Immune checkpoint inhibitors and corneal transplant rejection: a call for awareness

Immune checkpoint inhibitors (ICI) have ushered in a new age of immuno-oncology and provided hope to patients with a host of malignancies. It is important to note, however, that while they can show breathtaking efficacy in retarding and even potentially reversing malignant growth, they may also carry the risk of immune-related adverse events. One particularly devastating potential immune-related adverse event is the risk of rejection in organ allografts, and it is precisely this risk in the eye to which we seek to draw attention.

The cornea, the pristinely transparent tissue at the anterior-most surface of the eye, was the first tissue successfully transplanted. Today, almost 50,000 corneal allografts are transplanted every year in the USA [1], and it remains the most commonly performed allotransplantation procedure in the world. Due to the unique form of immune privilege the eye enjoys, corneal allografts can be transplanted without the need for systemic immunosuppression; this contrasts starkly with other solid organ allografts. A critical component of corneal immune privilege is the immune checkpoint molecule PD-L1. Therein lies the potential for corneal transplant rejection with the use of systemic PD-L1 and PD-1 inhibitors, increasingly common agents employed by the oncology community.

PD-L1 is highly expressed in the cornea. Its role as a coinhibitory signal that downmodulates T-cell responses is well established. Corneal endothelium, the most posterior layer of the cornea, abutting the aqueous humor within the eye, expresses PD-L1 and thereby suppresses PD-1⁺ T-helper cells via a contact-dependent mechanism [2]. In the eye, this is one factor amongst many that contributes to corneal immune privilege. Others include soluble immunosuppressive factors in the aqueous humor, the expression of Fas ligand on the cornea, the cornea’s avascular structure, and a complex immunoregulatory phenomenon called anterior chamber-associated immune deviation, which suppresses delayed-type hypersensitivity in the eye [3]. Furthermore, the iris pigment epithelium can convert transiting effector T cells into regulatory T cells (Tregs) [4]. In turn, these iris pigment epithelium-induced Tregs inhibit bystander T cells in the anterior chamber, which is adjacent to the cornea, by a number of mechanisms. Importantly, one of these is via expression of PD-L1 and its ligation of PD-1 on T cells. Perhaps unsurprisingly then, animal models of PD-L1 or PD-1 blockade (or knockout) have shown high levels of corneal transplant rejection. When anti-PD-L1 antibodies are administered systemically to mice with corneal allografts, the rejection rate is over 90% [5]. Conversely, when PD-L1 fusion proteins are administered to mice with fully MHC-mismatched corneal allografts, graft survival is significantly prolonged compared with controls [6]. When PD-L1 is knocked out of the donor tissue and then transplanted into a wild-type mouse, the rejection rate is 80%. When a wild-type graft is placed in a PD-L1 knockout host, the rejection rate is 100%. Similarly, when antibodies against PD-1 are administered systemically to mice with corneal allografts, the rejection rate is also 100% [7].

Taken together, these results strongly suggest a high risk of corneal transplant rejection with the use of PD-L1 inhibitors. As of writing, two cases of corneal transplant rejection secondary to an ICI have been described. In the first, a woman with unresectable squamous cell lung cancer experienced rejection of her corneal graft after the ninth cycle of nivolumab, a PD-1 inhibitor. Aggressive systemic and local corticosteroid therapy failed to salvage the graft, leading to total corneal opacification [8]. In the second, an 85-year-old woman with bilateral corneal grafts, one transplanted 16 years prior and the other 6 years prior, developed bilateral, simultaneous graft rejection 3 months after starting pembrolizumab for metastastic urothelial cell carcinoma [9]. Of note, rejection occurred despite excellent compliance with once daily fluorometholone in each eye, a common topical, ophthalmic corticosteroid used to chronically suppress the alloimmune response. Furthermore, she was asymptomatic, and was only diagnosed on a routine clinical exam by her corneal surgeon. She was treated with intensive topical dexamethasone drops with good clinical effect, only for rejection to recur 8 weeks after tapering off topical therapy. Ultimately, ICI therapy was discontinued due to the high risk of bilateral blindness. The second case is both deeply worrisome and highly instructive. First, rejection developed despite chronic topical suppressive therapy. Second, the patient was asymptomatic, leading to concern that subclinical graft rejection could go undiscovered for lengthy periods of time, perhaps even to the point of graft failure. Third, the patient was not a recent transplant recipient but rather had one graft for nearly two decades and the other for 6 years without any issue or prior rejection episodes.

Because of the risk of rejection, ICIs are relatively contraindicated in any organ transplant recipient, and allografts have generally been an exclusion criterion in clinical trials of these drugs. Systematic reviews of ICI therapy in solid organ transplant recipients have found rates of rejection to be alarmingly high, ranging from 37 to 41% [10–13]. These cases mostly represent data from kidney, liver and heart transplant patients. PD-1 inhibitors were consistently associated with higher rates of rejection than CTLA-4 inhibitors. While median time from initiation of therapy to rejection was just under 1 month in two of these series [7,8], patients could experience rejection as late as 5 months after beginning ICI therapy. Furthermore, increasing time from transplant to initiation of an ICI was not protective, meaning patients with stable, decades-old grafts as well as those with recent grafts are all at risk. Unfortunately, once ICI-induced rejection arises, outcomes are poor; graft failure ultimately occurred in 81–83% of cases. These are sobering facts. Interestingly, of all the visceral organs, renal allografts have the highest rates of rejection. Much like the cornea, the renal tubular epithelium highly expresses PD-L1, thereby inhibiting a local alloimmune response [14]. Based on the immunobiology of corneal transplants as well as the cases above, there is good reason to posit that there is a real risk to corneal graft recipients also.

Notwithstanding the above arguments, given the paucity of clinical data at this time, we believe it is premature to propose that ICIs be contraindicated in corneal graft recipients. But we do propose that the risks be weighed appropriately in the complex decision-making process that oncologists face. Ophthalmologists are fond of saying ‘life trumps sight' in cases where management of life-saving systemic disease clashes with our preferred management of ocular disease. Given where we are with the available data to date, the risk of ICIs is most likely worth taking in someone with a corneal transplant in one eye, good vision in the fellow eye and metastatic melanoma failing current therapy that threatens to kill the patient in a few months. But what of the patient with bilateral transplants, as in the second case or monocular patients who have good sight only through one corneally grafted eye? For these patients, rejection could be devastating.

What then can we recommend to the medical community for patients who are potential recipients of immune checkpoint inhibitor therapy? First, patients should be asked if they have ever had a corneal transplant, and if so in one or both eyes. Of note, corneal transplant recipients are almost never on systemic immunosuppressive agents and thus oncologists cannot rely on the patients’ medication list to attune them to the presence of an allograft. So natural and conspicuous a cue will be absent here. This recommendation could easily be incorporated into standard questionnaires for new patients and should cause little-to-no inconvenience. Second, when these patients are identified, the potential risks of augmenting the alloimmune response need to be discussed explicitly, especially in the case of binocularly grafted patients or those who are monocularly sighted. Third, should the patient be initiated on immune checkpoint inhibitors, they need to be referred to their corneal transplant surgeon for close monitoring. Episodes of rejection can often be overcome with aggressive local treatment if diagnosed early. In general, close monitoring of the eyes for the first year of therapy would likely be a reasonable standard. Only a small fraction of the patients with corneal transplants will develop a malignancy that would warrant ICI therapy, so such a practice would hardly be burdensome to the ophthalmic community. Corneal transplant patients typically have life-long relationships with their corneal surgeons, so arranging for prompt ophthalmic consultation would likely be straightforward.

Last, it will be important to keep track of these cases and report their outcomes. Although the precise prevalence of corneal transplants in the USA is unknown, with an incidence of 50,000 per year, nearly a million cases is a plausible estimate. Thus, oncologists can expect to encounter this clinical scenario over their careers. It will take the combined efforts of oncologists and ophthalmologists to better define the risks to our patients, and to manage any complications when they arise.