MOONLIGHT evolved into a platform trial of sequential, randomized and single-arm cohorts investigating the combination of dual immunotherapy with FOLFOX in parallel versus sequentially and versus chemotherapy alone, as well as triplet FLOT plus nivolumab.
Until now the relative contribution of adding the CTLA-4 checkpoint inhibitor ipilimumab to first-line combined nivolumab + chemotherapy for patients with advanced gastroesophageal adenocarcinoma has not been investigated. So far, also the triplet regimen FLOT in combination with nivolumab has not been evaluated.
Given the favourable response- and progression-free survival rates of the combination of anti-CTLA-4 and PD-L1 in the treatment of metastatic melanoma9 and in advanced chemo-refractory GEA,7 it seemed reasonable to compare the addition of nivolumab/ipilimumab to chemotherapy head-to-head with chemotherapy and in a sequential approach.
At the time MOONLIGHT was designed, data on the effectiveness of checkpoint inhibitors and their dependence on the level of PD-L1 expression were preliminary. Therefore, MOONLIGHT included patients independent of the PD-L1 expression status.
In MOONLIGHT, dual checkpoint inhibition plus Chemotherapy administered in parallel was associated with no improvement in activity compared to chemotherapy alone, with comparable survival and response rates, and there was no enrichment with increasing PD-L1 CPS cut-offs. Comparable results, with no additional benefit of adding the CTLA-4 checkpoint inhibitor Tremelimumab to FOLFIRI plus Durvalumab, independent of PD-L1 CPS in second-line GEA have been reported recently by the French DURIGAST Trial.17 In MOONLIGHT, for chemo alone, the median OS and median PFS were even numerically longer and were in line with survival rates reported by the CheckMate 64914 and KEYNOTE 859 trial18 for the standard chemotherapy arm. One explanation for this observation may be the higher number of treatment-related adverse events of grade 3/4 which were seen with dual immunotherapy in combination with chemotherapy (TRAEs 74% vs. 45%). It is known that combining anti-PD-L1/anti-PD-1 with anti-CTLA-4 increases the proportion of grade 3 to 4 AEs.7 Recently, published results also suggested an unfavorable benefit–risk profile with no improvement in survival but a higher number of treatment-related adverse events and deaths for nivolumab–ipilimumab combined with platinum-based chemotherapy relative to pembrolizumab combined with platinum-based chemotherapy as a first-line treatment for patients with advanced NSCLC.19
Another explanation for the failure of adding immunotherapy might be the generally low number of PD-L1 positive patients with less than 50% harboring a PD-L1 status of ≥1 (antibody clone 28-8), as trials in GEA have revealed that only the subset of patients with high-PD-L1 expression on tumor and immune cells substantially benefit from immune checkpoint inhibitors.14,16,18 Compared to PD-L1 expression rates reported from pivotal global phase III trials in HER2 negative GEA, ranging between CPS ≥ 70–8014,18 PD-L1 expression in European trials were indicated lower.20,21 A meta-analysis by Lin et al.22 found that the PD-L1 expression rate in studies from Asian areas was numerically higher (52.3%) compared to those from non-Asian areas, including Caucasians (32.7%), which might be due to higher frequencies of virus-associated malignancies in Asian areas. However, data are conflicting as trials in other malignancies such as lung cancer indicate, despite some regional differences, a relatively consistent PD-L1 expression across regions like Europe, Asia Pacific, America.
While there are some differences in PD-L1 expression rates between European and international studies, these differences may be also influenced by regional, methodological, and interlaboratorv factors.
Of note, OS and PFS did not seem to vary according to PD-L1 CPS. The reason why we didn’t observe a difference in the magnitude of survival benefit between patients with and with no PD-L1 expression might be the small number of patients, the generally low PD-L1 expression rate, and the chosen cut-off ≥1, as trials demonstrated that only the subset of patients with high PD-L1 expression on tumor and immune cells benefit substantially from immune checkpoint inhibitors.14,16,18
The second question, if an alternating regimen with immune maintenance therapy in patients with disease control after short term induction chemotherapy alone is as effective and less toxic as continuous chemoimmunotherapy showed a markedly longer OS with FOLFOX and nivo and ipi in parallel than that of FOLFOX induction followed by nivolumab and ipilimumab (16.5 months vs. 6.9 months) either in the overall or prespecified PD-L1–positive population (CPS ≥ 1%). Similar results have been reported in JAVELIN Gastric 100 trial, which failed to demonstrate superior OS with avelumab maintenance versus continued chemotherapy in patients with advanced GC or GEJ cancer.20 Of note, Javelin had a longer chemotherapy induction of 3 months and did not show a detrimental effect for IO maintenance therapy. Of note, median time of chemotherapy treatment in the sequential treatment was only 6 weeks (3 cycles) with only 38% of patients receiving chemotherapy re-induction after immunotherapy, meaning that in a disease as aggressive in terms of tumor biology as GEA, prolonged chemotherapy is necessary to control the tumor. Although the number of subsequent therapies was above average what is reported from phase III trials and was highest in the sequential treatment arm (65%), the subsequent chemotherapy could not compensate the early progression on immune- maintenance therapy. Despite associated with lower toxicity and a clearly more favorable safety profile versus the combined chemotherapy plus IO therapy, including lower rates of grade ≥ 3 TRAEs (74% v 45%), reductions in treatment-related hematologic AEs, and a lower incidence of neuropathy, our study doesn’t support the concept of chemotherapy induction followed by immunotherapy.
Third, MOONLIGHT addressed the questions of triplet FLOT regimen in combination with immunotherapy, which is of importance, as recently the French GASTFOX trial confirmed that the addition of docetaxel to oxaliplatin and 5-fluorouracil (mFLOT/TFOX) is more effective than oxaliplatin and 5-fluorouracil (FOLFOX) for patients with advanced gastric or gastroesophageal junction (GEJ) adenocarcinoma.23 This regimen is considered a new first-line treatment option for patients eligible for triplet regimens, at least for patients with PD-L1-negative tumors according to current guidelines.24 Although in label according to EMA approval, there have been no data on safety and efficacy of a triplet chemotherapy regimen in combination with immunotherapy so far. Data on the combination of immunotherapy with triplet taxan containing chemotherapy have been only conducted in a single arm phase II trial in locally advanced head and neck cancer, reporting favorable response and survival rates with good tolerance.25 In MOONLIGHT, the combination of nivolumab and FLOT in 52 patients with metastatic GEA was associated with a favorable ORR and an improved disease control in the early course of therapy as well as a prolonged survival rate. Besides, the safety profile of nivolumab plus FLOT chemotherapy was consistent with the known safety profiles of the individual treatments and no new safety signals were identified. Nivolumab plus chemotherapy-related grade three or worse events were 67% and the most common grade 3 AE was a neutrophil decrease in 44%. These data were consistent with the known safety profile of checkpoint inhibitors with doublet chemotherapy as shown in CheckMate 64914 and Keynote 85918 with similar frequencies of Grade 3-4 TRAEs (CM 649 in 60%; KN 859 59%).
Of note, baseline characteristics in this group were more favorable for the success of immuno-chemotherapy. By immunohistochemistry evaluation, PD-L1 expression degrees was highest in the FLOT plus nivolumab cohort with 65% harboring a PD-L1 CPS of ≥1. Both, Keynote 859 and CheckMate 649 determined a meaningful prolongation of survival treated with pembrolizumab/nivolumab in metastatic GEA patients with more than 1%/5% PD-L1 expression.14,18 Besides, FLOT plus nivolumab led to a numerically longer median OS in patients with a tumor PD-L1 expression with a median OS of 16.7 months in PD-L1 positive patients vs 10.6 months in PD-L1 negative patients. Between trial comparisons of triplet FLOT plus nivolumab in MOONLIGHT vs doublette FOLFOX plus Pembrolizumab in Keynote 859 with a median OS of 13 months in CPS ≥ 1 patients, anticipates an increased efficacy with the addition of the taxane. These results suggest that adding nivolumab to FLOT is an effective treatment option with good tolerance and may represent an option for selected patients with aggressive, rapidly progressive disease. However, given the small cohort size and lack of control for evolving treatment standards, results must be interpreted with extreme caution. A potential next step could be to evaluate mFLOT/TFOX or FLOT regimen in combination with immunotherapy or zolbetuximab in biomarker selected patients.
Moonlight had several limitations. First, patients were recruited independent of biomarker selection based on PD-L1 score and mismatch repair deficiency, although rare, was not measured. Overall, PD-L1 expression rate was comparable between arms but generally low compared to previous reported trials and the cohort with the most promising results (FLOT plus nivo) had the highest number of PD-L1 positive patients.
Second, sample size of each cohort was small and according to the statistical design, cohorts could not be compared statistically, as this study was not designed to compare these arms. Therefore, it is also not possible to speculate advantage of one or the other combination treatment for specific study subgroups which clearly limits the meaningfulness of this trial. The current study’s contribution is therefore exploratory rather than practice-changing
Another limitation is that the experimental arms (A1/A2) and A2 have not been compared to the current first-line treatment standard of doublet chemotherapy plus IO monotherapy, as this was not yet standard of care at the time the trial was designed.
In conclusion, the results of MOONLIGHT suggest, that in the first-line setting of metastatic GEA, the combined immune checkpoint blockade with chemotherapy is insufficient and toxic and should not be further explored in future studies.
In addition, brief duration of 6 weeks of FOLFOX followed by immunotherapy alone is inferior irrespective of PD-L1 expression. Future trials should evaluate immune maintenance therapy after longer induction with chemoimmunotherapy in PD-L1 selected patients.
Encouragingly, the combination of FLOT with nivolumab is feasible and appears to show good efficacy, further supporting the use of this regimen for selected patients who need triplet chemotherapy for early tumor control and might benefit from additional immunotherapy due to positive PD-L1 expression. However, the potential benefit of FLOT plus immunotherapy should also be carefully weighed against its notable toxicity.
