DKK1-Targeted Agent Shows Efficacy in Gastroesophageal Cancer

Samuel J. Klempner, MD, medical oncologist at Massachusetts General Hospital, discusses the preliminary results of a study of the investigational agent DKN-01 for patients with gastroesophageal adenocarcinoma.

The ongoing phase 2a DisTinGuish trial (NCT04363801) enrolled patients with advanced esophageal cancer to receive DKN-01 in combination with tislelizumab (BGB-A31) and capecitabine/oxaliplatin. Part A enrolled all patients with advanced disease for frontline therapy, while part B focused on second-line therapy following chemotherapy in those with high DKK1 expression.

Klempner says the objective response rate in the part A frontline population was 68%, but in the DKK1-high second-line population in part B, it was 90%, including deep and durable responses. While the progression-free survival (PFS) and overall survival (OS) data were not yet mature, the median PFS was 10.7 months at a data cutoff of September 15, 2021, which he says is comparable with that of nivolumab (Opdivo) and ipilimumab (Yervoy) in CheckMate649 (NCT02872116).

In addition to the correlation between DKK1 expression and efficacy, the study found no clear correlation between high DKK1 and high PD-L1 expression, which supports the hypothesis that antitumor activity can be attributed to DKN-01’s targeting of DKK1 and not the anti–PD-1 activity of tislelizumab.

TRANSCRIPTION:

0:08 | In terms of efficacy, that is one of the key findings of this presentation, in part A—this is the frontline population—the response rate in the modified intention-to-treat population was 68%. But really what is of interest is in the biomarker-high population, where you would expect to see the greatest efficacy, the response rate was 90%. Again, the numbers are relatively small, but some of these responses are quite deep and durable, as was shown in the spider plot on the poster. And the activity was higher in the DKK1-high patients, which is what you would hope to see if the biomarker is identifying a population that is predicted to have a better response rate.

0:58 | PFS and OS data [are] immature, but the median PFS is 10.7 months, which is encouraging in a phase 2 setting and compares favorably [with] the phase 3 trial data from CheckMate649, for example. But again, it's small numbers and we'll need larger numbers to confirm if this is a true observation. But I would say the take-home message is in this population, the biomarker does appear predictive of patients who are going to get the most benefit.

1:32 | The final point that I would make is it does not appear that DKK1 elevation correlates with PD-L1 expression, so it's not like the DKK1-high patients are also the PD-L1–high patients who would have been more likely to benefit anyway. So those 2 things appear independent, suggesting that the drug is not acting primarily via PD-L1–the efficacy is not mediated via the PD-1 inhibitor. We presume this reflects some activity of the DKN-01 agent added to the therapy.