Sir,

We have read with great interest the correspondence from Imafuku et al. The authors provide further evidence that sequential treatment of anti-PD-1 blockade followed by the application of the anti-CTLA-4 antibody ipilimumab is associated with an increased rate of severe immune-related toxicity (Danlos et al, 2015; Khoja et al, 2015; Aya et al, 2016; Bowyer et al, 2016; Furudate et al, 2016). This contrasts with the reverse treatment sequence where no increased frequency of immune-related adverse events has been observed (Weber et al, 2015; Ribas et al, 2016). The observations of Imafuku et al suggest that the timing between the administration of the last dose of an anti-PD-1 antibody and the first dose of ipilimumab is a critical factor with all patients who experienced high-grade immune-related toxicity, having received their first dose of ipilimumab within 1 month after the last administration of an anti-PD-1 antibody. In our patient cohort, the median time interval between therapies has been 32 days in patients who developed severe autoimmune toxicity vs 46 days in patients without toxicity, a difference that was statistically not significant. It should also be remembered that a high receptor occupancy is achieved after only a single dose of an anti-PD-1 antibody that persists for several months so that the turnover of PD-1 expressing immune cell populations may also be important in determining the immunological outcome next to the half-life of the antibody itself (Brahmer et al, 2010). The sequence in which PD-1 and CTLA-4 molecules are engaged on effector T cells also leads to vastly different gene expression profiles and one may propose to distinct immunological outcomes providing a potential explanation why differences in immune-related toxicity can be observed depending on the treatment sequence of checkpoint regulators (Das et al, 2015).

Imafuku et al did not present any efficacy data, but it should be emphasised that treatment with ipilimumab after anti-PD-1 failure has significant clinical activity with an objective response rate of 10–15%, which is in keeping with the treatment experience of ipilimumab in anti-PD-1 therapy naive patients. This has recently been confirmed in a larger data set of patients who received ipilimumab after progression on pembrolizumab in the Keynote-006 clinical trial (Long et al, 2016).

Overall, these data highlight that ipilimumab is a treatment option after progression on anti-PD-1 therapy and patients should be closely monitored for immune-related adverse events, particularly, if anti-CTLA-4 therapy is initiated shortly after the last application of an anti-PD-1 agent. An ongoing randomised clinical trial (NCT02731729) will provide prospective data regarding the efficacy and toxicity of single-agent ipilimumab therapy, or combination therapy of ipilimumab and nivolumab in patients who progress on anti-PD-1 therapy.