated with the degree of acne severity. The isotretinoin group likely had more severe acne than the isotretinoin-naive group, which would translate into a higher rate of ocular diagnoses among them, regardless of the treatment. We tried to eliminate this confounder by excluding patients already treated by ophthalmologists before the index date. Another limitation is the lack of data on contact lens use. Wearing contact lenses could lead to complications and influence the association between isotretinoin use and ocular diseases. However, this is most likely a nondifferential information bias among the study groups that would lead to underestimation of the true ocular AEs of isotretinoin therapy.

We had no information on dechallenge and rechallenge, which could have provided a more precise temporal relationship between AEs and isotretinoin use. Our protocol was in line with other investigations in which the World Health Organization classification was used to help determine the association of reported AEs with isotretinoin use.⁵ Nevertheless, the reporting of AEs to the medical services by our isotretinoin users enabled us to establish a general timing of those events from the beginning of treatment, giving a much more accurate assessment of the ocular effects of the drug than studies in which no such events were recorded.

The study results underscore the importance of primary and secondary prevention measures. Dermatologists and primary care physicians who prescribe isotretinoin are urged to also prescribe ocular lubricants (eg, preservative-free artificial tears) as a primary preventive measure. When patients taking isotretinoin are seen with ocular problems, ophthalmologists should ascertain the timing of the onset of symptoms and consider discontinuation of the drug if the symptoms progress or persist despite treatment. A follow-up visit to the ophthalmologist should be scheduled about 4 months after the first dispensed isotretinoin prescription. To ensure early detection and treatment, it is important that patients and caregivers should be informed of the potential ocular AEs associated with isotretinoin use.

Correspondence: Gabriel Chodick, PhD, Maccabi Healthcare Services, 27 Ha’Mered St, Tel Aviv 68125, Israel (hodik_g@mac.org.il).

Accepted for Publication: January 21, 2012.

Published Online: April 16, 2012. doi:10.1001/archdermatol.2012.352

Author Contributions: Ms Goldshtein had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Neudorfer, Shamai-Lubovitz, Chodick, Dadon, and Shalev. Acquisition of data: Goldshtein. Analysis and interpretation of data: Neudorfer, Goldshtein, Chodick, and Shalev. Drafting of the manuscript: Goldshtein and Chodick. Critical revision of the manuscript for important intellectual content: Neudorfer, Shamai-Lubovitz, and Shalev. Study supervision: Chodick and Shalev.