We found that three among the 19 patients (15.79%) on vandetanib, one among the 202 patients (0.50%) on osimertinib, and all five patients (100%) on ABT-414 showed vortex keratopathy. All patients except those deceased shortly after the initiation of chemotherapy complained of decreased visual acuity due to corneal epithelial changes. While old drugs showed a low incidence of corneal change causing vision deterioration, more recently used novel drugs in clinical trials revealed obviously high incidence rates of ocular side effects. Fortunately, full recovery of both vision and the cornea was confirmed in seven patients (Table 3).
Vortex keratopathy is a condition characterized by a whorl-like pattern of corneal deposits in the corneal epithelium. Well-known causes of the disease are amiodarone use or Fabry’s disease, which is also known as lysosomal storage disorder [5]. Beyond amiodarone, other drugs that are known to cause vortex keratopathy include vandetanib and osimertinib [5]. There have been few case reports of vortex keratopathy after EGFR inhibitor chemotherapy reported to date. Ahn et al. reported a case of vortex keratopathy after six one-month cycles of 300 mg/day of vandetanib [10]. Chia et al. reported vortex keratopathy presumed to develop eight months after the use of AZD9291, a third-generation tyrosine kinase inhibitor (TKI) currently known as osimertinib. Both case reports did not address the prognosis of the corneal lesions [12]. We reviewed a tertiary hospital database for patients who used any kind of EGFR or FGFR inhibitors. Although there could be some variation depending on the geographic region, country, and institution, we calculated the incidence of vortex keratopathy for each of the drugs. Furthermore, we confirmed that the keratopathy cleared or at least improved after discontinuation of the offending agents.
Drug-induced vortex keratopathy is typically associated with cationic and amphiphilic medications that penetrate lysosomes and bind to cellular lipids. It is believed that drug–lipid complexes are the intra-lysosomal inclusion bodies observed in the basal layers of the corneal epithelium [10, 13]. The mechanism of vortex keratopathy with EGFR receptor inhibitors was thought to be either deposits of chemotherapy agent-derived metabolites in the cornea or abnormal turnover or migration of corneal epithelial cells due to the inhibition of corneal EGFR [5]. Recently, interactions between vandetanib and lysozymes and their characteristics were described, providing support for the previously suggested role of drug metabolites [14].
Strikingly, every patient who used ABT-414 showed vortex keratopathy. This is quite alarming given that there were three cases out of 19 patients on vandetanib and one out of 192 patients on osimertinib in comparison who showed the same. Moreover, in one case, vortex keratopathy developed within only 22 days after the first infusion of the drug. In the AbbVie study, patients received prophylactic steroid eye drops three times a day for a week starting two days prior to each infusion and continuing for four days after. According to the phase I study of ABT-414, patients showed dose-related ophthalmologic toxicity outcomes such as dry eyes, blurry vision, eye pain, photophobia, watery eyes, and findings of microcyst development within the cornea. Although it is unclear whether this microcystic development refers to vortex keratopathy, the article suggested that steroid eye drops could be used to help reduce the incidence and severity of those side effects [6]. Despite previous reports on the usefulness of prophylactic steroid eye drops in reducing the incidence of ocular side effects from ABT-414 [6,15], this treatment did not seem effective in preventing the development of vortex keratopathy in our study, since all five patients with ABT-414 had developed such corneal lesions.
ABT-414 is an antibody-drug conjugate (ADC) consisting of three components: an EGFR-targeting humanized monoclonal antibody, a potent microtubule agent (monomethyl auristatin F [MMAF]) and a noncleavable maleimidocaproly linker that connects MMAF to the antibody [7]. In clinical development, ABT-414 was designed to limit binding to wild-type EGFR [15]. Therefore, MMAF is supposed to have a lower cytotoxicity, attenuated potency, and improved aqueous solubility as compared with its uncharged counterpart, monomethyl auristatin E (MMAE) [16].
However, the reduced toxicity does not appear to extend to ocular toxicity. According to a review article on ADCs with MMAF and MMAE published in 2015, there were five cases of corneal microcystic epithelial changes and four cases of corneal deposits or inclusions [17]. Interestingly, among the list of ADCs associated with ocular side effects, four of 13 used MMAF as a cytotoxin. However, none of the ADCs employing MMAE as a cytotoxin were associated with ocular side effects [17].
MMAF, an antimitotic auristatin derivative with a charged C-terminal phenylalanine residue [16], is a microtubule inhibitor that induces apoptosis in cells undergoing mitosis. However, a recent study showed that it may also disrupt nondividing cells in interphase.[15] Although it is not clear as to why MMAF preferentially disrupts corneal cells, the ocular toxicity of the substance seems to have promoted more instantaeneous formation of vortex keratopathy following dosing with ABT-414.
Despite the extensive number of medical records of this study, conventional EGFR inhibitors such as erlotinib, geftinib, afatinib, and cetuximab were not linked with corneal epithelial changes in any cases. There were few patients who had chart records of simple punctate epithelial erosions and none of these had evidence of vortex keratopathy. Although the mechanisms are not clear, the latest EGFR inhibitors seem to affect the corneal epithelium more directly than conventional drugs.
Although we could not calculate the accurate incidence of vortex keratopathy following the use of EGFR inhibitors as chemotherapy due to the nature of this study, the condition does not seem uncommon when dealing with recently developed agents. Vandetanib is a second-generation EGFR inhibitor and osimertinib is a third-generation EGFR inhibitor that can target T790M and EGFR TKI-sensitizing mutations while sparing wild-type EGFR.[18] ABT-414 is an investigational compound. Quite a few patients developed vortex keratopathy after using these three chemotherapy agents. There were also some new or investigational drugs that did not promote vortex keratopathy in patients such as olmutinib, naquotinib, rociletinib, AZD-3759, JNJ-61186372 and regorafenib. However, there is a possibility that lesions in conjunction with the use of these drugs could be found in a much larger cohort.
On the other hand, we suspected corneal dysmaturation in patients who used FGFR inhibitors. Corneal dysmaturation is a benign and indolent condition that leads to a frosted corneal epithelium or individual islands of opalescent epithelium. Fibrovascular corneal pannus is not present [19]. Unlike vortex keratopathy, corneal dysmaturation is scarcely reported in the literature, which complicated our ability to obtain evidence. Furthermore, although histological findings are important for accurate diagnosis, we did not have access to samples of corneal lesions since this was a retrospective study. Nevertheless, the opacification pattern of patient with FGFR inhibitor agent was different from that of vortex keratopathy and was clearly distinguishable upon comparison with that of EGFR inhibitors. As there is no supporting evidence of histopathologic specimens among our patients, we cannot rule those with keratopathy after FGFR inhibitor chemotherapy as having corneal dysmaturation. However, there have not been any reports regarding corneal changes associated with the use of FGFR inhibitors to the best of our knowledge. Therefore, our findings are meaningful that we found out the clinical features were different from vortex keratopathy. It would be unwise to draw conclusions from our study that actual incidence rates of this corneal epithelial change after FGFR inhibitor treatment are as high as we suggested since the total number of patients included was too small.
We also reviewed the records of NSCLC patients on regorafenib, a multikinase inhibitor that blocks FGFR1 and 2 and various other receptors, and pazopanib, a multikinase inhibitor that blocks FGFR, VEGR, and platelet-derived growth factor receptor, but did not find any cases similar to those of the three patients on FGFR inhibitors. We hypothesize that selective strong affinities of ASP-5878 and FPA-144 toward FGFR affected corneal changes in the three patients. Since we could not find any report describing ophthalmological findings following FGFR inhibitor chemotherapy, further research on the influences of FGFR inhibitors on the corneal epithelium is needed.
It is difficult to define the exact time lapse between the start of chemotherapy and the appearance of corneal epithelial changes because not all patients underwent regular follow-up examinations during and after chemotherapy. However, it is clear that these corneal lesions are able to develop quite fast, suggesting the possibility that previously reported cases might have been diagnosed far later than at the time of the actual onset of keratopathy. One of our cases on vandetanib developed vortex keratopathy 91 days after the first chemotherapy session, which is much quicker than previously reported. Separately, one case on ASP-5878 took only 55 days to develop a corneal lesion, while two patients on FPA-144, an enhanced monoclonal antibody against FGFR2b, took two months. Furthermore, nearly everyone in the ABT-414 group took less than one month to develop vortex keratopathy.
All patients who developed corneal epithelial changes were treated with recommended regimens as follows. According to the FDA of United States, the recommended dose of osimertinib is 80 mg in tablet form once a day until disease progression or unacceptable toxicity. In ABT-414, the recommended regimen is 1.25 mg/kg via intravenous infusion every two weeks over 30 to 40 min [6, 20]. In clinical studies of vandetanib, patients received vandetanib 300 mg once daily [21, 22]. In clinical trials of ASP-5878, varying oral doses of 2 mg twice daily to 20 mg twice daily were given to patients [8]. In our center, 12 mg twice daily was given orally. The recommended dose of FPA-144 is 15 mg/kg given by intravenous infusion every two weeks [23]. Although it is not clear, it seems that there are certain accumulated doses of drugs that evoke corneal epithelial changes as patients complained of decreased visual acuity after the lapse of certain length of time. As seven patients achieved full recovery and three showed partial recovery from corneal epithelial changes after discontinuation of the drugs, the decrease of accumulated dose of chemotherapy agents might be important for the prognosis of corneal lesions.
It is important to check which other chemotherapy agents the affected patients in the study took before receiving EGFR or FGFR inhibitors elucidate whether these agents had any influence in the development of corneal epithelial lesions. Table 4 presents prior chemotherapy agents used before EGFR and FGFR inhibitor therapy in the affected patients. First, considering geftinib and afatinib, we did not find evidence of vortex keratopathy in this study as we discussed and there was no report of such among the literature. Second, we reviewed the literature dealing with ocular side effects of the listed chemotherapy agents in Table 4. Although quite a few agents listed above had a variety of ocular side effects, none of them had any evidence specifically dealing with corneal epithelial changes (Table 5). Moreover, vandetanib and osimertinib already have collected a few reports between them on the topic of corneal epithelial changes. Therefore, we concluded it is reasonable to believe that the observed corneal epithelial changes that occurred in 12 patients in this study were induced by the EGFR or FGFR inhibitors highlighted.
Two cases on vandetanib and one case on osimertinib experienced a delay in reaching a diagnosis of ocular complications. However, this delay may have been due to late consultation with the ophthalmology department, as these patients did not have regular ophthalmologic examinations before they complained of decreased visual acuity. In our study, among 6,871 patients, only 16.89% had any record of ophthalmologic examinations. Furthermore, only one-third of these participated in a consultation or follow-up visit associated with their chemotherapy treatment. The rest of the medical records did not relate to the patients’ chemotherapy.
Lack of knowledge of corneal epithelial changes after the use of EGFR or FGFR inhibitors among clinicians may cause them to disregard the importance of consultation with the ophthalmology department before, during, and after chemotherapy. Moreover, visual impairment may not be considered important when compared with other systemic side effects of chemotherapy. Therefore, we are unable to estimate how many patients may have experienced symptoms without referral or diagnosis.
Furthermore, due to the nature of the study, patients were detected by retrospective chart review, which allows for possible selection bias. Subtle symptoms and clinical findings may have been caused by these agents yet may have been described as “puntate epithelial erosions or dry eyes,” which were not included as cases. Therefore, the percentages of patients affected by a given agent do not accurately summarize the true prevalence of ocular toxicity associated with that agent. In other words, the actual incidence of corneal epithelial changes after EGFR or FGFR chemotherapy could have been underestimated.
Another limitation of our study is a possibility that actual causes of corneal epithelial changes could be more complicated. The drugs discussed here are varied, with different mechanisms of action. Some results could be off-target effects or could be affected by other drugs that the patient took during chemotherapy. However, vortex keratopathy associated with osimertinib, vandetanib, and ABT-414 was already reported in previous articles. Also, ASP-5878 and FPA-144 are not multi-TKI. Therefore, the results of our study, although somewhat inaccurate, have significance in that we examined long-term follow-up records and confirmed a full recovery of keratopathy in many cases.
In our study, steroid eye drops were not effective in preventing corneal epithelial changes after treatment with ABT-414. Nevertheless, all patients showed improved visual acuity and corneal surfaces after discontinuing the drug. Reardon et al. also pointed out that, once ABT-414 treatment was held or discontinued, ocular symptoms gradually resolved spontaneously in the majority of patients. They speculated that the corneal epithelium would regenerate after discontinuation of ABT-414, eliminating microcysts caused by the drug [6]. Therefore, considering the possible side effects of steroid eye drops, it is wise not to use them for prophylaxis. Overall, patients undergoing chemotherapy with EGFR or FGFR inhibitors should be educated about the possibility of corneal epithelial changes that reduce visual acuity and reassured that the condition is generally reversible after the end of treatment.