Background: We have previously shown that undetectable ctDNA either at baseline or during therapy predicted response in mm patients (pts) treated with anti-PD1 antibodies (aPD1). Pseudoprogression, defined as radiological progression prior to response, occurs in 8% of pts treated with aPD1. We sought to determine if ctDNA could differentiate pseudoprogression from true progression, defined as continued clinical or radiological disease progression. Methods: Between July 2014 and May 2016, pts receiving aPD1 had serial bloods for ctDNA. Included pts either had RECIST PD at first restaging or early clinical progression. Those with untreated brain metastases were excluded from the analysis. ctDNA was quantified using digital droplet PCR for mutations (BRAF/NRAS) at baseline and during the first 12 wks of treatment. Based on our prior studies, ctDNA results were grouped in to ‘favorable’ and ‘unfavorable’ ctDNA profiles (see Table), and these were compared in pts with true and pseudoprogression. Results: 29 pts were included, 28 with RECIST PD at first restaging and one with early clinical progression. 9 (31%) pts had a subsequent RECIST PR or SD and were considered pseudoprogression and 20 (69%) had true progression. Of the pseudoprogressors, 7/9 pts remained in response with a median follow-up of 20 months (mths). 2/9 pts had disease progression at 7 and 18 mths, with ctDNA that remained detectable with a > 10-fold decrease during treatment in both patients. Of those with true progression and a favourable profile, 1 had a > 10-fold decrease in ctDNA by wk 12 and was switched to MAPK therapy prior to further imaging, and the other had an undetectable ctDNA at wk 6 which increased again at wk 12. The latter pt had a new lesion on first restaging CT scan despite PR in all existing lesions with true PD on second restaging at wk 24. Conclusions: ctDNA in patients with mm at baseline and early on aPD1 treatment differentiates pseudo from true progression.