With all the changes Astrazeneca made to its first-line lung cancer trial Mystic, and its reluctance to explain important details, it is hard for investors to know what to look for when data from this all-important study are revealed shortly.
Still, details have started to dribble out, and educated guesses can be made about many of the remaining gaps (see tables below). It has become clear that the most important results might not come out until next year, though a bigger question is whether the surprise accelerated approval of Keytruda plus chemo in the broad first-line setting has already rendered Mystic irrelevant.
That US approval was game-changing because it allowed Merck & Co’s Keytruda to be prescribed first line to patients with no PD-L1 expression, reducing to a scientific curiosity an earlier battle to capture different levels of PD-L1 expressers. Roche’s Impower-150 trial of Tecentriq plus chemo, due to yield survival data in the second half, could add further pressure here.
Mystic, which tests Astra’s Imfinzi alone or combined with tremelimumab, can in the near term do little more than match Keytruda plus chemo, with the possible advantage of giving an option to patients unable to tolerate chemo. Whether they would tolerate tremelimumab – Bristol-Myers Squibb’s similarly acting anti-CTLA4 agent Yervoy is notoriously toxic – is a separate point.
|Design of opel-label Mystic trial (NCT02453282)|
|Population||1,118 first-line NSCLC patients, stratified by PD-L1 expression (assumed to be above and below 25%)|
|Arm||Imfinzi + tremelimumab||Imfinzi monotherapy||Pt-based doublet chemo|
|Primary comparison||Versus chemo||Versus chemo||NA|
At least it is certain that progression-free survival (PFS) data from Mystic will be reported soon; there had been suggestions that, should PFS be negative, Astra would simply run Mystic straight to its overall survival (OS) readout in 2018, but the company has confirmed that, even if negative, PFS would be reported.
Astra’s sweeping changes to Mystic’s design promoted several analyses to primary endpoints, with the aim, as Astra kept repeating, of giving “optionality” (Astra’s Mystic mystery and other surprises in store, January 17, 2017). The group also insisted that there was sufficient power in Mystic for all of these.
However, while it seemed that PFS and OS were now primary measures, close inspection of Astra’s presentations reveals that PFS is only a primary for the combo arm, and thus this is the imminent focus; for Imfinzi monotherapy PFS is a secondary analysis, meaning that a positive effect here will hold no weight without a primary OS hit.
Beyond this the biggest Mystic mystery is the PD-L1 expression cut-offs to be tested – Astra has never revealed these, but many analysts are expecting expression >25% to form the basis of the patient stratification, supported by information already on Imfinzi’s label regarding the Ventana PD-L1 assay.
Therefore the narrowest, most strongly expressing group – assumed to be >25% – will be analysed first, with subsequent, hierarchical cuts capable of meeting statistical significance as long as the earlier PD-L1 level cutoffs show a positive effect.
|Making sense of Astra's Mystic mystery|
|Study arm||Primary endpoint||PD-L1 expression*||Note|
|H2 2017 events|
|Imfinzi + tremelimumab, >25% PD-L1||PFS||>25%||Stratification level; first hierarchical analysis*|
|Imfinzi + tremelimumab, <25% PD-L1||PFS||>5%||Capable of showing significance if >25% level is positive|
|Imfinzi + tremelimumab, <25% PD-L1||PFS||>1%||Capable of showing significance if >5% level is positive|
|Imfinzi + tremelimumab||PFS||All-comers||Capable of showing significance if >1% level is positive|
|Possible 2017/18 events|
|Not disclosed||Interim OS||Not disclosed||Unspecified number of interim OS analyses|
|Imfinzi + tremelimumab, >25% PD-L1||OS||>25%||Stratification level; first hierarchical analysis*|
|Imfinzi + tremelimumab, <25% PD-L1||OS||>5%||Capable of showing significance if >25% level is positive|
|Imfinzi + tremelimumab, <25% PD-L1||OS||>1%||Capable of showing significance if >5% level is positive|
|Imfinzi + tremelimumab||OS||All-comers||Capable of showing significance if >1% level is positive|
|Imfinzi monotherapy, >25%**||OS||>25%||Design specifies "PD-L1 expressing tumours" only|
|Note: *not disclosed formally; **for Imfinzi monotherapy PFS is a secondary endpoint only.|
An additional difficulty with Mystic applies to all immuno-oncology trials: pseudo-progression can confound a PFS benefit, while placebo subjects who progress and then switch to active treatment can erase an OS advantage.
At a press conference during the Asco meeting Astra’s head of immuno-oncology, Rob Iannone, confirmed that non-responding subjects in Mystic were not allowed to switch to the active arm within the study. However, this does not preclude them moving onto a different treatment, and confounding the data that way.
It must also be noted that Astra, having initially touted the Mystic readout as a critical transition point, has of late been reining back expectations, insisting that the data are not as binary as the market is perceiving. The group has also hinted that a PFS hit is not a dead cert, though PFS failure does not rule out an OS benefit next year.
Analysts are roundly ignoring the caution, with Leerink most recently calling Mystic a “highly anticipated... clear binary event”. EvaluatePharma’s sellside consensus sees 65% of Imfinzi’s 2022 sales of $2.6bn being generated in NSCLC.
With Astra stock running up 12% since the start of May the markets certainly see Mystic as strategically crucial, even if Astra no longer shares this view.