Immune checkpoint inhibitors in challenging populations

Abstract

Immune checkpoint inhibitors, including those targeting the PD-1/PD-L1 and CTLA-4 pathways, are revolutionizing cancer therapeutics. Both activity and toxicities largely stem from unleashing tumor- or host-specific cytotoxic T cells. Many patients seen in routine clinical practice did not qualify for, or were seriously underrepresented in immune checkpoint inhibitor clinical trials. Thus, a major gap in knowledge regarding the safety and efficacy of these agents persists in many populations, even following regulatory approval. To address this challenge, we aggregated and synthesized the available pre-clinical and clinical data surrounding immune checkpoint inhibitor therapy in challenging clinical populations to assist treatment decision making for both academic and community oncologists. Specifically, we focus on the safety and activity of immune checkpoint inhibitors in patients with autoimmune disorders, post-organ transplant, chronic viral infections, ongoing immunosuppressant use, organ dysfunction, pregnancy, brain metastases, extremes of age, and impaired functional status.

Introduction

Agents that block the interaction between programmed cell death-1 and its ligand (PD-1/PD-L1), and inhibit cytotoxic T lymphocyte antigen-4 (CTLA-4) are transforming the therapeutic landscape in oncology. These so-called immune checkpoint inhibitors target these key immune regulatory pathways and thereby unleash restrained T cell mediated anti-tumor responses. Anti-PD-1/PD-L1 directed therapies have now received regulatory approval in melanoma, nonsmall cell lung cancer (NSCLC), renal cell carcinoma (RCC) and head and neck squamous cell carcinoma (HNSCC). Ipilimumab (anti-CTLA-4) has a more narrow scope of activity as a single-agent, with regulatory approval only in melanoma. However, anti-CTLA-4 therapies may augment the activity of anti-PD-1 in melanoma and other cancer types, thus resulting in more widespread use.

Immune checkpoint inhibitors are appealing treatment options for patients and clinicians for several reasons. First, they have broad activity, demonstrating response rates ranging from 15% to 90% in over 10 different cancer types.¹ Second, they frequently induce durable disease control. Nivolumab, for example, has now been associated with a 34% 5-year overall survival rate in advanced melanoma, with similar durability observed in other cancers. Third, immune checkpoint inhibitors generally have favorable toxicity profiles (particularly using anti-PD-1/PD-L1 monotherapy). Although immune related adverse events (irAEs) may infrequently cause substantial morbidity and even mortality, many patients experience excellent quality of life with minimal symptoms while on therapy. Identifying reliable predictive biomarkers of efficacy and particularly toxicity has been a major challenge.

The safety and activity of immune checkpoint inhibitors has been well-characterized in numerous clinical trials. The average oncologist’s patient population, both in community and academic practices, however, is frequently comprised of many patients who would have been ineligible for these seminal clinical trials. Such “trial-ineligible” patients may now desire treatment, and, in our experience, this presents an extremely common source of confusion for both academic and community oncologists alike.

Several small studies have begun to explore the safety and efficacy of these agents in excluded or underrepresented populations, including those with dysregulated immune activation (pre-existing autoimmune diseases or hematopoietic/solid organ transplant), compromised immune function (long-term immunosuppression, chronic viral infections), and significant medical co-morbidities (organ dysfunction, old age, brain metastases). Despite these early efforts, there remains substantial uncertainty surrounding the safety and efficacy of anti-PD-1/PD-L1 and anti-CTLA-4 in these populations. Herein, we synthesize the current data to facilitate appropriate utilization of these novel therapeutics.

Autoimmunity

Dysregulated immunity mediates autoimmune disorders such as inflammatory bowel disease, autoimmune hepatitis, Guillain-Barre syndrome, etc. The hallmark toxicities of immune checkpoint inhibitors, irAEs, result from aberrant activation of autoreactive T cells against host tissues. Clinically, irAEs recapitulate or closely resemble various autoimmune disease. Although most irAEs resolve with corticosteroid administration, expectant monitoring, and/or hormone replacement, fulminant events occasionally lead to severe morbidity or even mortality.²

Naturally, the mechanism of action of immune checkpoint inhibitors led to fears that further immune stimulation would lead to clinically unacceptable immune activation in patients with pre-existing autoimmunity, in the form of underlying symptom flares or new autoimmune manifestations. Pre-clinical data supported these concerns, as CTLA-4 deficient mice succumbed to fulminant autoimmune activation with multi-organ involvement and a diffuse lymphoproliferative process.³ PD-1 knockout mice also developed immune mediated myocarditis (at least in the BALB/c mouse model). Additional pre-clinical and gene association data have also suggested that CTLA-4 and PD-1/PD-L1 axes may play some role in autoimmune disorders, although the precise roles have not been fully elucidated.^{4, 5} Thus, patients with active autoimmune disease were excluded from all clinical trials. This population, however, represents 20 – 50 million people in the United States alone. One study using Medicare data demonstrated that a full 13.5% of lung cancer patients had a concurrent diagnosis of an autoimmune disease, suggesting the urgency of exploring this population.⁶

To begin to address this question, our groups aggregated 30 patients with melanoma who had pre-existing autoimmune disease that received treatment with ipilimumab. Disorders included inflammatory bowel disease, rheumatoid arthritis, multiple sclerosis, and several others; 20% of patients were receiving steroids or other immunosuppressants at baseline. In this cohort, 27% experienced exacerbations of their autoimmune disease and 33% experienced conventional irAEs requiring treatment.⁷ The events that occurred resolved quickly with standard corticosteroid treatment algorithms, with the exception of one patient with psoriasis who died of colitis. This case was at least partially attributed to a long delay in seeking care and presentation following the onset of hypotension and multi-organ failure from volume depletion. Several patients with inflammatory bowel disease had low-grade disease flares, but were manageable with corticosteroids. An objective response rate of 20% was observed, consistent with ipilimumab clinical trials.

We also assessed 52 patients with pre-existing autoimmunity treated with anti-PD-1. Of these, 30% experienced an autoimmune flare (although only 4% required treatment discontinuation), and 29% experienced a classical irAE (8% discontinued treatment).⁸ Interestingly, patients with rheumatologic disorders (rheumatoid arthritis, lupus, psoriasis) tended to have low-grade disease exacerbations, whereas no patients with gastrointestinal or neurologic disorders experienced flares. Again, these irAEs and autoimmune flares responded to standard treatment algorithms, and no patients died of treatment related events. Of note, 33% of patients responded to treatment, again largely consistent with anti-PD-1 therapeutic trials.

Thus in our view, treatment with either anti-PD-1 or ipilimumab is feasible for patients with pre-existing autoimmunity, particularly in view of progressive, metastatic cancer. One caveat to these data are that our cohorts comprised patients for which clinicians were willing to give immune checkpoint inhibitors, and may not have included patients with the most severe autoimmune disease. We also would have great hesitancy using combination immune checkpoint blockade (e.g. ipilimumab and nivolumab) in patients with clinically significant autoimmune disease. Still, underlying autoimmune disorders are an important consideration and require close monitoring, but do not pose an absolute contraindication to treatment.

Transplant

Patients who undergo solid organ or hematopoietic stem cell transplant require fine-tuned modulation of immunosuppression to maintain allograft tolerance and prevent rejection or graft-vs-host disease (GVHD). Interestingly, pre-clinical data suggest that the PD-1/PD-L1 axis may be particularly critical to maintain organ tolerance, and PD-1 gene polymorphisms are associated with superior graft survival.^{9, 10,11} The relationship appears complicated, however, as high expression of PD-1/PD-L1 on various T cell subsets has been associated both with inferior survival and lower rates of GVHD following allogeneic stem cell transplant.^{12, 13} Unfortunately, patients also develop cancer following transplant at a high rate, at least in part due to their chronic immune suppression, and often need systemic therapies. Immune checkpoint inhibitors in this setting are considered exceedingly high risk to produce catastrophic post-transplant complications by breaking immune tolerance.

Several prospective trials and case reports, however, suggest that these agents may be tolerated in some patients. Ipilimumab, a generally more toxic drug than anti-PD-1/PD-L1 agents, appears to have superior safety in this setting, surprisingly. An initial phase I study showed excellent tolerance of a single dose of ipilimumab (ranging from 0.1mg/kg to 3mg/kg) in patients with relapsed hematologic malignancies following allogeneic stem cell transplant without any episodes of GVHD in 29 patients.¹⁴ A follow up study of ipilimumab 3mg/kg or 10mg/kg for four doses was conducted in a similar population of 28 patients with relapsed hematologic malignancies following stem cell transplant. At these doses, ipilimumab induced dose-limiting graft vs. host disease (GVHD) in 14% of patients and irAEs in 21% (including one death), but also induced several durable disease responses.¹⁵ Several other case reports of advanced melanoma treated with ipilimumab following solid organ or stem cell transplant have suggested reasonable tolerance with only a single episode of rejection in a renal transplant patient.^16–21

Anti-PD-1/PD-L1 has also been used safely in several patients following hematopoietic stem cell transplant.^{16, 22–26} However, several cases of renal and cardiac allograft rejection, and serious GVHD have also been reported soon after commencing anti-PD-1 in patients with various advanced malignancies.^{17, 27–33} In particular, based on these published case reports, renal transplant patients seem to be at high risk of rejection. It is unknown, however, how many patients have been treated without complications. One could speculate that the risk of rejection may be greater in allografts that are less well-matched, have been in situ for a shorter period of time, and require higher doses of immunosuppression. We are currently conducting a more systematic multi-center effort to characterize the safety of these agents and risk factors for toxicities. Needless to say, immune checkpoint inhibitors carry a high risk of graft rejection, and should be used with extreme caution and in a multidisciplinary setting in patients following transplant with full disclosure surrounding the risks.

Chronic Viral Infections

Hepatitis B, hepatitis C, and HIV infections have also been near-universal exclusion criteria for immune checkpoint inhibitor trials. These chronic infections may suppress T cell function and could theoretically compromise efficacy (particularly in the case of severe HIV/AIDS with low CD4+ T cell counts). Interestingly, there is pre-clinical and limited clinical data suggesting these agents could assist in viral clearance in infected patients.^34–38 In fact, therapeutic trials to test these agents in HIV or hepatitis C have been conducted (NCT02028403, NCT00703469), although the results have yet to be published. The role of immune checkpoint inhibitors in the management of chronic viral infections, particularly in view of numerous effective anti-viral therapy options, is unclear.

Very little experience with immune checkpoint inhibitor therapy in patients with advanced cancer is available at this time. Initial data from a trial of nivolumab in patients with hepatocellular carcinoma (Child-Pugh score ≤B7) included patients with chronic hepatitis B or C. This study demonstrated a 10% rate of grade 3/4 liver enzyme elevation, but reported an overall manageable safety profile.³⁹ Responses to treatment were observed in patients with hepatitis B and C. Currently, a phase III study comparing nivolumab with sorafenib in patients with advanced hepatocellular carcinoma is underway (NCT02576509), and will likely provide more insights into the safety and efficacy of nivolumab in this setting. Another ongoing trial is evaluating pembrolizumab in patients with HIV and advanced cancers; this study allows a variety of metastatic cancers and requires CD4+ T cell count of >200 (NCT02595866). Pending the results of these trials, we do not view that HIV or hepatitis C is a contraindication to treatment with anti-PD-1. Patients with HIV/AIDS and low CD4+ T cell counts should be monitored closely both for activity and for immune reconstitution-type phenomenon, particularly if on concurrent anti-retroviral therapy.

Chronic immunosuppression

Immunosuppressants could potentially hinder immunotherapy responses by impeding various facets of T cell function, including activation and effector function. Accordingly, ongoing high-dose corticosteroid or other immunosuppressant use has excluded patients from immune checkpoint inhibitor trials. Thus, a paucity of experience exists regarding the safety and efficacy of these agents in patients receiving chronic immunosuppressants. By contrast, extensive experience surrounding the use of replacement (physiologic) dose corticosteroids exists, given the frequency of primary or secondary adrenal insufficiency experienced with ipilimumab; activity appears equivalent in this population.^{40, 41}

In patients receiving supra-physiologic doses of steroids, our study of patients with prior autoimmunity suggested that responses may be less frequent in patients receiving high-dose steroids or other disease modifying therapies (15%) than in those not requiring these agents (44%).⁸ Further, a study of ipilimumab in patients with metastatic melanoma with brain metastases showed that patients rarely experienced responses when requiring corticosteroids (although this could represent a more aggressive and refractory disease course).⁴² Thus, we attempt to wean patients to replacement doses of corticosteroids (≤10 mg of prednisone daily or equivalent) prior to starting immune checkpoint inhibitors whenever possible.

Organ dysfunction

A vast number of patients with cancer have medical co-morbidities, including chronic organ dysfunction. For example, an estimated 10% of the U.S. population has chronic kidney disease, a number which is undoubtedly higher in the cancer population (particularly in patients with genitourinary malignancies). While no clear contraindication to immune checkpoint inhibitors exists for patients with renal, hepatic, or cardiac dysfunction, these patients were largely excluded from clinical trials, presumably due to difficulties in characterizing safety signals. In contrast to many cytotoxic chemotherapeutics and biologic agents, which undergo hepatic or renal clearance, immune checkpoint inhibitors and other monoclonal antibodies are metabolized to peptides and amino acids by circulating phagocytic cells.⁴³ Thus, renal or hepatic dysfunction would be expected to have minimal impact on drug levels, metabolism, or clearance.

To address the safety and efficacy of immune checkpoint inhibitors in patients with organ dysfunction, we recently assessed our experience with 27 patients who had pre-existing cardiac, hepatic, or renal dysfunction treated with anti-PD-1/PD-L1 agents. Somewhat arbitrarily, we included patients with creatinine ≥2.0 mg/dl or creatinine clearance <30mL/min, with liver enzymes or bilirubin ≥ 3× the upper limit of normal or radiographic evidence of cirrhosis, or with cardiac ejection fraction <45%. In this cohort, we did not observe an excess incidence or unique pattern of organ-specific irAEs.⁴⁴ Several patients experienced volume overload and congestive heart failure exacerbations that resolved with supportive care. Whether this represents systemic inflammation leading to cardiac strain and volume overload, excess fluid-volume associated with infusions, or simply the natural history of the disease is unclear. In three patients receiving hemodialysis, we did not observe any irAEs or other concerning adverse events. Durable responses occurred in several patients. Other clinical trials have permitted mild organ dysfunction (e.g. creatinine clearance 30 – 60mL/min) and have also reported efficacy comparable to patients without these co-morbidities.^{39, 45} One study of cisplatin-ineligible patients with urothelial carcinoma treated with atezolizumab reported equivalent response rates in patients with impaired renal function, although differential toxicity was not reported in this study.⁴⁶ Thus, in general, we do not consider even severe organ dysfunction as a contraindication to anti-PD-1/PD-L1 in patients with adequate performance status and functional reserve.

Recently, fulminant myocarditis and other cardiac events have been reported following immune checkpoint inhibitors (particularly combination PD-1/CTLA-4 blockade).^{2, 47} It is not at all clear whether pre-existing cardiac conditions predispose patients to these more severe events. Indeed, two of the most severe cases lacked cardiac risk factors other than hypertension.² Some studies now mandate cardiac monitoring early on combination treatment (e.g. weekly troponins and baseline electrocardiograms). However, it remains unclear whether to withhold treatment if cardiac co-morbidities are identified. One potential approach would be to use single agent anti-PD-1 rather than a more aggressive combination in patients with more severe cardiac co-morbidities, although this recommendation is not supported by firm data.

Extremes of age

Although very elderly patients were not specifically excluded from clinical trials, this population is underrepresented in nearly all studies.⁴⁸ One could speculate that age could influence immune cell function, as well as the composition of the tumor microenvironment.⁴⁹ Efforts are ongoing to understand the interaction between young vs. aged immune cells and interactions with anti-PD-1. Clinically, the impact of age on immunotherapy response has not been systematically described. No clear differences have been noted in numerous trials between response rates in patients >65 or <65. To assess at a more granular level, our group performed a retrospective review of patients to assess whether age influenced the rates of therapeutic responses or toxicities. In this cohort, we observed similar progression-free survival, overall survival, and toxicity rates in patients by decades.^{50, 51} Further, a meta-analysis of nine randomized controlled trials that enrolled 5265 patients to immune checkpoint inhibitors or various control arms, improvements in clinical outcomes were seen in both younger and older patients receiving immune checkpoint inhibitors compared with control therapies.⁵² This study, however did not directly compare outcomes in younger versus older patients. A 28% response rate was also recently reported in urothelial bladder carcinoma patients > 80 years of age treated with atezolizumab (compared with 23% for all patients).⁴⁶ We have also reported cases of clinical benefit and response in patients >90 years old, including a 90 year old treated with combination ipilimumab and nivolumab who experienced a complete response, and a 95 year old with a partial response following pembrolizumab.⁵³ We view that old age is not a contraindication to immune checkpoint inhibition, and rather that functional status is a more relevant consideration. However, the overall toxicity profile, efficacy, and relative risks and benefits compared with other therapies have not been studied comprehensively in tumor types other than melanoma, and will need further study.

On the other extreme, the safety and efficacy of immune checkpoint inhibitors for pediatric patients is incompletely studied. One completed phase I study tested ipilimumab in 33 pediatric patients with refractory solid tumors (primarily melanoma and sarcoma) and noted no responses, but stable disease in 18% of patients.⁵⁴ Toxicity profiles were similar to those observed in adults, but frequently occurred early, even after the first dose (uncommon in adults). Heterogeneous immune checkpoint expression, including PD-1 and PD-L1, has been described in a variety of pediatric cancers, although these tumors typically have relatively low mutation burden (associated with lower response rates in some cancers).^55–59 Numerous studies are ongoing in pediatric patients with a variety of cancer types involving anti-PD-1 with or without anti-CTLA-4.⁵⁶

Poor Performance Status

Patients who present with advanced cancer and poor Eastern Cooperative Oncology Group performance status (ECOG PS) present a challenging conundrum to oncology practitioners, and in many cases overlap with other populations already discussed. In general, cytotoxic chemotherapy (particularly later lines of therapy) is associated with substantial toxicities, impaired quality of life and short lifespan in patients with ECOG PS ≥2.^60–63 By contrast, immune checkpoint inhibitors, particularly anti-PD-1/PD-L1 monotherapy, often have favorable toxicity profiles, even in patients with poor PS. Moreover, some diseases like Hodgkin lymphoma, Merkel cell carcinoma, and melanoma have response rates in the range of 40–90% (albeit in clinical trial populations), further complicating the decision.^64–67

Very little published experience exists to guide oncologists in terms of whether to use immune checkpoint inhibitors in patients with poor PS. One study of atezolizumab in cisplatin-ineligible patients with urothelial carcinoma reported a response rate of 25% in patients with ECOG PS 2 (compared with 23% in all patients).⁴⁶ In our anecdotal experience (largely in melanoma patients), we have observed excellent responses in this population, albeit less frequently than in patients with preserved PS. This issue will need further study, particularly given the high costs of therapy. In our practice, we typically have a frank discussion about the likelihood of benefits and toxicities, and often involve palliative care early. We would be less likely to recommend treatment to patients with concurrent high-dose steroids (e.g. in the setting of progressive brain metastases) and more likely to treat patients with more responsive cancers (e.g. Hodgkin lymphoma, melanoma). An accurate predictive biomarker of response would have particular utility in this population to guide treatment decision making.

Brain Metastases

Patients with brain metastases represent a challenging clinical population with traditionally poor prognosis, and may present with limited PS or ongoing steroid use. While it remains unclear whether immune checkpoint inhibitors cross the blood-brain barrier, animal models suggest they may have modest penetrance and clinical studies have confirmed they can induce intracranial responses.⁶⁸ A phase II study of ipilimumab was conducted in 72 patients with melanoma and brain metastases, including 51 without neurologic symptoms and not receiving corticosteroids (cohort A), and 21 with stable symptoms on corticosteroids (cohort B).⁴² In cohort A, 18% of patients had disease control at 12 weeks, compared with only 5% (1 patient) in cohort B.

We have also recently reported a retrospective analysis on 89 patients with Stage IV melanoma treated with standard of care pembrolizumab, 36 (40%) of whom had brain metastases prior to commencing therapy.⁶⁹ The overall survival data was consistent with that seen for other pembrolizumab data, with 6-month overall survival of 86% and estimated median survival of over 20 months. Additionally, patients with brain metastases treated with radiation therapy and/or surgery prior to pembrolizumab (median 5 months, HR 0.27, 90% CI: 0.12 – 0.64) had similar degree of time to progression (TTP) as patients without brain metastases (median 6 months, HR 0.19, 90% CI: 0.08 – 0.42), compared to patients with untreated brain metastases (median 1.2 months). Additionally, concurrent radiation therapy with pembrolizumab was performed in 15 patients without unexpected toxicity. This data supports the use of pembrolizumab in patients with brain metastases and suggests that surgery or radiation prior to systemic therapy is ideal, but concurrent therapy with radiation is feasible.

Early results from a phase II study of pembrolizumab in patients with brain metastases and melanoma and NSCLC (limited to PD-L1 expressing tumors) have also demonstrated efficacy. In this study, 22% of melanoma and 33% of NSCLC patients experienced intracranial responses, and 3 melanoma patients (17%) developed grade III neurologic toxicities.⁷⁰ A phase II study of ipilimumab in combination with nivolumab in patients with brain metastases is ongoing (NCT02320058). Other studies are ongoing to prospectively evaluate interactions between brain radiation therapy and immunotherapy, as early results have speculated that combining these modalities may augment efficacy and perhaps toxicity.^71–73

Pregnancy

Metastatic cancer in pregnancy is a devastating condition leading to difficult decisions involving the mother and the developing fetus. PD-1/PD-L1 interactions appear to play a key role in maintaining fetal tolerance; indeed placenta is often used as a positive control for PD-L1 expression given its strong and ubiquitous expression. In animal studies, anti-PD-1/PD-L1 clearly increased the risks of spontaneous abortions.^{74, 75} However, in surviving animals, no increased risks of birth defects were noted. At this time, anti-PD-1 agents are categorized as pregnancy category D by the Food and Drug Administration, whereas ipilimumab is pregnancy category C (due to the less clear role of the CTLA-4 axis in fetal immune tolerance). Use of these agents in pregnancy, particularly anti-PD-1/PD-L1, would likely pose great risk of spontaneous abortion, with unknown maternal risk and unknown risks of birth defects, and would ultimately be an individualized decision made in light of the risks and potential benefits.

Conclusions

The use of immune checkpoint inhibitors across the spectrum of human cancers is rapidly expanding. Thus, characterizing their efficacy and safety in “real world” patient populations not included in clinical trials is a critical objective. Available studies suggest that these agents have often have acceptable safety profiles even in trial-ineligible populations, with the likely exception of anti-PD-1 in solid organ transplant recipients. Larger prospective studies would help extend and validate these experiences but may be challenging to complete. Further, identifying reliable predictive biomarkers of efficacy and toxicity may also help assist with treatment decision-making. In the interim, clinicians should consider these data when making treatment decisions in challenging patient populations, balancing the risks of toxicity with potential benefits, and making such decisions in a multidisciplinary setting.