Avid Bioservices Inc (CDMO)

[keep_trying]

Well here is something for those interested in events around September 24 to check out. Posters may recall that I indicated I was looking at the Kaplan Meier plot that Dr. Gerber presented that showed the 2nd line NSCLC results for the Docetaxol control arm and the two Bavi arms, side by side. A few others were doing the same and shared their insights. I subsequently completed that assessment and now offer two graphics for critical review by posters on this board, keeping in mind that PPHM has already affirmed on September 24 that the KM plot data includes coding errors and should not be used for investment decisions.

One graphic is a resolution enhanced copy of the trial KM plot (not my work) to make the censored patient tick marks more discernible. The other is a spreadsheet graphic I prepared by taking readings off the enhanced KM plot and citing status for each patient for each of the three arms in the plot indicated time progression (order of time since initiation of treatment).



9/7/2012 Gerber/Shan K-M Graph – Enhanced so can see censored marks easier…


In essence, I extracted an NSCLC Bavi trial patient status spreadsheet from Dr. Gerber's September 7 presentation KM plot data. Note that this spreadsheet may include some KM plot data read errors, so posters that are interested are encouraged to compare the graphic with the spreadsheet to see if they can call out such things and share them with this board. I think I read things right, but......

The KM plot was enlarged enough to overlay scaling and take readings for patient status to the nearest 0.1 months. A cross check was also done by using the math for balancing the number of patients who died plus survivors at each point on the timeline, knowing that the increment on the Y axis when a patient dies would be 1 divided by the number of survivors. That calculated fraction should coincide with the Y axis shift for each patient's status on the time since initiation of treatment time line.

I used the term "survivors", which would be the sum of patients that had already posted they were censored before a patient's logged point on the time line plus the number of patients still alive at that point. Over time, all the alive patients eventually became censored on the KM plot such that censored equals survivors at completion of the KM plot period. Deaths plus survivors equals number of patients in the trial arm. This cross check tool proved very useful in assigning which tick mark censored patient belonged in which trial arm when the early month plots were overlaying upon themselves. The trial arm that needed a censored patient to balance the math cross check broke any tick mark assignment ties.

There is also a question about which patient in the count represents the median patient for the overall survival plot when there was an even number of patients in the control arm. The 1 mg/kg Bavi arm included 40 patients, raising question whether the 20th patient breached the MOS or the 21st. Posters can see that the break doesn't matter that much, since the Control arm shows it was MOS evented fairly quickly while both of the Bavi arms were one or two patients away from eventing at timing of the July 2012 indicated last update timing from Dr. Gerber's September 7 presentation in Chicago.

Since the median selection essentially splits the data pool in half, I looked at the data grouping for notworthy items in the first and second half of the trial arm patient group. It takes some study "soak time" of the data for the significance of data table information to stand out, which is why, I presume, that the Kaplan Meier plots are the common vehicle for communicating such trial data.

Best wishes and IMO.
KT

Some items I found noteworthy for call outs from the KM plot derived table:

1. Data was read from the Kaplan Meier plot and rounded to the nearest 0.1 months. While data cross checked against total trial arm counts, visual reads of the graphics could have introduced errors.
2. The most progressed censored patient in the control arm is at 14.6 months, in contrast to the 16.8 months and 18 months for the 1 mg/kg and 3 mg/kg arms respectively.
3. Median Overall Survival evented in the control arm, per Dr. Gerber and Shan analysis, when the 19th death was recorded at 5.6 months. Reading the 20th death for MOS would given an MOS of 6.1 months, but censored patients before then could change MOS status.
4. There were three censored patients in the Control Arm within the 5.6 month MOS period such that if those particular patients died before 5.6 months, the MOS could be shortened to as low as 4.7 months. However, the control arm MOS cannot exceed 5.6 months (6.1?) if those three patients remain lost to follow up or survive past 5.6 months
5. At the July 2012 latest update timing cited in Dr. Gerber's September 7 presentation of KM plot data in Chicago, there were 27 patients of 38 in the control arm noted as deceased and 11 censored. There were no control arm patients reported as alive at month 15 or 18, suggesting that the control arm data was mature before 15 months. However, an update of KM plotting for the control arm beyond July, such as for August and September, might be doable by PPHM when they issue the post investigation results. Two more months of trial progression data should enhance contrasts between the control and Bavi arms. Also, Clinical Trials.gov indicated that the Bavi NSCLC trial started and stopped recruiting between October 2010 and October 2011 providing the rough time period during which a patient would have started participation on a time line plot.
6. 8 of the last half (20 through 38) of patients were posted as censored in the control arm, the remaining 11 having died. 3 of the first half (1 through 19) patients in time progression were posted as censored in the control arm, the remaining 16 having died. At last reporting (July 2012) the control arm had 11 censored and 27 deceased patients indicated on the KM plot.
7. The 1 mg/kg Bavi arm has 40 patients, making the 21st (20th?) patient's death cause for the MOS having evented. As of the July update, only 19 patients in the 1 mg/kg Bavi arm had died, requiring at least one or two more patient deaths to be documented to reach an MOS endpoint similar to that reached in the control arm. However, PPHM reported the interim MOS as 11.1 months, which correlates with when 17 patients in the 1 mg/kg Bavi arm had died. Patients in the first half timing order (1 through 20 of 40) had 14 deceased and 6 censored patients, indicated at timing of 9.2 months. Considering the blanking out of censored patients for interim analysis, PPHM estimated the 1 mg/kg Bavi arm MOS at 11.1 months.
8. If the 1 mg/kg Bavi arm were to confirm all six patients that were censored in the first half grouping as having deceased, the MOS could be as short as 9.8 months. If just three of the six censored patients in the first half grouping continued to survive or be censored beyond the PPHM indicated MOS, the MOS would extend beyond Dr. Gerber's interim 11.1 months MOS.
9. The 3 mg/kg Bavi arm has 39 patients, making patient number 20 that dies the basis for establishing the MOS. The 20th patient in the time sequence from the KM plot is at 7.6 months, but the arm has not reported the 20th deceased patient yet as of the end of July 2012 update. At the July update, the 3 mg/kg arm had reported only 19 deaths, leaving one more death being needed to bracket around the indicated MOS. 12 of the first 20 patients were deceased, with 8 patients censored. The last half grouping (patients 21 through 39) added 12 more censored patients with 7 additional deaths.
10. Considering the censored patients, PPHM has estimated the 3 mg/kg Bavi arm MOS at 13.1 months. This corresponds with when 16 patients had died and 16 patients were censored. If all eight censored patients in the first half of patients plotted on the KM plot died without surviving past the indicated time, the final MOS could be as short as 7.6 months. However, if only three or less of the Censored patients died before 13.1 months, the MOS would extend beyond the 13.1 months identified by Dr. Gerber for the interim KM plot.