Seven questions CTL019 adcom should consider (and two it won’t)

With the US advisory panel convening on Wednesday July 12 to discuss Novartis’s CTL019 in paediatric leukaemia, CAR-T therapy could be on the verge of its first ever approval.

The Swiss firm comes armed with unprecedented remission rates in refractory patients; nevertheless, approval is not a given – toxicity and unreliable manufacturing are two issues that loom large, for instance. While the panel’s briefing documents are not yet available, here in no particular order are the topics likely to occupy panellists as they deliberate CTL019’s fate.

1. What about the relapses?

For CTL019’s approval in acute lymphoblastic leukaemia (ALL) Novartis will rely on the Eliana trial. Here a single dose of CAR-T cells put 83% of successfully infused subjects into complete remission – a striking finding given that 56% had failed stem cell transplant, and patients overall had progressed after one to eight lines of prior therapy.

However, the primary endpoint involves remissions at six months, and CAR-T is known to have a high relapse rate (Kite flies, but durability remains the key unknown, September 27, 2016). Last month Novartis revealed that the relapse-free rate of Eliana responders at six months was 75%, implying a six-month CR rate nearer to 60%.

Thanks to the advent of novel therapeutics paediatric ALL survival rates have surged in recent years, but CTL019 targets the relapsed/refractory setting, and the panellists will no doubt focus on the benefit that CAR-T therapy is giving over and above what this subset of patients would normally experience.

2. Are key Eliana patients being excluded?

Novartis has consistently used the number of successfully infused subjects as the denominator for its remission rate calculation. But the true intent-to-treat number is higher, because it includes enrolled subjects in whom it proved impossible to infuse CAR-T cells.

If the 16 Eliana subjects who discontinued before infusion are included the six-month remission rate falls to about 50%. The adcom is virtually certain to focus on the true ITT population as the baseline number, and might well carry out its own data analysis.

One reason why so many patients in Eliana (and in CTL019’s lymphoma study Juliet) discontinued pre-infusion was that the relatively long manufacturing process opened up a window in which their disease progressed. This had the added effect of enriching the populations for the fittest patients – something the adcom will also bear in mind.

3. Is efficacy down to bridging chemo/subsequent transplant?

Largely because of the long vein-to-vein time brought about by complex manufacturing, doctors in Eliana used bridging chemotherapy to maintain patients’ status while they awaited the CAR-T cells. This introduced a possible confounding factor: could subsequent responses have been due more to chemo than to CAR-T?

Moreover, given the problem of relapse, responding subjects frequently go on to receive a stem cell transplant. The adcom will similarly want to drill down into the actual benefit that a patient can expect to derive specifically from being put into a disease-free state by the CAR-T cells.

4. How controllable are CAR-T toxicities?

However the data are cut, the efficacy of CTL019 is clear. For the panellists, however, this must be balanced against the risk at which patients are put by undergoing such a procedure.

The side effects of CAR-T therapies are well documented, most often comprising cytokine release syndrome and neurotoxicity; moreover, these tend to correlate with efficacy and with baseline disease burden. The adcom will want reassurance that these toxicities are controllable as nurses have gained expertise with CAR-T.

Expect the panellists also to drill down into the precise numbers of grade 3 and 4 effects – and treatment-related deaths – that have occurred on all trials of CTL019 (Spotlight – Putting a number on CAR-T deaths, June 26, 2017).

5. How reliable is manufacturing?

Failure to produce CAR-T cells, often because a patient has insufficient viable T cells to begin with, was a problem with academic manufacturing. The panellists will want to be comfortable with Novartis’s claim that the commercial process has allayed many of the difficulties.

However, it is undeniable that CAR-T manufacturing is still an inexact science; the starting cell subpopulations are a game of chance, and cell expansion is largely hit and miss.

One presentation at last year’s Ash meeting concerned a retrospective analysis of relapses on a CTL019 study run by University of Pennsylvania, which analysed the integration of the CAR-T construct down to every last cell used. This found that in two patients the CAR had accidentally been inserted into a malignant B cell, which after losing the CD19 antigen grew out into a CTL019-resistant clone.

The fact that a single mistakenly transduced cell can cause relapse should be scrutinised at length by the adcom, which might well wonder how many other CD19-negative relapses that have not been analysed in this way might have been the result of similar manufacturing mistakes.

6. How close is manufacturing to the real commercial process?

Given the questions around manufacturing, the panellists will want to be confident that the manufacturing process used in Eliana was near or equal to that expected to serve the commercial setting. Most groups are aiming for full automation and much shorter production time, but none is there yet.

In Eliana CAR-T cells were produced at two plants – Morris Plains and a site in Germany – before being shipped to 25 hospitals in North America, Europe and Asia/Pacific (Ash – Novartis makes its case for driving a CAR into the real world, December 5, 2016).

7. Is there read-across from other CAR-T studies?

Panellists will have access to extensive data, not just from studies of CTL019. It is possible that manufacturing and data analysis methods, along with baseline patient characteristics, will be compared across trials.

And perhaps most important here is the cerebral oedema that led to the discontinuation of Juno’s JCAR015. Expect the adcom to review documented CTL019 side effects to ensure that none might have involved this serious toxicity; the FDA has separately talked of setting up a CAR-T safety database.

Other considerations

It is also worth pointing out two other issues crucial to the future of CAR-T therapy: whether a treatment priced at up to an expected $600,000 per patient per year can be viable, and whether bridging patients to transplant, rather than providing an actual cure in its own right, is sufficient.

These two issues are, of course, closely connected (Vantage Point – CAR-T value calculation in the firing line, June 27, 2017). But it is the adcom’s duty only to decide on the risk-benefit equation, irrespective of a treatment’s commercial potential; and pricing is, of course, outside the remit of the US FDA.

Without knowing who the adcom’s members are – they have yet to be announced – it is difficult to speculate on the depth into which panellists will go into these issues. Difficulty in convening a panel entirely free of conflicts might yield a somewhat generalist group of paediatricians.

Be that as it may, just five years after a striking response in the first CTL019 patient changed CAR-T from an academic curiosity into one of the most promising oncology treatments, the approach is on the verge of hitting the big time.

To contact the writer of this story email Jacob Plieth in London at [email protected] or follow @JacobPlieth on Twitter.

Jacob Plieth and the Stat journalists Adam Feuerstein and Damian Garde will take part in a live blog of the CTL019 adcom on July 12. Details will appear on the EP Vantage website.