Expert Financial Analysis and Reporting

Key Points:

• I think that an imminent press release dealing with topline results from an interim look at the ZUMA-1 trial will not have much meaningful data even though it is highly anticipated by investors. We will have to wait for the full data set that will be presented at American Society of hematology (ASH) in December 2016 to draw meaningful conclusions.

• I think that the major data point reported in the press release that investors will focus on will be the CR rate for patients as determined at three months after KTE-C19 cell infusion. I project that the rate will be between 25% and 35%. If I am correct, the upper end of this range would be considered satisfactory but probably a little below investor expectations and the lower end would be disappointing.

• Kite has strongly suggested that based on the interim look it will file a BLA by yearend and gain approval by mid-2017. Unless the CR rate is something over 40%, I think that management will begin to walk this back so that investors will shift to expecting approval in late 2017 or early 2018.

• I also think that investors have expected Kite to gain the first approval of a CAR-T drug in r/r DLBCL. However, Novartis is conducting a similar trial with its similar CAR-T drug (CTL 019) on the same timelines as ZUMA-1. I think that investors may change to the point of view that these drugs will be approved at about the same time in late 2017 or early 2018.

• I continue to feel that the $3 billion market capitalization of Kite is based on the most optimistic of possible scenarios. Over the balance of 2016, I expect that investor expectations will come down. I would not own the stock at these prices.

Investment Opinion

Kite will be reporting interim results of the phase 2, ZUMA-1 trial of its lead product KTE-C19 by the end of September. See my report “A Detailed Analysis of the Design of the All-Important ZUMA-1 Trial” for a detailed description of that trial. The most bullish analysts are projecting a very positive outcome in this trial that will result in approval in 2017 and sales of as much as $1 billion of sales by 2020 and are using terms such as “medical miracle” to describe KTE-C19. There is a lot riding on ZUMA-1. Kite is expecting dramatically positive results from this interim analysis. It has suggested that it will file a BLA by yearend based on the interim data and that KTE-C19 could be approved in mid-2017.

I would point out this is the first trial that Kite has ever conducted. It is a small, open-label phase 2 trial without a control group that will enroll 112 patients. This type of trial design is often used to create information to allow the design of a larger, controlled, randomized phase 3 trial. However, the FDA has approved drugs based on similar trial designs if they showed striking efficacy in aggressive cancers without effective drug treatments. For example, the checkpoint modulators Opdivo and Keytruda were approved for some indications based on similar types of trials.

Despite keen investor anticipation, the press release will give only limited information on the results in this trial. The full data set will be presented at the ASH in December 2016. I think that the most important number that investors will focus on in this press release will be the CR rate at three months post infusion of KTE-C19. So what are expectations for this data point?

The trial is designed on the expectation that 40+% of patients will achieve a CR at about six months after cell infusion. Kite indicated that the CR rate for seven patients treated in the phase 1 part of ZUMA-1 was 57% at three months, but a more rigorous analysis suggests it was 38%. I looked at data from an NCI trial for a drug very similar to KTE-C19 and also at data from a trial by Novartis with its directly competitive product CTL 019. I concluded that these small single center trials suggested a CR rate of perhaps 35% to 40% might be expected. However, as I explain in detail later in this report I think that the interim results for the multi-center ZUMA-1 trial will not be as good. I am projecting a 25% to 35% CR at three months after cell infusion. I think that if the CR at three months is 35%, investors will view it as positive or slightly disappointing. If the CR is 25% over less, it would be very disappointing.

The CR rate has to be looked at along with the duration of response. The CR rate at three months may be significantly higher than at six months or a year. My view is that the duration of response will have to be maintained six months at a minimum for the CR to be considered clinically important. I believe that a minimum, FDA will want to see the CRs maintained for six months. As previously noted, the press release will report the CR rate at three months. Along with most investors, I would expect the rate of CRs to go down over time. I do not expect anything meaningfully new to be reported on side effects.

Summing it all up, I think the data that will be reported in the press release may have frustratingly little meaningful information. We will have to wait for ASH to get more informative data on the trial.

Overview

Kite has said that it will report data by late September for its lead drug KTE-C19 in the phase 2 part of ZUMA-1. Kite is currently selling at a $3 billion market capitalization which is noteworthy for a company that has yet to complete a clinical trial. Investors have been excited by the early phase 2 results for CAR-T drugs similar to KTE-C19 at the NCI and academic centers. Small, single center studies have shown striking therapeutic efficacy in certain hematological cancers albeit with severe, sometimes life threatening side effects. Many in the medical and investor communities consider KTE-C19 and other CAR-T drugs to be paradigm changing. However, there have been no large and well conducted trials that validate this belief.

This data stems from an interim look at the first 51 patients treated in the r/r DLBCL cohort of the phase 2 part of ZUMA-1; the full trial is not scheduled to complete until March 2017. Kite’s expectation is that patients treated with KTE-C19 will achieve a 40+% CR rate. Kite also believes that the median duration of response will be six months or more. If these two parameters are achieved Kite believes that KTE-C19 will be approved. This is an open label trial without a control arm so the data must be compared to historical data from other trials. The statistical plan hypothesizes a CR rate of 20% for current standard of care. However, the SCHOLAR-1 meta-analysis conducted by Kite suggests that standard of care may only produce an 8% CR.

Timing of Data Release

Kite has said that it will issue a press release on topline results for the phase 2 part of the ZUMA-1 trial in late September 2016. This will be followed by a presentation of detailed data at the American Society of Hematology (ASH) meeting in December 2016. ASH is a very important forum for releasing trial results to the key opinion leaders in hematology. ASH has a policy of restricting the amount of information on a clinical trial that a company can issue in press release prior to presenting the data at ASH. Because of this restriction, there may be some uncertainties and differences in interpretation of data by investors. It is probable that Kite will only provide data on: (1) overall response rate, (2) complete response rate, and, (3) information on toxicity, particularly death or major adverse events. While this is important information, the key to understanding the data is in the details that won’t be forthcoming until ASH.

The release will include data on the first 51-patients enrolled in the r/r DLBCL cohort of the study. I understand that the reported response rates will reflect the best response to therapy over the initial three months of follow up on an intent to treat basis. Some patients enrolled in the trial may not receive an infusion because their disease worsens before the cell infusion or because the CAR-T production process fails. These patients will be included in the denominator used in the calculation of CR and ORR.

Historical Results in Other Studies of r/r DLBCL

In a recently published report, I cited results from three sources. NCI did executed a phase 1/2 study that using a product that provided the technology for KTE-C19. This study looked at r/r DLBCL and other r/r CD 19 positive B-cell cancers such as PMBCL, CLL, TFL and others. Novartis has reported on a small trial with its CAR-T product CTL 019 which is very similar (possible the same) to KTE-C19. Finally, Kite reported data on seven patients treated with KTE-C19 in the phase 1 part of ZUMA-1 which is the only human data on this product. Here are highlights from those studies.

• Using a source who had no relationship with Kite, I gained access to data relating to the NCI trial as of August 15, 2014. It gave efficacy results on a combined group of 17 r/r DLBCL, r/r PMBCL and TFL patients. The CR rate for this group was 35%, but the data was not broken out individually for the three cancers so we don’t know the CR for r/r DLBCL.

• In Kite’s 2015 10-K they reported that the response rate for 14 patients in the same NCI trial with r/r DLBCL and r/r PMBCL was 43% (6 of 14). Again there was no breakdown for r/r DLBCL patients versus r/r PMBCL. The duration of response was 7 to 9 months for responders.

• In the phase 1 part of ZUMA-1, Kite stated that 57% (4 of 7) patients with r/r DLBCL experienced a CR. Three of these patients had an ongoing response at nine months, but the response was brief for the other. Eight patients were actually enrolled in the trial, but Kite did not include the results of one patient whose disease progressed too rapidly and couldn’t be infused with cells. On an intent to treat basis the number enrolled would have been eight instead of seven. A more conservative calculation would be that 3 of 8 (38%) experienced a clinically meaningful CR.

• Novartis reported on fifteen r/r DLBCL patients who were treated with CTL 019. Seven of fifteen or 47% experienced a CR three months after cell infusion.

These are very small data samples and one must be cautious in drawing conclusions, but we still have to make a judgement. Kite said that the CR rate was 57% in the phase1 component of ZUMA-1, but for reasons I just explained I think that 38% is a better number to use. The NCI data is muddled because it doesn’t break out r/r DLBCL from other cancers but the 35% CR rate is probably in the ballpark for r/r DLBCL. The Novartis data was based on three month response rates and it is possible that the CR rate will drop as time goes on so I think that 47% is too high. My interpretation is that this data suggests a CR of 35% to 40% was achieved.

Why the CR Rate Results in the Phase 2 Component of ZUMA-1 Could Be Less Than 35% to 40%

There are a number of reasons to expect that the CR rate for phase 2 of the ZUMA-1 trial will be less than 35% to 40% as just estimated.

• The NCI and Novartis studies were done at single centers which were experienced in the use of CAR-T therapy. ZUMA-1 will enroll patients at 25 to 30 centers many of which have never treated a patient with CAR-T. There is always unintended bias with single center studies in selecting patients and reporting data. This results in a phenomenon well understood by investors that results from multi-center trials seldom (never) replicate the results from single centers.

• The response rates in the pivotal portion of the ZUMA-1 trial will be evaluated on an intent to treat basis. The trials I talked about were less rigorous in that investigators would exclude patients who might be unevaluable for any number of reasons. For example, Kite in reporting phase 1 results from ZUMA-1 did not include a patient who progressed too rapidly after chemotherapy to be given a cell infusion. Such a patient would be evaluated in this trial because it is based on intent to treat. There is also a likelihood of cherry picking patients most likely to benefit and omitting those less likely to.

• The number of patients treated in these phase 1 trials is so small that it is difficult to have strong confidence in the results. Just a shift of a patient or two could have a significant effect on the CR rate.

There is no rigorous way to estimate what the CR rate will be in the data about to be released. One can only make an educated guess and mine is that the CR rate will be on the order of 25% to 35%.

What Is A Reasonable Expectation For Safety?

It is unlikely that Kite will provide much detail on the safety profile of KTE-C19. They would likely have to release information on the incidence of mortality if any and any unusual or unexpected adverse events. Otherwise the PR may just include a blanket statement to the effect that KTE-C19 AEs were similar to those observed in the past. For many of the centers participating in this trial, they will be using CAR-T therapy for the first time. Because of this and because of the known propensity of this therapy to cause severe life threatening side effects, investigators likely will err on the side of caution and this is likely to lead to more side effects being reported than in earlier single center trials.

In reporting safety results for the phase 1 portion of ZUMA-1, Kite reported that the incidence of > grade 3 side effects from cell infusion was more than 35%. I think this is misleading in two ways. First of all, it excludes side effects that are related to chemotherapy pre-conditioning which probably causes more frequent and more severe side effects than cell infusion. Second of all, it doesn’t give any information on grade 4 side effects that are life threatening and require hospitalization. I think that Kite will follow the same non-transparent approach in the upcoming data release. We will have to await the presentation at ASH to get an accurate view of the side effect profile of KTE-C19 which is its Achilles heel.

When Will A BLA Be Filed?

Kite has suggested that it may file a BLA by yearend based on the data from this interim look at 51 patients. By yearend, the last patient treated (72^nd) in the r/r DLBCL cohort would have been followed for about 5 months and the first patient for about 17 months. The therapeutic benefit of KTE-C19 is based primarily on obtaining a CR and maintaining it (durability of response). The FDA will probably want to see data on as many patients as possible and for as long as possible which would bias the agency away from approving the drug on interim data.

While Kite can submit a BLA based on this interim data, this does not mean that FDA would accept or act on it. My view is that FDA will want to see data on all patients and will want to have some conviction on durability of response. This would argue for letting the trial run to conclusion in March 2017 and might lead to a BLA filing in 3Q, 2017.

Another issue that FDA may consider is that Novartis also has a trial in r/r DLBCL with its similar CAR-T drug CTL 019 on the same timeline as ZUMA-1. It seems to me that FDA will want to see the data from both trials to see the similarities and differences. This also makes the agency less likely to act on ZUMA-1’s interim data.