Long-awaited early results of the first Crispr-edited Car-T therapy failed to match Crispr Therapeutics’ gravity-defying valuation yesterday. The treatment-related death of the only subject given the highest dose of CTX110 so far overshadowed the data, and Crispr’s stock fell 14%.
Still, the evidence disclosed points to the death being a one-off event, and the data do broadly support Crispr’s approach. More concerning is the fact that on a cross-trial basis CTX110 has barely been able to match the data unveiled in May on its key allogeneic competitor, Allogene’s ALLO-501, and established autologous players.
The headline numbers for Crispr, from the phase I Carbon study in B-cell lymphomas, reveal that 11 subjects have been treated across four CTX110 doses. There have been four remissions, all complete, including two among four subjects given the third dose, 300 million cells, on which Crispr is now focusing.
The only subject given dose four, twice as high, also went into complete remission, but died 52 days later. Crispr said it had paused enrolment at this dose.
In terms of efficacy the immediate comparison, across trials, of course, will be made with the ALLO-501 data in lymphoma that Allogene unveiled at Asco. Here 19 patients were evaluable across three dose levels, and there were 12 responses, seven of them complete; there were four serious adverse events, but no deaths.
A key consideration regarding the efficacy is that both datasets are extremely immature. The median follow-up in the Allogene trial was 3.8 months, with just two subjects past six months, while three patients in the CTX110 study are three months beyond dosing.
This is important because relapses are known to be an issue for patients who initially respond to Car-T therapy, though this is a bigger problem for childhood leukaemias than for lymphomas. One of the CTX110 responders, at the second dose level, relapsed before six months.
Crispr said one of the subjects given 300 million cells had been redosed, presumably before recording a three-month complete response, and redosing is now an option in all cohorts. Allogene had redosed one relapsed partial responder, who then went into complete remission.
A separate issue is how these extremely early data compare against Novartis’s Kymriah and Gilead’s Yescarta, autologous Car-T therapies both approved for lymphoma. The US labels for these indicate ORRs of 50% and 72%, including 32% and 51% rates of complete remissions.
|Selected anti-CD19 Car-T therapies in lymphoma|
|Company||Gilead||Novartis||Bristol Myers Squibb||Allogene/Cellectis||Crispr Therapeutics|
|Type||Autologous (CD28 co-stim)||Autologous (4-1BB co-stim)||Autologous (4-1BB co-stim, defined cell composition)||Allogeneic (4-1BB co-stim; LV transduced, Talen-mediated TCR & CD52 knockout)||Allogeneic (CD28 co-stim; Crispr-mediated Car knock-in in TRAC locus, with β2M knockout)|
|Treatment-related deaths||3 (hemophagocytic lympho histiocytosis, anoxic brain injury, cerebral oedema)||3 (CRS in setting of stable or progressive disease)||4 (diffuse alveolar damage, pulmonary haemorrhage, multiple organ dysfunction syndrome, cardiomyopathy)||None||1 (HHV-6 reactivation)|
|*As these are approved for r/r DLBCL the efficacy numbers are as cited on the US labels, using remission criteria that are stricter than in the compared clinical trials; **90% of 92 subjects got bridging chemo, but the remission data cited on Kymriah's label consider only those who had measurable disease after bridging, or who had not been bridged; ^includes 151 subjects bridged with chemo.|
Crispr took some time on an analyst call to give details about the patient death, which it said occurred after grade 2 cytokine release had resolved. But the subject then developed encephalitis, and was found to have had reactivation of herpesvirus, a seemingly unusual event. Crispr put the death down to viral reactivation in an immunocompromised patient.
How they differ
ALLO-501 and CTX110 are both derived from healthy donors, and comprise a fairly standard anti-CD19 Car-T construct. While the former uses lentiviral transduction and Talen nucleases, the latter uses one-step Crispr Cas9 editing, meaning that the Car can be knocked in specifically at the genetic locus coding for the endogenous T-cell receptor, whose knock-out is achieved more easily.
But if the CTX110 and ALLO-501 data so far look borderline what is the point of an allogeneic therapy? Increasingly the argument is that allogeneics are more convenient, with a ready treatment available off the shelf and no need for bridging chemo, and an option for patients with poor-quality T cells, though few now claim that there could be a cost advantage versus an autologous therapy.
Behind CTX110 Crispr has CTX120, an anti-BCMA Car, and CTX130, which targets CD70, and the first clinical data from both are expected next year. Crispr’s first clinical trial has validated its approach, but it is far from the home run that an $8bn market cap had priced in.