The results of this trial of compassionate use ipilimumab at MSKCC are largely consistent with the results presented in abstract form for several other phase 2 trials of similar doses/schedules of ipilimumab.20–22
Specifically, our objective RR of 12% (as adjudicated by the proposed immune-related response criteria) is very comparable to the objective RRs of 5.8% to 15.8% adjudicated by standard modified WHO criteria that are reported in other phase 2 trials. These trials also reported CR + PR + SD rates of 27.1% to 35.1%, similar to our immune-related CR + immune-related PR + immune-related SD rate of 41%. Finally, grade 3 to 4 immune-related adverse events were noted in 20% to 38.6% of patients in these trials, consistent with the rate of 29% we describe here.
In addition, the results of a phase 1 to 2 trial of 2 doses and schedules of tremelimumab, another anti–CTLA-4 antibody, were recently published.23
This trial reported an objective RR of 9% by Response Evaluation Criteria in Solid Tumors criteria, a CR + PR + SD rate of 39% to 41%, and a grade 3 to 4 toxicity rate of 13% to 27%. Median time to progression was about 1.9 months, with median OS of 9.97 to 11.53 months. With the exception of the longer OS reported in this trial, which was likely because of the enrollment of only treatment-naive patients with relatively low LDH, and despite the caveats of comparing different phase 2 trials, these results are otherwise strikingly similar to ours. Certainly, the median OS of 7.2 months that we report in our heavily pretreated patients also compares favorably to standard first-line therapies for melanoma, such as temozolomide or dacarbazine.24
In this study, results were reported according to immune-related response criteria because these were used for clinical decision making. When we also evaluated responses using traditional modified WHO criteria, we noted strong agreement between both criteria. As might be expected, the only discrepancies occurred when 2 patients who were adjudicated to have immune-related SD would have experienced PD by the modified WHO criteria because of the development of new metastatic lesions (which increased the total tumor burden used in the immune-related criteria by <25%). As such, there was no change in the objective RR or best response rate of SD or PD. There was a slight difference in the Week 24 CR + PR + SD rate of 30% (15 of 51 patients) versus the Week 24 immune-related CR + immune-related PR + immune-related SD rate of 33% (17 of 51 patients), as well as a small change in the median PFS by modified WHO versus immune-related criteria (2.5 vs 2.6 months, respectively).
Our results suggest that patients who develop grade 3 to 4 immune-related adverse events are more likely to experience clinical benefit at Week 24 compared with those with no or mild immune-mediated toxicity. These results are consistent with prior reports25–27
and are also biologically consistent with the belief that CTLA-4 blockade acts in a nonspecific fashion to de-repress the immune system, resulting in the frequent but not absolute co-occurrence of antitumor effect and autoimmune-like toxicity. Of note, the median number of ipilimumab treatments that patients with grade 3 to 4 immune-related adverse events received (which was 4) was identical to the median number of treatments patients with grade ≤2 immune-related adverse events received. The comparable duration of therapy suggests that the increased grade 3 to 4 immune-related adverse event rate in patients with clinical benefit was not likely to be because of increased exposure to ipilimumab.
Remarkably, our analysis shows that the ALC correlates strongly both with clinical benefit and OS. The ALCs obtained at baseline and after 1 ipilimumab treatment appear to be prognostic, but these early ALCs may be more of a reflection of the extent of prior therapy and tumor burden. However, the suggestion that a low baseline ALC may be associated with poorer OS does have implications for the sequencing of therapy in future trials that combine ipilimumab with chemotherapy.
The ALC obtained after the second ipilimumab dose appears to be a particularly informative biomarker. None of 8 patients with an ALC <1000 cells/μL experienced clinical benefit at Week 24; these patients also had statistically and clinically significantly inferior OS compared with patients with an ALC ≥1000 cells/μL. There was a difference in objective RR between both groups (18% vs 0%), but this was not statistically significant (P = .33), possibly because of the small number of objective responses that were noted in this trial.
The relatively small number of patients in this trial precluded us from performing a detailed multivariate analysis, although the ALC remained a significant prognostic marker when we controlled for baseline LDH. Nevertheless, the observation that other prognostic factors (number of prior therapies and the primary site of the tumor) were not correlated with OS suggests that the ALC does represent an independent biomarker.
There is a strong biological rationale that the ALC value is correlated with benefit from ipilimumab because it directly blocks CTLA-4 expressed on various lymphocyte populations. Presumably, a threshold ALC of 1000 cells/μL reflects the underlying capacity of the immune system to be adequately activated by ipilimumab to mediate clinically meaningful antitumor effects. Our observation also complements recent data indicating that the rate of change of ALC with ipilimumab therapy is positively associated with the CR + PR + SD rate.28
In particular, the observation from this larger data set that patients whose ALCs drop after initiation of ipilimumab have a 0% CR + PR + SD rate is consistent with our observation that a minimum ALC level may be required for patients to derive clinical benefit from ipilimumab.
Overall, these data have strong implications for clinical practice, as they suggest that the approximately 20% of patients with an ALC <1000 cells/μL will not benefit from additional ipilimumab therapy. Such patients could be spared the potential toxicity of further therapy and could be switched to an alternative treatment plan.
In conclusion, this report of patients with advanced melanoma treated at MSKCC in a trial of compassionate use ipilimumab is consistent with other phase 2 evaluations of anti–CTLA-4 antibodies. Furthermore, our results suggest a single ALC measurement obtained after 2 ipilimumab treatments is strongly predictive of clinical response to ipilimumab. A larger retrospective multivariate review and prospective validation of this biomarker are to be strongly considered, as they have the potential to identify patients who are unlikely to benefit from ipilimumab therapy, while benefiting those who are likely to benefit.