Sunesis Pharmaceuticals — Update 22 June 2016

Sunesis released results at the European Hematology Association (EHA) meeting in June 2016, detailing the results from a Phase I/II trial of Qinprezo (vosaroxin) in combination with Dacogen (decitabine, Otsuka) in patients with AML and high-risk MDS. This is an ongoing open-label investigator-sponsored trial being conducted by MD Anderson Cancer Center. The target population of the trial is previously untreated patients over 60 unsuitable for intensive chemotherapy. A total of 63 patients were enrolled in the trial, the majority (56, 89%) with AML and with a median age of 69.

Dacogen is currently only approved in the US for the treatment of MDS, although it is approved in the EU for newly diagnosed AML patients over 65 with unfavorable cytogenetics (see our initiation report on Sunesis for a discussion of the Dacogen EU approval).

The trial was initiated with patients receiving 90mg/m2 Qinprezo twice per cycle in combination with Dacogen at 20mg/m2 per day in a five-day cycle. However, the Qinprezo dose was reduced to 70mg/m2 following a high rate of grade 3 mucositis (four out of 16 patients in the Phase II portion of the trial). Interestingly this lower dose showed substantially increased median survival compared to the higher 90mg/m2 dose (Exhibit 1). The 70mg/m2 dose almost tripled the median survival for these patients to 16.1 months. The median survival for patients on the same regimen of Dacogen alone was 7.7 months.1 23 of the 41 patients in the 70mg/m2 cohort are still alive and being monitored and the median overall survival for this group could potentially increase.

Although the difference in median survival between doses was substantial, the results were not entirely surprising considering the patient population selected for this trial was not suitable for chemotherapy, and although these drugs are more tolerable than the intense 7+3 regimen (cytarabine and daunorubicin) standard of care for AML, adverse events are not uncommon. Consistent with this, the rate of adverse events during the trial was high (Exhibit 2). A majority of patients (82%) experienced mucositis, although the reduction in dose appeared to reduce the rate of grade 3 or 4 mucositis from 25% in the 90mg/m2 arm to 17% at the lower dose. The rate of early death (at less than eight weeks on study) was also substantially reduced by going to a lower dose: 27% on 90mg/m2 arm compared to 5% on 70mg/m2 (Exhibit 3). This difference in death rates is likely treatment related and could account for the difference in overall survival between the two doses.

Another element of the safety profile is the high rate of grade 3 and 4 infections observed during the trial. 56% of patients on the trial had major infections, while 3% and 10% had fungal or other infections. Infectious disease is common in patients with these disorders because by their very nature the treatments are myelosuppressive. In previous studies, Dacogen alone resulted in 61% of patients acquiring grade 3 or higher of pneumonia or bronchopneumonia, compared to 48% in patients treated with the standard of care low-dose cytarabine.1 Similarly, during the Phase II REVEAL-1 trial of Qinprezo, 24 (45%) patients got sepsis, 28 (25%) got pneumonia, and 10 (20%) got other infections of grade 3 or higher on 72/90mg/m2 respectively.2

The response rates observed in the trial were better than either Qinprezo or Dacogen alone in similar populations (Exhibit 3). These previous trials only examined patients with AML, although because the fraction of patients with MDS in the current trial is small, we believe this comparison is appropriate. Although detailed breakdowns of response rates that are directly comparable to previous studies are currently unavailable for the two different dosing regimens, the combined complete response (CR) rate was 49% for the current trial compared to 25-30% for Qinprezo alone or 16% for Dacogen alone. When incomplete platelet responses (CRp) are included in this analysis, the rate increases to 66%, which is higher than the combined rates for Qinprezo and Dacogen alone, suggesting the potential of synergy.

Recently AbbVie presented data on the use of its Bcl-2 inhibitor Venclexta (venetoclax) in combination with Dacogen or Vidaza (azacitidine) for the treatment of a similar patient population at the American Society of Clinical Oncology (ASCO) meeting in June 2016.3 The Phase Ib/II trial focused on the front-line treatment of AML patients unfit for intensive therapy with intermediate or poor-risk karyotypes and found an overall response rate (ORR, CR + CR with incomplete marrow recovery [CRi] + partial response [PR]) of 76% in this population (n=34). By comparison in the Sunesis trial, patients with intermediate or high-risk karyotypes (n=36) showed a 72% response rate with a more stringent definition of ORR (CR + CRi + CRp). We find these results to be comparable to the extent that parallels can be drawn between the two studies, considering the Qinprezo study includes a small number of MDS patients.

Sunesis announced during the presentation of these results that it intends to perform a follow-up Phase III study comparing Qinprezo plus Dacogen to both Qinprezo plus cytarabine and the 7+3 regimen in frontline AML. A program or programs of this design could potentially be used to support approval in the US or EU for a frontline indication in elderly patients, although we see the approval environment for AML in Europe as significantly more forgiving considering previous approvals for Vidaza and Dacogen for AML there.

The most impactful upcoming event expected for Sunesis is the decision from CHMP regarding the approvability of Qinprezo in Europe for relapsed and refractory AML patients over 60. We expect the decision around YE16, but, although unlikely, variations in the length of clock-stop periods during the EMA review process could delay the decision. Based on the precedent of Dacogen and Vidaza approval in Europe, we have a favorable outlook (60% chance of approval) for the ultimate decision (see our previous report for a detailed comparison). It is unlikely, however, that there will be additional feedback from the EMA or the company regarding the proceedings until the CHMP decision.

We additionally expect TAK-580 results to be reported by partner Takeda in H216, and the initiation of the SNS-062 Phase I/II clinical program around YE16.

We are increasing our valuation to $156.2m or $1.80 per basic share ($1.46 per basic share), from $150.1m or $1.74 per basic share ($1.41 per diluted share). We have increased the probability of success in front-line AML in the US (from 20% to 25%) and Europe (from 40% to 45%) because we find the Dacogen combination data encouraging and the proposed three-arm Phase III follow-up trial could provide an additional route to approval. We see the US approval as higher risk because of a history of approval denials for AML including for Dacogen and Vidaza, whereas in Europe Dacogen and Vidaza have been approved for classes of AML. This increase in success probability is partially offset by an increase in the predicted spending for the trial of approximately 50% or $9m to support the additional arm (accounted for under US and EU frontline AML lines in our NPV). The remaining changes to our model reflect the advancement of our NPVs to Q216 and the change in net cash.

Sunesis ended Q116 with $40.0m in cash and marketable securities and $11.7m in debt following the April 2016 debt offering and refinancing of existing debt. Results for the quarter were largely within our estimates, and we are not updating our 2016 numbers at this time except to reflect Q116 results and an updated share count. We have increased R&D spending in 2017 and onward by approximately $1.5m per year to reflect the increased cost of the proposed front-line AML development plan, partially offset by reductions in other R&D spending unallocated to individual programs. We still predict that the company will need $95m in additional financing to reach profitability in 2021, and record this as illustrative debt in our model ($30m in 2017, $30m in 2018, and $35m in 2020).