Kazia Therapeutics — GDC-0084 – encouraging initial efficacy

Kazia Therapeutics is an Australian biotechnology company focused on oncology drug development, listed on both the ASX (KZA) and Nasdaq (KZIA). It has two products in development: GDC-0084 for brain tumours (glioblastoma) and Cantrixil for refractory ovarian cancer. This note focuses only on GDC-0084 as important new clinical data have been presented.

Kazia presented a poster on the Phase IIa (NCT03522298)1GDC-0084 dose and initial cohort expansion results at the annual meeting of the Society for Neuro-Oncology (SNO), held from 20–24 November 2019. These data are from the first stage of the study, which was dose ranging and found a maximum tolerated dose of 60mg as opposed to 45mg determined by Genentech in the original Phase I study.2 Patients take GDC-0084 daily as 4, 15mg capsules in one dose in the morning.

The data presented here are from Stage 1 (September poster). Patients all had surgical tumour debulking and combined radiation and temozolomide (TMZ) therapy using the Stupp regimen (Stupp et al, 2005).3 Stage 2, involving about 20 patients, is recruiting. Data on PFS and OS from the Stage 2, cohort, plus information on the activity of GDC-0084 if taken with food (currently it is taken before food), will be available in 2020. The extent of patient survival governs the time taken to read out data: longer survival means longer to read out.

These results show a very encouraging survival profile in eight evaluable patients (out of nine) with median PFS of 8.4 months (Exhibit 3).All these patients are effectively resistant to standard TMZ chemotherapy as they produce the enzyme O6-methylguanine-DNA methyltransferase (MGMT) because they have an unmethylated MGMT gene promoter. MGMT repairs the DNA damage done by TMZ making patients resistant to that treatment (Jiapaer et al, 2018); about 61% of glioblastoma patients are TMZ refractory. If GDC-0084 can retard brain tumour growth in TMZ refractory patients, it could find a widespread, standard role.

In a large study, Hegi et al (2005) looked at TMZ resistance due to MGMT. Patients with the active form of the gene could repair TMZ damage and had a median PFS of 5.9 months. Patients with a deactivated MGMT gene did not repair the TMZ damage to their DNA so their tumour cells had a greater response to the drug and the patients showed a longer median PFS of 10.3 months.

The small number of patients in the Stage 1 part of the GDC-0084 Phase IIa study means that the median PFS cannot be contrasted reliably to previously published figures. Clearly, these early data look better than the 5.9 months in TMZ-resistant patients but as yet are below the 10.3 months for TMZ-susceptible patients with methylated, MGMT promoter Hegi et al (2005). The data are very encouraging and merit a full comparison study. Stage 2 will yield more robust data.

The important aspect for patients and their families – we need to remember that all these cold statistics have tragic individual stories behind them – is survival and quality of life. The Kaplan Meier OS curve (Exhibit 4) shows good survival in this small group with two deaths of eight cases reported. Of these, the two longest survivors were disease free after more than 15 months. As glioblastoma is retarded by GDC-0084 not killed by it, at least directly, this implies possible longer-term chronic use for a proportion of patients; if correct, this is commercially important as it might generate a large base of daily users by keeping them alive.

The OS curve (Exhibit 4) does not reach the 50% probability level, it sticks at 71%, so median survival cannot be determined. In their larger study, Hegi et al found that active MGMT gene patients, as studied here, had a medium overall survival of 12.7 months. Again however, the Stage 1 cohort is small so firm conclusions cannot be drawn.

There are many factors behind overall cancer survival, not just MGMT. For example, younger patients survive better than older ones. However, the median age for any brain cancer is 59 (SEER) and the Stage 1 cohort contained 55% of patients less than 59 years old so the sample does not seem skewed; in fact there was one patient in his 80s. Another key factor is the tumour size before surgery: under 42mm has a better prognosis than over 42mm; the tumour sizes of the patients in the study are not available. The EORTC website has a detailed consideration of such factors. There are also other mutations that have an impact on survival outcomes;4 these are more common in younger patients and were not controlled in this study.

The crucial aspect of this result is that it enables Kazia to progress into a pivotal trial with a higher level of confidence. The current Phase IIa Stage 2 cohort will provide useful additional data. The Stage 1 data itself, while encouraging, has no randomised comparison group as a control – it was dose ranging – and, as management notes, historical controls are not necessarily a good guide to the outcome in such small patient groups.

Nonetheless, given the recurrent and aggressive nature of glioblastoma and the selection of patients who are more resistant to TMZ (due to the unmethylated MGMT promoter) as part of study inclusion criteria, we are impressed by the outcome so far.

The trial now being planned to start in H120 will compare GDC-0084 to adjuvant TMZ therapy and could be pivotal. This is planned by Kazia to be a randomised, two-arm study with 228 patients. The study will compare maintenance therapy with GDC-0084 vs standard-of-care TMZ. Patients will first undergo surgery to remove the bulk of the tumour followed by a radiation therapy combined with TMZ (Stupp regimen). Patients will be then randomised to receive maintenance therapy with either GDC-0084 or TMZ. TMZ is normally given for only six months due to toxicity.

The primary endpoint is likely to be either progression-free survival or overall survival. The average of five literature median PFS and OS values in GBM patients with an unmethylated MGMT promoter is PFS of 5.2 months and an OS of 13.8 months (June 2019 note).

Normally, the FDA requires two pivotal studies. In a small indication like glioblastoma, which is poorly controlled by current therapies and has short survival times, one well-run trial could be enough for approval if the data are robust and the product has a safe profile.

Kazia is also evaluating, using a series of alliances, the use of GDC-0084 in metastatic brain cancer and childhood disease, Exhibit 5. This spreads the clinical and financial risks, but such third-party studies can become prolonged.

Some of these are structured as initially dose ranging with an ‘expansion’ cohort. One study (NCT03765983) is in breast cancer metastases combined with Herceptin but at a lower 45mg dose than the 60mg now used by Kazia. The Memorial Sloan Kettering (MSK) sponsored academic study, is a combination with radiotherapy. Radiotherapy is standard in brain tumour surgery. However, 30–50% of patients still progress.

The St Jude paediatric trial (NCT03696355) uses GDC-0084 two to three months after radiotherapy; this trial has completed dose escalation and has a maximum tolerated dose of 27mg/m2 . For example, a 20kg child has a body surface area of 0.7m2 so the dose would be 18.9mg compared to the 60mg adult dose.

Outside brain cancers, other companies are developing PI3K inhibitors (Zhao et al, 2017). No other PI3K inhibitor can penetrate the brain, so only GDC-0084 could be used for brain cancers, but the class as a whole has shown anti-cancer efficacy. For example, Aliqopa (copanlisib, Bayer) is approved as an iv infusion for the treatment of adult relapsed follicular lymphoma and Zydelig (idelalisib, Gilead) is approved as an oral treatment for chronic lymphocytic leukaemia (CLL) and follicular lymphoma. Copiktra (duvelisib, Verastem) was approved in October 2018 again for leukaemia (CLL) and some lymphomas. In 2019, the FDA approved Piqray (alpelisib, Novartis) as an oral combination therapy with fulvestrant for metastatic breast cancer therapy for patients with the PI3KCA mutation. This is the same mutation as in the Kazia-MSK trial. The approval was based on a progression-free survival of 11 months vs 5.7 months on fulvestrant alone. In glioblastoma, Roche/Genentech has a Phase I PI3K pathway inhibitor (Ipatasertib5) in a combination dose and safety trial (NCT03673787, data, if released, Q121). No other PI3K inhibitor brain cancer studies are in progress according to the clinicaltrials.gov database.

Our assigned clinical probability of success for GDC-0084 is 25%. We have retained this because the Stage 1 data, although highly encouraging, are based on too few patients to be conclusive. The further patients in the expansion cohort will firm up the results. Our base case valuation of A$137m assumes a GDC-0084 market launch in 2024 following a single pivotal study. This scenario also assumes that GDC-0084 is out-licensed to a marketing partner in 2023 -24. In this scenario, global sales for GBM reach US$1,050m in 2030. Please see earlier notes for detailed assumptions. This base case is now equivalent to A$1.91/share (A$1.81/share including dilution) after the share placement in October increased shares in issue from 62m to 72m. Kazia is also listed on Nasdaq under the code KZIA, with each Nasdaq-listed ADR representing 10 ordinary shares.

There is an alternative scenario of a GDC-0084 launch in 2026 rather than 2024 if a further pivotal study is needed for FDA approval. On revised assumptions, this scenario valuation becomes A$86mor A$1.19/share (undiluted).

Kazia had A$5.4m cash on 30 June 2019, Exhibit 6. The operational FY19 cash use was A$6.8m after R&D tax rebates of A$1.4m. We still project higher cash costs in FY20 and FY21 as a pivotal trial is being planned and will need to be funded. Currently, Kazia does not have partnering income but good follow-on data from Stage 2 of the current trial could enable a deal.

In late October, Kazia completed a A$4m funding, A$3.8m net, by issuing 10 million new shares at A$0.40 each. We estimate Kazia might spend A$15–20m from mid-CY20 onwards to finance the large GDC-0084 study planned to start probably, in H1CY20 - although most clinical costs will occur from H2 CY20. Following the equity raise in October 2019, we have assumed long-term debt of a minimum A$11m in FY21 in our forecasts (formerly A$15m), but this could be from a mix of equity, partnering or grants. We have not assumed funding for a Phase II Cantrixil study.