Combination immunotherapy

Despite IO offering durable treatment options for many patients living with metastatic melanoma, nearly half of treated patients will not respond, even with combined checkpoint inhibition. T-Vec is distinct from other IO agents currently available, in part, as it is locally administered but still demonstrates the capacity to induce response in non-injected lesions through systemic immune activation. This antitumor effect is likely due to both the cytotoxic capacity of T-Vec, resulting in increased neoantigen exposure to host antigen presenting cells (APC) following tumor lysis, as well as its ability to modulate the tumor immune microenvironment (TME). In the phase II study of T-Vec in melanoma, post-treatment tumor biopsies were obtained, and treated patients were observed to have lower levels of immunosuppressive cell types including CD4 + FOXP3+ regulatory T-cells and myeloid derived suppressor cells (MDSCs) within the TME of injected lesions as compared to reference controls [5]. In two patients for whom biopsies of both injected and non-injected tumors were available, the non-injected lesions demonstrated a phenotype containing intermediate levels of these immunosuppressive subtypes as compared to biopsy data from untreated patients. These findings suggest that T-Vec may promote a more effective antitumor immune response.

However, while non-injected lesions do regress with T-Vec, the systemic response rates are significantly lower than those seen with PD-1 checkpoint inhibition. Modest response rates in visceral metastasis seen with T-Vec suggest that this agent may be most effective when used in combination with other IO agents, including checkpoint inhibitors, which can re-activate exhausted T-cells. Ipilimumab, as well as pembrolizumab, have both been evaluated in the phase I setting with T-Vec, with significantly higher response rates appreciated, particularly at non-target sites, than with T-Vec alone. When combined with ipilimumab, 52% of non-target index lesions demonstrated regression with an immune related response criteria (irRC) overall response rate of 50% (4 complete responses, 5 partial responses), which is dramatically higher than the approximately 10% response rate seen with ipilimumab alone [5]. While patients with brain metastasis were excluded from these studies, our patient’s response to T-Vec and pembrolizumab suggests that this combination can result in a systemic immune response capable of treating disease within the CNS.

Management of melanoma brain metastasis with immunotherapy

As was the case for our patient, complications associated with CNS metastases is a leading cause of cancer related mortality in melanoma. Response data for melanoma brain metastasis with IO is limited as these patients were excluded from landmark IO clinical trials.

However, following the promising preliminary response data in the phase I setting, multiple dedicated trials that focus specifically on patients with melanoma brain metastasis are underway. These studies are designed to evaluate treatment efficacy on CNS related endpoints including intracranial response (ICR), functional status, and survival without neurologic decline, and will provide critical insight on the use of IO for asymptomatic and symptomatic brain metastasis [16]. Currently, a three-arm phase II trial, which includes two cohorts of patients with asymptomatic brain lesions treated with either ipilimumab/nivolumab for four doses followed by nivolumab biweekly (cohort A) or nivolumab monotherapy (cohort B), is underway. The study also includes a third cohort (cohort C) receiving nivolumab monotherapy with symptomatic or refractory brain metastasis or leptomeningeal disease, a subpopulation of patients that have never been included in earlier trials. The principal endpoint is ICR with secondary endpoints including extracranial response, overall response rate (ORR) and PFS/OS. Preliminary results initially reported at the ASCO 2017 Annual Meeting, and recently updated at the 2017 World Congress of Melanoma meeting, are promising with patients treated with ipilimumab/nivolumab achieving the highest ICR of 46%, double the 20% observed in cohort B. Cohort C had the lowest ICR (6%) [12, 13]. As would be expected, grade 3–4 toxicities were most common in Cohort A (43%), driven by ipilimumab, and similar to those seen in prior studies. Of note, neurological adverse events were rare (6%), occurring in only 4 patients, with one patient developing radionecrosis [12]. One seizure and two headaches were deemed to be related to treatment [12]. A similar ICR rate of 56%, was seen in patients treated with ipilimumab/nivolumab on CheckMate 204, also presented at the ASCO 2017 Annual Meeting [14].

The prognosis of melanoma brain metastases patients requiring WBRT is poor with 3 to 4 months survival. Melanoma is considered to a highly radioresistant tumor, and most of these patients have continued cranial progression. On our review of the literature we are unable to identify a published complete ICR to WBRT alone in the setting of multiple melanoma brain metastases [17–20]. However there is mounting evidence that the radiation response may be synergistic when used in combination with IO [21, 22]. Radiation, like oncolytic viral therapy, offers the possibility of increased tumor antigen presentation and can participate in IO-instigated in situ vaccination. Therefore, radiation could serve to augment an effective systemic immune response. Frequently referred to as the “abscopal effect,” tumor regression has been documented at non-irradiated sites and been attributed to an anti-tumor immune response [23]. However, unfortunately, this phenomenon has not been to-date reliably invoked in clinical management and, in fact, in some instances immune suppression has also been documented in the context of radiation therapy. These variable findings may be, in part, due to the total dose of radiation delivered or the dosing schedule [24–26]. Several studies are currently underway, which will hopefully provide additional insight into how radiation can be used to enhance the activity of IO (NCT02716948, NCT02858869).

Predictive & prognostic biomarkers in melanoma

With new IO agents in development and the rise of combination IO, there is a need for accurate biomarkers to assist clinicians in determining which patients are most likely to respond to IO, as well as which patients would be expected to respond to alternative IO therapy at progression of disease. For ipilimumab, an increase in tumor-infiltrating lymphocytes (TILs) as well as a high genomic mutational load have been shown to be predictive of response to therapy [27, 28]. Certain specific mutations, including the NRAS mutation detected in our patient, also have been identified as associated with response to IO [29]. For anti-PD-1 therapy, TIL profiling, specifically with an increased density of CD8 T-cells, as well as high PD-L1 expression (≥50% by immunohistochemistry), can identify patients with the greatest probability of achieving a clinical response. However, a significant percentage (~ 12%) of patients with low PD-L1 expression (< 10%) will respond to treatment [30–32]. Overall, none of these biomarkers are sufficiently precise at determining which patients are likely to respond and more importantly, which IO agent might be expected to be most effective. As such, novel genomic and tissue based evaluation techniques are currently in development with the goal of addressing this need [33].

Among these approaches is qmIF, which can provide a higher resolution evaluation of the tumor immune microenvironment as compared to traditional IHC. We include this data as a representative example of the type of analysis that can be performed. Although this patient’s tumor showed a significant degree of TIL infiltration, described by the evaluating pathologist as varying from focally “non-brisk” to “brisk,” the population of CTL (CD3 + CD8+), which are known to be critical in anti-tumor response to programmed death-ligand 1 (PD-L1) therapy, were consistently low [28]. Furthermore, PD-L1 expression, on tumor cells, albiet of limited utility, was also absent. While clinically validated biomarkers are required, we speculate that despite having a low CTL density, the CD3 population (presumably comprised of CD4+ cells, including T regulatory cells) detected with qmIF, may have been suggestive of the potential fora response to immunotherapy. Of note, while our patient would not have been a candidate for MASTERKEY-26 trial due to symptomatic brain metastases, our findings are consistent with the corollary studies performed on tumor specimens obtained on this trial. In this phase Ib cohort (n = 21), response to combined T-Vec/pembrolizumab did not correlate with baseline CTL infiltration [34]. These findings, would suggest, that oncolytic viral therapy may be capable of augmenting the activity of anti-PD-1 therapy even in patients with a “non-favorable” TME.

In this case, qmIF was performed on this patient’s primary lesion prior to receiving any immunotherapy treatment, however this analysis can be performed on the TME of metastatic tumors, including the CNS, and may provide additional insight into immunoediting mechanisms unique to the CNS. Clinicopathologic studies indicate that the CNS is not an “immune-privileged” organ and multiple studies have demonstrated that CNS metastasis frequently resembles primary tumors with regards to immune infiltrate [35]. TIL infiltration, when seen in the CNS, similar to extracranial lesions, has demonstrated prognostic importance when found in the TME of melanoma brain lesions [36].

However, as new therapies are developed, access to the TME should not be assumed to be the same between extracranial and intracranial metastases as there are unique features of the TME in the CNS which warrant specific evaluation for their functional and prognostic implication. These include microglial cells which are known to be capable of antigen presentation and can express PD-L1, as well as the absence of high endothelial venules, which are postulated to enhance immune cell recruitment into the TME [9, 37].