BETTER USE OF IMMUNOTHERAPY

Cytotoxic T lymphocyte antigen 4 (CTLA-4) and programmed cell death 1 (PD-1) are normally important in attenuating immune responses to maintain immunologic homeostasis. Immune checkpoint inhibitors that block these immunoregulatory targets therefore lead to enhanced immune activation, controlling tumors but also occasionally leading to unintended inflammation of normal tissues, termed immune-related adverse events (irAE). In this issue, Menzies et al.^[1] and Spain et al.^[2] add to our understanding of irAEs in patients with advanced melanoma and provide questions for further study.

Specifically, Menzies et al.^[1] retrospectively examined the safety of PD-1 inhibitors (pembrolizumab and nivolumab) in two unique patient populations largely excluded from prior PD-1 clinical trials– patients with pre-existing autoimmune disorders (n = 52) and patients who had prior significant toxicity from ipilimumab (n = 67), defined as any grade 3 or greater toxicity or any toxicity requiring steroids. The authors' main conclusion is that PD-1 therapy is reasonable in these higher risk patient populations because patients can still benefit from treatment. Although the toxicity may be somewhat higher, it is manageable.

Clinicians treating patients with immunotherapy have been wondering about the safety of immune checkpoint inhibition in patients with underlying autoimmune disorders for many years. Since no prospective data were available, investigations into this issue began with retrospective case reports. Some case reports suggested ipilimumab may be safe,^[3] but in others, ipilimumab was associated with worsening of underlying autoimmunity.^[4] The largest retrospective series reported 30 patients with advanced melanoma and autoimmune disorders treated with ipilimumab.^[5] In this study, a proportion of patients had benefit from ipilimumab, although it resulted in some flares of underlying autoimmunity and other, more traditional irAEs not obviously related to the patients' underlying autoimmune disorders.

It is reassuring, although not too surprising, that similar findings were also present in the series by Menzies et al. of PD-1 treated patients, with even lower rates of grade 3/4 toxicity than ipilimumab. In patients with metastatic cancer and underlying autoimmune disorders, metastatic cancer is often the greater threat to life expectancy. The weight of evidence suggests these patients should be offered the opportunity to benefit from immune checkpoint inhibitors, and that drugs targeting PD-1 seem preferable to ipilimumab.

Menzies et al. extended their observations of PD-1 safety by examining an additional cohort of patients who had prior ipilimumab toxicity. In this group, Menzies et al. concluded that recurrence of the specific prior ipilimumab toxicity is rare with subsequent PD-1 therapy. While there may be a higher rate of PD-1 toxicity in general in this group, toxicity is manageable and patients can benefit from PD-1.

These results are interesting but need to be considered in the changing landscape of melanoma treatments. Due to superior efficacy, PD-1 monotherapy (or in combination with ipilimumab) is generally given prior to ipilimumab monotherapy in many countries with access to this treatment sequence.^[6,7] Therefore, the greatest clinical significance may lie in extrapolation of these results to future patient populations where ipilimumab is given first and then subsequent PD-1 therapy is considered. For example, ipilimumab is approved in the United States as adjuvant therapy for resected stage III melanoma and was recently shown to have an overall survival benefit in this setting.^[8] In patients who elect to receive adjuvant ipilimumab, experience toxicity, and subsequently have metastatic disease recurrence for which they are considering PD-1 therapy, the data from Menzies et al. provide some reassurance that re-exacerbation of any prior ipilimumab toxicity may be unlikely. Whether PD-1 monotherapy is safe to deliver after prior PD-1 + ipilimumab combination immunotherapy toxicity remains an ongoing question, but one of relevance given the high rate of treatment discontinuation among patients treated with PD-1 and ipilimumab in combination.

While Menzies et al. focused on safety of PD-1 inhibition in unique patient populations, the article by Spain et al.^[2] focused on the issue of neurologic irAEs as a less well-characterized event in clinical trial reporting. Among 352 patients treated at a single institution with either ipilimumab, nivolumab, pembrolizumab, or the combination of ipilimumab and nivolumab, Spain et al. characterized 10 patients who experienced neurologic toxicity thought to be related to immune checkpoint inhibition. Though neurologic irAEs were rare in this series (2.8% of patients overall), this rate is higher than the rate often reported in clinical trials (~<1 style="font-size: 0.75em; line-height: 1; max-width: 100%;" sup="">[9,10]

Fortunately most of these neurologic toxicities resolved with immunosuppression.

There have been a number of prior case reports describing various neurologic irAEs arising with checkpoint inhibition such as demyelinating polyradiculoneuropathy,^[11]posterior reversible encephalopathy syndrome,^[12] and a Guillain-Barre like syndrome (GBS)^[13] particularly notable since GBS resulted in a treatment-related fatality in an adjuvant study of ipilimumab.^[8] Other case series reporting neurologic toxicities with ipilimumab^[14] and PD-1^[15] have also been published with incidences similar to the cohort reported by Spain et al.

The report from Spain et al. adds to the literature in several important ways. First, the authors propose immunosuppressive treatment algorithms for neurologic toxicity. Notably, these proposed algorithms for neurologic toxicity differ from standard irAE management algorithms as the authors' algorithms specifically incorporate immunosuppression of B-cell/humoral immunity with treatments such as intravenous immunoglobulin (IVIG) and plasmapheresis as second line immunosuppression to consider after initial steroids. This recommendation is consistent with B cell immunosuppression reported in prior case reports of neurologic toxicity such as with the CD20 antibody rituximab.^[16] Additional research is needed, however, into the exact mechanism of neurologic irAEs to best understand how specific immunosuppressive modalities may be used to mitigate toxicity, while ensuring preservation of the antitumor immune effect.

Of additional importance, Spain et al. were meticulous in reporting neurologic toxicity by specific grade and not reporting 'all grade toxicity' and 'Grade 3–4 toxicity' as is commonly done in clinical trials. While grade 3–4 toxicity is clearly a significant category of neurologic toxicity to report, grade 2 neurologic events are often incredibly different from grade 1 neurologic events. For example, per Common Terminology Criteria for Adverse Events (CTCAE) version 4.0, a grade 1 peripheral motor neuropathy is asymptomatic whereas a grade 2 event is associated with moderate symptoms and/or limitations of instrumental activities of daily living. Toxicity data from clinical trials would likely be more informative if grade 2 toxicities were similarly reported separately from 'all grade toxicity'.

Both Menzies et al. and Spain et al. should be congratulated for their efforts retrospectively addressing essential clinical questions related to irAEs in patient populations for whom little clinical trial data exist. Their efforts raise ongoing questions as to how we can most rapidly learn from our growing clinical experience with immune checkpoint inhibition across the oncologic spectrum. First, the data from Menzies et al. suggest that patients with underlying autoimmune diseases should not necessarily be excluded from clinical trials. One could envision including these patients to learn about the risk/benefit profile within these patients, particularly in later phase trials or within independent cohorts after initial safety and dose-limiting toxicity assessments have been established from earlier phase clinical studies. Second, since data from Menzies et al. suggest a very low likelihood of PD-1 exacerbating prior specific ipilimumab toxicity, clinical trials should not exclude patients simply because they received immunosuppression with drugs like infliximab which is appropriate treatment for certain irAEs. Finally, large groups of patients are needed to determine patterns in clinical presentation and best management practices among patients with rare irAEs such as neurologic toxicity. Increasingly sophisticated technology platforms may be able to eventually support multicenter, international databases, as suggested by Spain et al., to accomplish these goals. The number of patients treated with immune checkpoint inhibition continues to increase, and we will continue to learn from our growing collective experience.