Provenge PhIII Trials – The Alternative Explanation of Survival Results

One-page Summary

 Placebo patients lost, on average, 73% of their circulating mononuclear cells through each apheresis/infusion
procedure. At least 50% of apheresed cells were lost in purifications steps at Dendreon‟s manufacturing facilities
and a further 2/3 were kept back from patients and frozen.
 Placebo patients lost these 73% of cells three times at 2 week intervals.
 The human body has several mechanisms to restore overall T-cell counts, and this lymphocyte removal would
have no consequences in healthy individuals through most of their lives. But the dramatic age-related changes in
the immune system that occur around the age of 65-70, imply that the replacement T cells in circulation are not of
the same subtype or diversity as the T cells that were removed.
 There is strong evidence to suggest that both t-cell diversity and the t-cell subtype which the elderly are impaired
in replacing, play a central role in tumor suppression.
 It is therefore unclear whether the 4 month survival difference between the arms in the three Provenge PhIII trials
can be attributed to Provenge efficacy or should be attributed to harm inflicted on the “placebo” patients by the
removal of a vital part of their innate cancer-fighting defenses.

 If placebo immunodepletion is indeed the intervention responsible for the survival difference between the groups,
it would imply that Provenge treatment itself is harmful to patients due to the >50% of their cells lost during
Provenge manufacture, and is shortening, not extending the lives of prostate cancer patients. There is evidence
that this immunodepletion might reduce survival by >5-6 months. Until it can be proven that immunodepletion
did not harm placebo patients, there is risk to every prostate cancer patient being exposed to this treatment.

There are many features of the Provenge PhIII trial data that are unexpected/inexplicable if the 4-month survival
difference is attributed to Provenge efficacy. All of these features would be expected if the survival difference is attributed
to immunodepletion being harmful to cancer patients >65 yrs of age:

 Patients younger than 65 from any arm of the trials had a median survival of 28-29 months, compared to 23.4
months for the Provenge patients over 65, and 17.3 months for the placebo patients over 65. Age is not prognostic
for survival in prostate cancer. These data are mystifying if Provenge “works”. The dramatic immune changes that
occur at age 65-70 suggest that patients <65 could still recover from the „placebo‟ immunodepletion without
obvious consequence, whereas older patients‟ immune systems could not recover to their previous state, and with
this change lost much of their cancer-fighting ability.
 The absence of survival difference between treatment arms in patients <65 (~one quarter of all patients) is
inexplicable if Provenge works by its proposed mechanism (unless they had better prognosis, for which there is
no evidence publicly available).
 Comparison of median survivals and baseline prognostic factors in the Provenge asymptomatic and minimally
symptomatic mCRPC trials to the 3 most recent major trials in all-stage CRPC, strongly suggests that the placebo
arm in particular should have lived longer than they did.
 This same comparison also suggests in particular that the early (<1 year) death rate in placebo patients was much
higher than would have been anticipated given the early stage of their disease.
 Provenge has never been shown to have anti-tumor effects either in vivo or in vitro despite a decade of efforts to
shed light on how it might be working. Thus to believe Provenge works is to admit that our understanding of the
immune system in cancer is extremely limited. But the person making this admission cannot therefore
simultaneously claim that immunodepletion did not damage the placebo group. The proposed mechanism by
which immunodepletion could result in shortened survival, though by no means „proven‟, is at least consistent
with our best and most recent scientific research on immune aging and cancer immunology.
Document Overview:
1. Placebo Arm active intervention p2
2. Explanation of Immune Aging and Cancer p6
3. Evidence in Trial Data does not favor Provenge efficacy p10
4. Summary & Conclusions p21
 Appendixes p22

1. Placebo arm Immunodepletion = NO Placebo

All three of the “placebo-controlled” phase III studies for Provenge had the same „placebo‟ design. First,
patients in both arms underwent 1.5-2 blood volume mononuclear cell leukapheresis 1, 2. Assuming 5.25L of
blood (men in this study were a little above-average of 5L), this equated to 8-10.5 L of blood leukapheresed.
For the drug arm Provenge was prepared using all of the cells collected in this manner and infused in its entirety
back into the patients. Placebo patients, however, had 2/3 of their leukapheresed cells frozen (in order that they
could be offered Provenge manufactured from the thawed cells upon disease progression) and received back
only 1/3 of the cells that had been removed by leukapheresis 3, 4. Trial participants underwent this procedure 3
times at 2-week intervals, each time with the placebo patients receiving only 1/3 of their cells back.
Dendreon has provided cumulative cell product parameters administered in the pooled Phase III trials 5. As
shown in the table, the median TNCs (total nucleated cells) infused into Provenge patients over 3 doses was
9.8x109, whereas for placebo patients it was 3.4x109.

Approximately 96% of these nucleated cells were shown to be T cells, B cells, monocytes, and natural killer
cells (Appendix A). Thus placebo patients were given a median of 6.19 x109 fewer lymphocytes/monocytes
back than Provenge patients.

1
Small et al, Placebo-Controlled Phase III Trial of Immunologic Therapy with Sipuleucel-T (APC8015) in Patients with Metastatic,
Asymptomatic Hormone Refractory Prostate Cancer. J Clin Oncol (2006) http://jco.ascopubs.org/cgi/content/full/24/19/3089
2
Small et al, Immunotherapy of Hormone-Refractory Prostate Cancer with Antigen-Loaded Dendritic Cells J Clinical Oncology, 2000
http://jco.ascopubs.org/cgi/content/full/18/23/3894
3
March 2007 CTGT Advisory Committee Meeting, Sipuleucel-T Briefing Document
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4291B1_01.pdf
4
Higano et al, Integrated Data From 2 Randomized, Double-Blind, Placebo-Controlled, Phase 3 Trials of Active Cellular
Immunotherapy With Sipuleucel-T in Advanced Prostate Cancer. Cancer 2009;115:3670–9
5
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p56:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
Further, as detailed in Appendix B, over 65% of nucleated cells were lost in the two buoyant density
centrifugation steps that transform the raw apheresis product received at the Provenge manufacturing facility
into the finished product that is re-infused into patients (this loss of cells applies to both Provenge and placebo
patients).
This loss, added to the 2/3 of their cells that are frozen, mean that the median placebo patient is in fact losing
7.8x109 lymphocytes/monocytes 3 times at 2-week intervals.
Mononuclear cell leukapheresis procedure spins blood at just such a speed as to preferentially precipitate the
desired lymphocyte and monocyte fraction, leaving most of the neutrophils and other granulocytes in the blood
re-entering the patients. So although Provenge and placebo infusions were 96% pure infusions of lymphocytes
and monocytes, these cells only constitute approximately 30-40% of the white blood cells in circulating blood.
Dendreon provided patient baseline white blood cell counts 6:

Using the 5.25 L estimation for the volume of blood in this population, the men in this trial can be calculated to
have had an average of 6x103 * 5.25x106 circulating white blood cells at baseline, or 31.5x109. Assuming 66%
of these to be granulocytes, patients had an average of 10.7x109 circulating lymphocytes/monocytes at baseline.
Thus, on average, placebo patients had 7.8x109 of their baseline 11x109, or 73%, of their circulating
lymphocytes/monocytes removed with each treatment cycle. Thus over 3 such cycles, placebo patients
unwittingly underwent the removal of a significant number of the baseline T cells, B cells, natural killers cells
and dendritic cells upon which they were singularly dependent to fight their prostate cancer.
n.b. these calculations are based upon medians, and individual patients will have undergone far greater and far
less cell removal than this median.

Nobody has questioned the potential impact of this imbalance immunodepletion

It is clear in the 160+ FDA documents relating to the Provenge BLA, from the 2007 panel transcript and from
conversations with investigators that the potential impact of this immunodepletion has never been considered.
There are many studies speaking to the inability even of young, healthy volunteers to fully recover from
repeated immunodepletion:

 Wright et al. Lymphocyte depletion and immunosuppression with repeated leukapheresis by

continuous flow centrifugation. Blood 1981;58:451-8.

6
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p28:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
  Kolf et al. Sustained decreases in lymphocyte counts in serial long-term leukapheresis donors.
Transfusion 2003;43:S28A-9A.
 Koepke et al. The safety of weekly plateletpheresis: effect on the donors’ lymphocyte population.
Transfusion 1981;21:59-63.
 Senhauser, et al. Immune system changes in cytapheresis donors. Transfusion 1982;22:302-4.
 Strauss et al. Effects on donors of repeated leucocytes losses during plateletpheresis. J Clin Apheresis
1994;9:130-4.
 Heal, et al. Long-term follow-up of donors cytoapheresed more than 50 times. Vox Sanguinis
1983;45:14-24.
 Wolf et al, Leukapheresis for the extraction of monocytes and various lymphocyte subpopulations
from peripheral blood. Vox Sanguinis, 2005
 Strasser et al, Mononuclear cell variability and recruitment in non-cytokine-stimulated donors after
serial 10-liter leukapheresis procedures. Transfusion, 2005

In the most recent of these studies by Strasser et al, using leukapheresis machines such as those in use in the
Provenge trials and still today, 13 young, healthy men underwent four 10L leukapheresis procedures at 2-week
intervals. The average 1010 MNCs dispensed from these individuals matched exactly the 9.9x10 9 MNCs in the
apheresis product dispensed from patients to make Provenge (see Appendix B for details).
2 weeks after the last LP procedure, the men in the trial had statistically significantly lower MNC counts:
 CD3+ T cells (x109/L) 1.29 (before) 1.00 (after) -22.5% (p=0.006)
 CD3+4+ T cells (x109/L) 8.20 (before) 6.66 (after) -18.8% (p=0.03)
 CD3+8+ T cells (x109/L) 3.89 (before) 3.21 (after) -17.5% (p=0.04)
 CD19+ B cells (x109/L) 2.74 (before) 2.27 (after) -17.2% (p=0.24, NS)
Only 5 of the 13 donors (in either case) were available for 2- and 3- year MNC counts (n.b. not the same 5
donors in each comparison to baseline). Compared to the first predonation count of the study, they found:
 CD3+ T cells diminished by 25% at 2 years (p=0.05) 20% at 3 years (p=0.31)
 CD3+4+ T cells diminished by 18% at 2 years (p=0.08)
 CD3+8+ T cells diminished by 35% at 2 years (p=0.07)
 CD19+ B cells were only lessened by 3% at 2 years and 12% at 3 years.

Although the “placebo” patients in the Provenge studies received 28% of these cells back 3 days after each
removal, and underwent one fewer such procedure than the men in the Strasser trial, it is clear that we do not
know the potential impact of this intervention on overall survival.

While the circulating cell loss might not appear dramatic in the 40-somethings in this study, the age-related
changes in the immune system are dramatic (details in next section), and the impact of this immunodepletion on
the elderly patients with mCRPC in the Provenge studies could potentially have been catastrophic.

Consequence of placebo being an active intervention

In 1999 when investigators and the FDA agreed upon the trial design, scientists did not yet have much of the
understanding that we have gained over the last 10 years on the age-related changes in the immune system.
Leukapheresis is performed at blood donation centers on hundreds of volunteers who have agreed to
specifically donate white blood cells and, despite the significant cell count declines 2 years later shown in the
study cited, has resulted in no negative consequences sufficiently obvious to have been noticed to date. For
most of a person‟s life, as the body‟s mechanisms to replace cells of all types are quite robust. But dozens of
experiments are now showing the profound changes that occur in the immune system around the age of 65-70,
and showing the decline in the body‟s ability to replace certain cell subtypes after this age. It will be shown,
perhaps even in the next few years, if immunodepletion by leukapheresis does indeed, as one might expect, alter
the immune systems of the elderly in ways that it does not impact younger people. Any alterations involving the
cellular subtypes that are involved in identifying or killing cancer cells have obvious implications for the
potential harm that leukapheresis might pose to elderly patients fighting metastatic tumors. Certain studies
ongoing today could even be directed to this purpose 7.

While the placebo harm cannot be “proven”, it would explain (as shown below) all of the unusual features of
the PhIII trial results. These self-same features cannot be plausibly explained in the context of Provenge
efficacy, and have either been dismissed as extreme chance occurrences, not considered at all, or glossed over
as consequences of our poor understanding of cancer immunology. Lacking any other factors that speak to the
relative claims of each intervention to have caused the survival difference, scientific and logical first
principles dictate that placebo harm is the more likely intervention to have resulted in this dataset.

Until it can be proven that placebo patients were not harmed, the studies upon which Provenge was approved
are invalid and we have no proof this drug is beneficial. Furthermore, if this immunodepletion is accountable
for the shorter survival of placebo patients, it follows that Provenge treatment is also harmful since they lose at
least 75% of the number of cells taken from placebo patients due to the large number of cells lost in Provenge‟s
manufacturing process. There is data to suggest Provenge intervention could shorten life expectancy by
greater than 5.5 months.

The burden of proof lies with the company that designed and ran the trial and the agency that approved
the protocol: they must prove that placebo immunodepletion was not harmful before they can justify
claims of Provenge efficacy. Until then, treating a single further patient would be unethical and a
liability. And until then, taxpayers should not be made to pay for a single Medicare patient undergoing this
intervention.

7
http://clinicaltrials.gov/ct2/show/NCT00104325 , http://clinicaltrials.gov/ct2/show/NCT00073060?term=apheresis&rank=7 ,
http://clinicaltrials.gov/ct2/show/NCT00067054?term=apheresis&rank=11
2. Immunodepletion in Elderly Cancer Patients
To gain a better understanding of the potential that immunodepletion could have significant consequences in
elderly patients with solid tumors, I have drawn upon the thousands of studies in the published literature that
form the foremost frontier of our current understanding of the processes in question.

The potential mechanism I shall outline focuses on T-cells. I fully acknowledge the separate roll that each of the
cell types lost in immunodepletion might play that could also provide plausible mechanisms, but since T cells
are arguably those that have received the most research attention and whose mechanisms in cancer surveillance
we understand the best, as well as those with well-defined age-related changes, I believe they are a justifiable
focus. I acknowledge that the immune system is highly complex and our understanding of it is very far from
complete. Thus, there are many other possible parallel or more important mechanisms in other cell types which
could explain the link from lymphocyte depletion to impaired cancer-specific immunity.

With a plausible mechanism described, one can look to the many features of the trial data and see whether these
can be more plausibly explained by one explanation for the survival difference than the other (Provenge
efficacy or imbalanced cell depletion).

T Cell Populations and Homeostasis

 Diversity is the Key to Immune System Strength: Generation and maintenance of a massively diverse
repertoire of T cell antigen receptors is essential for the immune system to be able to respond to the universe
of potential antigens, yet the number of distinct T cell receptors (TCRs) expressed by the estimated 1011-1012
T cells in the human body is not known. Using TCR gene amplification Arstila et al8 have shown that the
naïve cells are massively diverse, while the memory subset are orders of magnitude less so.
 Activation by its corresponding antigen causes a T cell to proliferate and respond: T cells are activated by
TCR recognition of a short linear peptide anchored within the peptide-binding groove of a major
histocompatibility complex (MHC) molecule present at the surface of an antigen-presenting cell (APC). The
best known of these are dendritic cells (DC), although monocytes/macrophages, B cells, and neutrophils are
also able to present antigens. Upon activation, naive T cells produce IL-2 and other cytokines, and enter
differentiation pathways resulting in the development of various T cell subsets. An effective T cell response
requires significant proliferation of the activated cell to accomplish the clonal expansion critical for
combatting the source of antigen.
 After antigen is destroyed, a few T cells retained as memory T cells: At the height of such a response, the
antigen-specific T cells may account for 25% of the population in the spleen and even more locally9.
Although most have a life-span of only a few days, a small pool will survive as long-lived memory cells.
Thus, controlled cell death (apoptosis) restores the numerical balance in the immune system, but results in a
small shift in the ratio of naïve to memory T cells since the retention of memory T cells allows a more
effective response to rechallenge by the same pathogen. Memory T cells possess different stimulatory
requirements than naïve T cells, but both naïve and memory cells respond less effectively with age.
 New naïve T cells only made in the thymus: The generation of T cell receptor diversity depends entirely on
the production of new, naive T cells in the thymus 10. Stem cells seeded into the thymus from the bone
marrow perform a random rearrangement of the mini-genes that determine the single specific TCR that that

8
Arstila et al. A direct estimate of the human αβT cell receptor diversity. Science. 1999
9
Butz et al, Massive Expansion of Antigen-specific CD8 t cells during an acute virus infection, Immunity (1998)
http://jco.ascopubs.org/cgi/content/full/18/23/3894
10
Nossal et al, Negative Selection of Lymphocytes - Cell (1994)
 cell then expresses. After the acquisition of CD4 or CD8 molecules needed for recognition of MHC class 1
and 2 cell surface markers on APCs these cells are then released into pool of circulating naïve cells. From
this pool, some cells are selected by antigen to proliferate.
 Expansion of Naïve cells in periphery increases to compensate for thymus decline with age: After its
central importance in establishment of a diverse T cell repertoire, the thymus has been demonstrated to
decline in function and output with age. After the ages of 40 to 50 years old, virtually the entire T-cell
supply is generated from clonal expansion of existing naïve and memory T cells. The replicative stress
associated with repeated clonal expansion can result in shortened telomeres and cellular senescence and lead
to phenotypic changes that decrease the functionality of these important cells. Both mechanisms contribute
to failure of the elderly to respond to new antigenic challenges such as cancers, poor vaccine responses and
increased morbidity with newly arising infections, such as is seen with antigenic shift or drift of the
influenza virus.
 While absolute T cell numbers are kept constant, relative proportions of subsets change: Homeostasis of
naïve and memory populations is most probably regulated independently11,12 coupled with preferred survival
of recent thymic emigrants13. By maintaining two pools of T cells with a very different repertoire, the
immune system combines two conflicting needs: a recognition of a wide array of novel antigens and an
efficient and timely response to previously-encountered antigens.
 Naïve counts decline and memory counts increase with age: Antigenic challenge through life leads to the
progressive contraction of the pool of naive T cells. Consequently, the peripheral T cell repertoire of elderly
persons (aged 70) consists predominantly of oligoclonal memory cells, and the overall T cell diversity is
significantly reduced relative to that seen in younger people. Naïve T cells have a half-life of 6 to 12
months14 15.
 Gradual decline throughout life collapses suddenly around age 65-70: Contraction of T cell receptor
diversity is not a linear process with age. A diverse naïve CD4+ T cell compartment is maintained for
decades by slow naïve cell replication, but this only partly compensates for declining thymic output16. This
period is followed by a dramatic and sudden collapse of diversity leading to a severely contracted repertoire
after the age of 70 years17.
 Most naïve T cells will be found in the blood or lymph, not in tissues populated with memory t cells: The
extent to which naive T cells migrate through the tissues is quite limited (Mackay, 1991). Their preferred
migratory pathway is from the blood back to the lymphatic tissue, where the chance of a rare antigen-
specific cell meeting its antigen is optimal. Most of the T cells that do wander into the tissues are memory
cells. As their numbers have been considerably increased by prior antigenic activation and clonal expansion,
they can afford to patrol the body widely to nip any reinfection in the bud.

11
Tanchot et al, Differential Requirements for Survival and Proliferation of CD8 Naïve or Memory T Cells, Science (1997)
12
Tanchot and Rocha, The organization of mature T-cell pools Immunol. Today (1998)
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1692750/pdf/10794050.pdf
13
Berzins et al, The Role of the Thymus and Recent Thymic Migrants in the Maintenance of the Adult Peripheral Lymphocyte Pool-
J. Exp. Med. (1998) http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2212318/pdf/97-2072.pdf
14
Hellerstein, Measurement of T-cell kinetics: recent methodologic advances, Immunol. Today 20 (1999) 438–441.
15
Macallan et al, Rapid turnover of effector-memory CD4(+) T cells in healthy humans, J. Exp. Med. 200 (2004)
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2212011/pdf/20040341.pdf
16
Wallace et al, Direct measurement of T cell subset kinetics in vivo in elderly men and women, J Immunol 173 (2004)
http://www.jimmunol.org/cgi/content/full/173/3/1787
17
Naylor et al, The influence of age on T cell generation and T cell receptor diversity, J Immunol 174 (2005)
http://www.jimmunol.org/cgi/content/full/174/11/7446
 T cell homeostasis and age18. The schematic diagram, based on published data, illustrates how
parameters of T cell homeostasis change with age. sj TRECs, as a marker of thymic output, show a log
linear decline with age. Peripheral turnover is high in the newborn, steady during adulthood, and increases
late in life. TCR-b chain diversity in the naive CD4+ compartment is maintained up to the age of 65 to 70
years, after which it collapses. The memory compartment has about tenfold less diversity in TCR-b chains
and also contracts at an older age.

 There is much evidence that Immune aging is related to increased incidence of cancer: A direct effect of
immunosenescence on cancers can be easily conceptualized and many experimental data seem to support
this contention. In cases of failure to clear tumors, it is hypothesized that one reason may be decreased
immunosurveillance. This idea is derived from the findings that tumor regression was observed in
immunocompetent hosts while cancer incidence increased in immunocompromised hosts.
 T cells central to recognising and killing cancer cells: It is now established that the immune system has
cells, particularly CD8+ cytotoxic T lymphocytes (CTLs), that can recognize tumor antigens and kill
tumors19, 20. Nevertheless, a major problem is that these T cells are either not induced or only weakly

18
Goronzy, T Cell development and receptor diversity during aging, Current Opinion in Immunology, 2005
19
Schreiber, H (1999) in Fundamental Immunology, Tumor Immunology, ed W.E.Paul (Lippincott-Raven Publishers, Philadelphia),
pp 1237–1270.
 induced, i.e., the T cells are not evident in the systemic circulation. One possibility is that there is
inadequate tumor antigen presentation by dendritic cells for eliciting T cell immunity 21. Another is that
tumor-reactive T cells are tolerized by the tumors19, 22.
 Evidence that Cancer mutation allows it to escape T cells that have recognised it: Cancer cells are
genetically unstable. This often leads to the selection of tumor variants that can escape the destruction by
CTL. Thus, a large established tumor may contain cells with various genetic alterations. When CTL are
applied to tumors, tumor variants that can not be recognized by CTL will prosper. Another related
mechanism by which the tumor can evade recognition is the loss of the tumor antigen recognized by CTL 23
24
. A third evasion mechanism has also been demonstrated.
Bai et al 25 have seen evidence for this phenomenon in mice. They frequently observed recurrences of
tumors in mice that had initially responded to therapy with high numbers of transgenic Tcells. Analysis of
the P1A antigen on the recurrent tumors revealed a variety of mutations in the P1A epitope. These mutations
abolished T cell recognition of the tumor cells, demonstrating antigenic drift of tumor antigens as a
mechanism for tumor evasion of T-cells in vivo.
 New naïve T cells required if tumor has evaded prior T cell responses: A least in mice, T cell response to a
the first antigenic epitope encountered can be rendered ineffective by selection of tumor antigenic variants,
which would require activation of a new naïve t cell to combat it. The iterative nature of this recognition and
evasion puts a singular emphasis on the naïve T cell pool in mounting an effective cancer response.
 Hypothesis of T Cell destruction of cancer also central to proposed mechanism for Provenge Several
novel strategies are being explored to induce tumor-specific T cell immunity. DC vaccination, as being
pursued by Dendreon, is one of these. The results of many of these studies speak to the importance of T
cells, and in particular naïve T cells, to mount an effective CTL-mediated tumor response. Immature DCs
capture antigens but lack full T cell–stimulatory activity26. In the presence of appropriate stimuli, such as
inflammatory cytokines, the DCs mature. DCs upregulate T cell adhesion and costimulatory molecules as
well as select chemokine receptors that guide DC migration into lymphoid organs for priming of antigen-
specific T cells. The use of DCs as adjuvants is supported by many animal experiments with primarily
mature DCs27. These studies have shown that the injection of tumor antigen–loaded DCs reliably induces
tumor-specific CTL responses, tumor resistance, and in some cases, regression of metastases27.
 Experimental evidence for proposed mechanism: Using a defined DC vaccine combined with detailed
immunomonitoring, Thurner et al28, provided proof that vaccination with mature DCs expa nds tumor-
specific T cells in advanced melanoma patients. In addition, they found some evidence for the direct
interaction between CD8+ CTLs and tumor cells as well as for escape of antigen-negative metastases.

20
Van den Eynde, B.J., (1997) T cell defined tumor antigens. Curr. Opin. Immunol. 9:684–693, pmid:9368778
21
Schuler, G., R.M.Steinman(1997) Dendritic cells as adjuvants for immune-mediated resistance to tumors. J. Exp. Med
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2199101
22
Pardoll Cancer vaccines - Nat Med. 1998 May;4(5 Suppl):525-31.
23
Yee el at, Adoptive T cell therapy using antigen-specific CD8+ T cell clones for the treatment of patients with metastatic melanoma:
in vivo persistence, migration, and antitumor effect of transferred T cells. Proc Natl Acad Sci U S A. 2002
http://www.pnas.org/content/99/25/16168.full
24
Uyttenhove et al, Escape of mouse mastocytoma P815 after nearly complete rejection is due to antigen-loss variants rather than
immunosuppression. J Exp Med. 1983 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2186952/pdf/je15731040.pdf
25
Bai et al - Antigenic drift as a mechanism for tumor evasion of destruction by cytolytic T lymphocytes, J Clin Invest, 2003, 111,
1487-96. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC155049/pdf/JCI0317656.pdf
26
Banchereau, J., (1998) Dendritic cells and the control of immunity. Nature. 393:245–252, pmid:9521319.
http://www.rockefeller.edu/labheads/steinman/pdfs/1998-nature.pdf
27
Lotze, M.T., (1999) in Dendritic CellsBiology and Clinical Applications, Dendritic cell therapy of cancer and HIV infection, eds
M.T.Lotze, A.Thomson (Academic Press, San Diego, CA), pp 459–485.
28
Thurner et al, Vaccination with mage-3A1 peptide-pulsed mature, monocyte-derived dendritic cells expands specific cytotoxic T
cells and induces regression of some metastases in advanced stage IV melanoma, J Exp Med. 1999
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2195739
3. Provenge Trial data explained by Immunodepletion

1. Exceptional Survival of all patients <65 is unprecedented

Patients <65 in the pooled data had median survivals of 28-29 months, compared to 23.4 and 17.3 months
for the Provenge and placebo groups in the >65 population. In no other mCRPC trial has age been shown to
be a prognostic factor, let alone one of the magnitude implied by these data.

In TAX327 (as shown in subset survival charts in bullet #3 below), the 504 men aged ≤68 had 17.6months
median survival and the 502 men aged ≥69 had 18.1 months median survival. Younger age clearly did not
confer a survival benefit to these patients. Halabi et al also did not find age to be a prognostic factor in their
analysis of many large mCRPC trials.

Since we have no evidence of anti-tumor effects of Provenge, these data could be looked at differently:

Removal of 54% of circulating Removal of 67% of circulating

T cells (x3 at 2 week intervals) T cells (x3 at 2 week intervals)
n Survival N Survival
<65 106 29.0 months 66 28.2 months
≥65 382 23.4 months 183 17.3 months

This inexplicably long survival in patients younger than 65 in both arms, several months longer than the
survival of the Provenge arm, truly demands plausible mechanistic explanation. No such explanation can be
given accompanied by the proposition that Provenge is efficacious. [it has never been disclosed that the
<65s enrolled had better prognosis than the >65s, which would be highly improbable since these are 25% of
the patients enrolled and age is not correlated with prognosis].

This is exactly what would have been anticipated if the proposed mechanism for immunodepletion harm in
patients over 65 was in effect during the trial:
 The patients under 65 whose immune systems could recover from immunodepletion (whether the larger
# of cells lost by the placebo patients or the smaller number of cells lost by the Provenge patients)
without harm, lived out the relatively long 28-29 months that reflected the good their early,
asymptomatic baseline disease characteristic. Patients younger than 65 still could have enough diversity
in both t-cell compartments, and sufficient residual ability to replace lost naïve t-cells that they are able
to recover to a state of reasonable immune competence after their immunodepletion.
 The patients over 65 that lost ~54% of their circulating T cells in each of 3 Provenge interventions, lived
23.4 months.
 The patients over 65 that lost ~67% of their circulating T cells in each of 3 placebo interventions, lived
17.3 months.
In the older population that have undergone the dramatic immune deterioration which occurs at 65-70 years of
age, it could be proposed that the loss of lymphocytes in the immunodepletion harmed their immune systems in
a way from which they could not restore their t-cells anywhere close to where they would need to be to mount
any meaningful anti-tumor response.
More specifically, as proposed, the few naïve t-cells remaining in the elderly appear to reside preferentially in
the circulation & lymph. With thymic output at its nadir, and clonal expansion of the few remaining t-cells both
slow and impaired, the greatest impact of the immunodepletion from which they cannot recover could well be
the particular loss of naïve number and diversity. As described above, there are several known mechanisms of
tumor immune evasion that necessitate the recognition by repeated new naïve T cells to mount an effective
response.

Since age is not prognostic for survival in prostate cancer, in the absence of any other prognostic data
to explain why these younger patients had a better survival1, the simplest implication of these data
taken at face value would be that lymphocyte depletion shortened survival in the placebo group
by ~11-12 months, and the less dramatic, yet still significant lymphocyte loss from the Provenge
group shortened their survival by 5-6 months.

2. Provenge did not show benefit in patients younger than 65

FDA Reviewer‟s comments:”In Study D9902B, there were a total of 126 subjects in the ITT population who
were <65 years of age; of these subjects, 77/341(23%) were in the sipuleucel-T arm and 49/171 (29%) in
 the placebo arm. In the subgroup of subjects who were less than 65 years of age, the observed hazard ratio
of 1.411 (95% CI: 0.869, 2.290) suggests a trend in survival in favor of the control group, compared to the
sipuleucel-T group.”
The reviewer then proceeds to an exploratory analysis of the pooled data from all 3 trials as shown below.

The reviewer draws this conclusion “The above analyses of the data from all three studies (D9901, D9902A,
and D9902B) support the hypothesis that the subgroup of subjects who were less than 65 years of age also
benefit from treatment with sipuleucel-T.”.

The pooled analysis does not, however, support this assertion. 29 vs 28.2 months survival does not indicate
benefit. It is clearly anomalous that no survival-benefit was seen in these patients, and cannot be explained
by any known mechanism associated with Provenge‟s proposed efficacy. It would be an extreme statistical
aberration if this was the result of chance in these 172 patients (23% of all patients).

The potential mechanism by which the differential immunodepletion between treatment arms might not
affect those patients under 65 is explained above.

3. Should patients have lived longer?

Direct comparison of all available patient baseline prognostic factors and their associated survival to these same
data from other CRPC trials should give an indication as to whether the placebo patients lived less long, or the
Provenge patients lived longer than would have been anticipated. This would speak directly to the relative
likelihood of each intervention as being more probably responsible for the survival difference.
Provenge study participants, defined as having “minimally symptomatic” or “asymptomatic” CRPC, were
specifically recruited to represent a patient population prior to the point in progression at which chemotherapy
would generally be initiated and has demonstrated survival benefit. The first 3 listed trials in which this
comparison was made were all testing chemotherapy regimens and enrolled CRPC patients with all levels of
symptoms. Thus by intent enrolled a population with more advanced disease and worse prognosis than the
Provenge studies.
For each of the 4 trials I have provided all the data on known prognostic factors that were available for both
populations.
TAX327
The TAX 327 study, compared docetaxel Q3w, docetaxel Q1w and mitoxantrone (all with prednisone) in men
with metastatic CRPC and demonstrated a median survival of 19.2 months for docetaxel q3w vs. 17.8 months
for docetaxel q1w vs 16.3 months for mitoxantrone. 29
Study enrolled March 2000 to June 2002, and involved centers in 24 countries. IMPACT enrolled from July
2003-October 2007. Stage migration would therefore favor IMPACT patient survival.

TAX327 IMPACT control arm

Taxotere Q3W Doc discretion/tax + pred30
n=335 n=171
Enrollment Criteria
Pain all Absent or minimal pain31
Opioids all No opioids allowed
Metastases Excl: brain mets Excl: lung, liver and brain mets
Life expectancy all At least 16 weeks
Baseline Characteristics
Age (median) yrs 68 70
Serum PSA (median) ng/ml 114 42
Gleason Score ≤ 7 42% [57%] 76%
8-10 31% [42%] 24%
n/a 26% -
„Substantial pain‟ * 45% 0%
Bone mets 90% 91.8%
Lung/liver mets 22% 0%
Survival Results
Overall Survival 19.2 months 21.7 months
* Pain PPI ≥ 2 or analgesic score >10

It is clear from the enrollment criteria and baseline characteristics that IMPACT‟s minimally-symptomatic
patients with very little pain, no visceral mets and low serum PSAs represent a far less advanced population
than TAX327 in which 45% of patients had substantial pain at baseline, 22% of patients had visceral disease
and baseline serum PSA levels and Gleason Scores were also higher.

29
Tannock et al: Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engl J Med 351:1502-1512, 2004
30
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p27:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
31
http://www.clinicaltrials.gov/ct2/show/NCT00065442?term=provenge&rank=1
This figure shows survival among various subgroups treated on the TAX 327 trial. At left are the subgroups defined in this exploratory
analysis the number of patients in each subgroup, and their median survival time in months, independent of treatment. At right is a
Forrest plot showing the median hazard ratios and their 95% CIs for survival on the docetaxel arms compared with the mitoxantrone
arm.

Across treatment groups:

 Patients with pain died on average 7 months earlier than those without. (45% vs 0%)
 Patients with visceral disease died on average 6 months earlier than those without. (22% vs 0%)
 Patients with PSA ≥ 115 died 5.6 months earlier than those with PSA below that level. (114 vs 42)

Taken together with the patient baseline characteristics (TAX327 vs IMPACT) repeated above in parentheses
and potentially favorable stage migration, one could find it surprising that IMPACT placebo patients lived a
mere 2.5 months longer than their far sicker counterparts in TAX327.

Above are the survival curves from the two trials (IMPACT on left): It is also unexpected, given their explicitly
earlier stage in disease progression and far better baseline prognostic factors, that as many patients (28%) had
died at one year in the IMPACT placebo group as had amongst the taxq3w patients (27%). Could this suggest
that there is an unusual early death rate in the IMPACT placebo arm?
CALGB 90401
The CALGB 90401 study comparing docetaxel + Avastin to docetaxel alone presented results at ASCO in June
and showed 22.6 and 21.5 months survival in the two arms, respectively (not significant).
CALGB90401 control arm IMPACT control arm
Docetaxel + prednisone32 Doc discretion / Docetaxel + pred33
n=525 n=171
Enrollment Criteria
Progression since last intervention Progression after hormonal therapy
Pain all Absent or minimal pain34
Opioids all No opioids allowed
Metastases No known brain mets Lung, liver and brain mets excluded
Cytologically-positive effusions all Pleural effusions or ascites excluded
Life expectancy all At least 16 weeks
Baseline Characteristics
ECOG status 0 55 %35 81.3 %
ECOG status 1 41 % 18.7 %
ECOG status 2 5% 0%
PSA (ng/dl) 85 42
Percent on opioid pain medication 35% 0%
No pain at baseline Data not public 52.6%
Lung/liver mets (~19%§ ) 0%
Survival Results
Overall Survival 21.5 months 21.7 months
PS 0 median survival 23.8
PS 1 median survival 17.2
PS 2 median survival <17.2
PSA ≤ 85 24.5
PSA > 85 18.4

§
In similar populations in TAX327, MSKCC and SWOG9916, 22%, 16% and 19% of patients, respectively, had lung and liver
mets at baseline, so these likely indicate an approximate representation in the CALGB population.

CALGB 90401 enrolled from April 2005 to Dec 2007. IMPACT enrolled over a similar timeframe from July
2003 to Oct 2007 (60% of patients enrolled after Nov 2005), so stage migration is not a confounder in this
comparison.

Investigators noted that OS with docetaxel/prednisone in this trial was longer than previously reported (21.5m
vs 19.2m in TAX327) and proposed it may be due to “stage migration or a good risk population (47% of
patients had 24 month predicted survival of >30%)” 32.

As healthy as these patients may have appeared to the investigators in comparison to the TAX327 patients, the
IMPACT “placebo” patients had undeniably better prognostic factors:
 ECOG 0 in 81% vs 55%. [amongst the CALGB patients, patients with ECOG 0 lived 6.6 months longer
than those with ECOG 1]

32
ASCO slides obtained from Roche/Genentech “CALGB 90401: A randomized double-blind placebo controlled phase III trial
comparing docetaxel, prednisone and placebo with docetaxel, prednisone and bevacizumab in men with metastatic castrate resistant
prostate cancer (mCRPC)”
33
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p27:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
34
http://www.clinicaltrials.gov/ct2/show/NCT00065442?term=provenge&rank=1
35
Break-down of ECOG 0,1 group as communicated at ASCO oral presentation
  PSA 42 vs 85 [CALGB patients below this median lived 6.1 months longer than those above it]
 Zero vs 35% of patients at baseline on opioid pain medications – cancer-related pain is considered a
powerful prognostic factor in metastatic prostate cancer: Halabi et al showed in a multivariable analysis
that pain interference predicted OS. In their study population, compared with men with lower pain
interference scores, the adjusted hazard ratio (HR) for death of men with high pain scores was 1.43
(95% CI, 1.17 to 1.74; P .001 – a 7.4month median OS difference) 36
 And IMPACT had no patients with visceral metastases, versus an estimated 19% for the CALGB
population, a prognostic factor which lead to a 6 month survival disadvantage for patients with
liver/lung metastases in TAX327.

Given the earlier disease stage and far better prognosis for IMPACT patients, why did they only live the same
length of time (21.7 vs 21.5 months) as the more advanced and sicker CALGB tax-only arm?

SWOG 9916
The SWOG 9916 study randomized 770 patients with progressive mCRPC in the United States to taxotere +
estramustine or mitoxantrone + prednisone daily. Taxotere demonstrated a 1.9 month survival benefit over
mitoxantrone (17.5 months vs. 15.6 months, P=0.02)

SWOG9916 enrolled patients from Oct 1999 to Jan 2003, while IMPACT enrolled from July 2003 to Oct 2007
(60% of patients enrolled after Nov 2005). Stage migration would favor IMPACT survival.

SWOG 9916 control arm IMPACT control arm

Docetaxel + estramustine37 Doc discretion / Docetaxel + pred38
n=338 n=171
Enrollment Criteria
Pain all Absent or minimal pain39
Opioids all No opioids allowed
Metastases Excl brain mets Lung, liver and brain mets excluded
Life expectancy all At least 16 weeks
Baseline Characteristics
Age (yrs) 70 70
PSA (ng/dl) 84 42
Bone pain > grade 2 36% 0%
Bone mets 84% 91.8%
Lung/liver mets 18% 0%
Survival Results
Overall Survival 17.5 months 21.7 months

The prognostic factors of pain, PSA and visceral metastases again favor IMPACT patient prognosis, and this is
borne out in their longer OS (a little spurious to claim with any conviction this should be even longer).

36
Halabi et al, Pain Predicts Overall Survival in Men With Metastatic Castration-Refractory Prostate Cancer, J Clin Onc, 2008
http://jco.ascopubs.org/cgi/content/full/26/15/2544
37
ASCO slides obtained from Roche/Genentech “CALGB 90401: A randomized double-blind placebo controlled phase III trial
comparing docetaxel, prednisone and placebo with docetaxel, prednisone and bevacizumab in men with metastatic castrate resistant
prostate cancer (mCRPC)”
38
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p27:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
39
http://www.clinicaltrials.gov/ct2/show/NCT00065442?term=provenge&rank=1
Yet given their vastly better survival prospects, along with the tailwind of stage migration, why have almost the
same proportion of patients (28%) died within one year as in the SWOG 9916 cohort (27%)? Asymptomatic
patients ought, intuitively, to outlive symptomatic patients quite decisively in the first 6-12 months as they must
spend some months progressing to the stage at which symptomatic patients enrolled.

Berry et al – Mitoxantrone in asymptomatic CRPC

Wanting to have at least one other trial in asymptomatic CRPC for comparison, below are the data from a Phase
III study of mitoxantrone + low dose prednisone vs low dose prednisone alone40. This study recruited from
March 1997 to Jan 1999, which makes D9901 (recruited Jan 2000 to Oct 2001) the most appropriate of the
Provenge PhIII trials to compare it to. Though numbers are small, in the absence of any better asymptomatic
data, they are noteworthy.

Mitoxantrone + D9901 Control Arm IMPACT control arm

prednisone Doc discretion (50% Doc discretion / Docetaxel
n=56 Docetaxel) + prednisone41
n=45 n=171
Enrollment Criteria Asymptomatic HRPC Asymptomatic HRPC minimally or
asymptomatic HRPC
Baseline Characteristics
Median Age 70 71 70
PSA (ng/dl) 56 47.9 42
ECOG status 0 75% 82% 81.3 %
ECOG status 1 23% 18% 18.7 %
ECOG status 2 2% - 0%
Bone mets 86% 91% 91.8%
Soft tissue mets 18% (lymph nodes only) 73% 49%
Lung/liver mets 6% 0% 0%
Survival Results
Overall Survival 23 months 21.4 months 21.7 months
12-month survival 82% 67% 72%

40
Berry et al. Phase III study of mitoxantrone plus low dose prednisone versus low dose prednisone alone in patients with
asymptomatic hormone refractory prostate cancer. J Urol 2002;168:2439-43.
41
Clinical Review, Sipuleucel-T, Review Completion Date 04/28/2010, p27:
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214540.pdf
In the D9901 „placebo‟ arm 50% of patients received Docetaxel, and 63% of patients received any
chemotherapy. D9901 patients had a slight edge over the mitoxantrone arm from all prognostic factors: lower
PSA, lower ECOG scores, absence of visceral mets, and a 3 year stage migration advantage. Combining this
prognosis with the fact that 50% of these patients received the only chemotherapy known to confer a survival
benefit, whilst none of the mitoxantrone patients did, it is quite unexpected to find that D9901 “placebo”
patients had a worse survival: 21.4 vs 23 months.

Given the above, the <12-month survival of 18% in these asymptomatic patients on mitoxantrone (and the
slower decline of their survival curves), makes the D9901 “placebo” patients appear as though they must be
harboring some exceptional ill-health that cannot be seen in their baseline characteristics, which would lead to
the demise of 33% within one year.

If those mitoxantrone patients in the 90s had the 2.9 month survival benefit of Docetaxel, might they have lived
at least to 25.9 months, and hint at the minimal level of survival we ought to be expecting from the unusually
healthy asymptomatic patients in our placebo arm?

Halabi Predicted Survival

Dendreon has presented the IMPACT survival data at ASCO, to FDA, to investors, and elsewhere directly
accompanied by the Halabi predicted survival for the two arms.

Survival Halabi Actual

(months) Predicted
Provenge 21.2 25.8
(n=341)
Placebo 20.3 21.7
(n=171)

These data have been used to illustrate the fact that Provenge patients lived longer than their predicted survival,
whereas placebo patients did not.
However, it has been shown in the original Halabi paper 42 that when testing the model in a validation
population, a predicted survival of 22.8 months corresponded with an actual survival of 27.2 months. The
model is admitted and demonstrated to increasingly underestimate survival of patients as they live beyond 14-
15 months.

One could extrapolate from this, therefore, that the observed 25.8 month survival in the Provenge group vs 20.3
month predicted survival cannot be cited in support of the notion that Provenge extended life.

Furthermore, the Halabi model was developed to predict survival in a mixed group of CRPC patients, and
cannot therefore be validly applied to an exclusively asymptomatic and minimally symptomatic population.
Given that the Halabi model uses neither pain, nor location of metastases as predictive factors, yet both are
known as highly correlated with survival, and the IMPACT population excluded both of these patient groups, it
is furthermore likely the Halabi model would systematically underestimate the survival of an asymptomatic
population excluding these negative prognostic factors.

4. The 4-month apparent “survival benefit” in PhIII trials is the only evidence of Provenge efficacy

Provenge has never shown anti-tumour effects either in vivo or in vitro (see excerpts from FDA review below).
Provenge has shown no efficacy in animal models. The only piece of evidence that Provenge prolongs survival
is the 4.2 month “survival benefit” vs the more immunodepleted arm in its PhIII trials.

Thus, invalidating the PhIII “survival benefit” invalidates the basis upon which the drug was approved and is
currently being used.

42
Halabi S, et al. Prognostic Model for Predicting Survival in Men With Hormone-Refractory Metastatic Prostate Cancer. J Clin
Oncol. 2003 Apr 1;21(7):1232-7. http://jco.ascopubs.org/cgi/content/full/21/7/1232
To quote the FDA reviewer 43: “The fact that they are able to get a response to PA2024, but consistently not to PAP
tumor antigen is troubling”
“Summary of immune monitoring. It is difficult to draw conclusions from the immune monitoring data for several
reasons. For some assays only a few patients were examined, and in cases where a reasonable number of patients was
examined, no response against PAP was seen. Only PA2024 seems to consistently generate an antigen specific response.
It is puzzling how the stimulation index data is presented. If the peripheral blood T cell response is really as high as they
are suggesting in some of their graphs, then that would represent a very strong response to PA2024. It is not clear why if
they can get that level of response why they do not see much more of a response to PAP. Dendreon intends to perform
some --b(4)-------------- assays and maybe --b(4)---------- for the ongoing D9902B study, but as yet have not performed
any of the assays. Hopefully, the results will be clearer in those studies. During a telecon with Dendreon to discuss some
of these figures that were included in Dendreon‟s advisory committee briefing package it was asked if they had any
evidence of a specific response to human PAP. They stated that, no, they do not yet have any evidence.”

To quote the CMC reviewer: “Dendreon performed a limited evaluation of the immune response to sipuleucel-T.
Immune monitoring was not performed in all trials. For example, no studies were done in D9902A, and so far it does not
appear that any analysis has been performed in D9902B, --b(4)-----------------------------------------------------------------------
------ Dendreon has stated that one limitation is in getting patient samples sent back to them for analysis. Another problem
has been in consistency in the T cell assays. The thought is that if more samples could be assayed at the same time, then
consistency may improve.
Even when immune monitoring was performed in a trial, it was not routinely performed and the number of patients
examined is somewhat small. Also, some assays appear to have only been done once. It is not clear that Dendreon has put
a high priority on measuring the immune response in patients in their trials. Considering that there appears to be very little
tumor antigen-specific immune response in the vaccinated patients, one would think that this would be a high priority.
Finally, it is difficult to fully interpret the data presented because few details were included on how the studies were
performed and figure legends were not provided.”44

43
CMC Review p74-75
http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214565.pdf
44
P 74 http://www.fda.gov/downloads/BiologicsBloodVaccines/CellularGeneTherapyProducts/ApprovedProducts/UCM214287.pdf
4. Summary and Conclusion

Each trial arm received a different active intervention, neither of which can be proven to have had no effect, but
one of which is likely to have had a greater role in determining the 4 month survival difference. The large
amount of data from the trials speaks to the greater likelihood of one intervention being the primary driver.

Provenge Intervention: Provenge confers a 4.2 month survival advantage in minimally symptomatic mCRPC
Immunodepletion Intervention: Placebo patients were immunodepleted by the repeated removal of 73% of
their lymphocytes & monocytes. Provenge patients were also immunodepleted, yet to a lesser degree, due to the
loss of >50% of their leukapheresed cells during Provenge manufacture. Immunodepletion of this magnitude in
patients >65 cannot be recovered from, causing in rapid tumor progression and death.

Immunodepletion
Provenge Hypothesis –
Hypothesis – placebo (and to a
Observation from Trial and Experimental Data
Provenge lesser extent
extends life Provenge) patients
were harmed
Patients <65 in both arms lived far longer than even the Provenge
group  
Provenge appears to confer no survival benefit to patients <65
(marked cliff in T cell diversity and naïve cell populations at 65)  
Median survival of the placebo group was less than would have
been predicted by comparing to other mCRPC studies, such as  
TAX327, SWOG9916 and CALGB 90401
Early deaths at ≤12 months in the placebo arm in particular appear
worse than extrapolation from these other trials would predict  
Provenge shows no in vitro or in vivo anti-tumor effects, and
showed no benefit in disease progression  

The balance of the data would appear to weigh in favor of differential immunodepletion as the cause for the
survival observed in this trial. There is no plausible explanation for trial data from reasonable hypotheses
regarding how Provenge might be working.

No future studies should be run with imbalanced immunodepletion of this nature. Efforts should be made to
reduce the number of cells lost during manufacturing processes similar to this one.
Appendix A

92% of Final Product Cells were lymphocytes and monocytes

Marker Cell Type % of Final

Product 45
CD3 T-Cells 63 %
CD19 B-Cells 6%
CD56 Natural killer 12 %
cells
CD14 Monocytes 15 %
? Other ? ? granulocytes? 4%
TNC 100%

45
March 2007 CTGT Advisory Committee Meeting CMC Briefing Document, p9
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4291B1_04a.pdf
Appendix B

Cells Lost During Provenge Manufacture

These steps are performed to remove stray neutrophils and red blood cells that are extracted from patients
despite the leukapheresis centrifuge having been set at such a speed as to spin down primarily the desired
mononuclear cell fraction (these unwanted cells will contribute to the incoming leukapheresis TNC count, yet
are likely only a few percent of the final product TNC count – data redacted from public document 46).

“to help describe and visualize the level of variation between individual patient leukapheresis units and between different lots
of the final product, the applicant presented the data in the form of box and whisker plots. Data plotted in this way is described
in terms of the median, quartiles of the meida, and minimum and maximum quartile extremes, with outliers plotted
individually”

46
March 2007 CTGT Advisory Committee Meeting CMC Briefing Document, p5 & p9-10.
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4291B1_04a.pdf
Dendreon redacted all the data on cell counts from the 2007 CTGT Advisory Committee Meeting CMC
Briefing Document. This is the only publicly-available source for this information. The redacted data
approximated below was be reverse-engineered from the two cell count charts in Figures 4 and 5 (reproduced
above) of the CMC briefing document. Despite the poor quality of the scan and the redaction of the left-hand
scale of Figure 5, by extrapolation from the known median cumulative CD54+ cell number in 3 doses of
Provenge (1.877x109), it can be deduced that the baseline is zero and the horizontal lines represent 2x10 9
increments.

Apheresis Product Provenge dose % loss to

Cell count Ω % of TNC ᴪ Cell count Ω % of TNC ᴪ Manufacturing
(x109) A (x109) B APH to FP
T-Cells - CD3 5.0 46 % 2.1 63% 58%
B-Cells - CD19 0.8 7% 0.2 6% 75%
NK cells - CD56 1.4 13% 0.4 12% 69%
Monocytes - 2.7 25% 0.5 15%
81%
CD14
Other 1.0 9% 0.15 4%
80%
(calculated)
TNC 10.9 100% 3.4* 100% 69%

Cells lost per Cells lost per %

Baseline WBCs intervention intervention difference
Provenge Patients Placebo Patients Provenge
% # in Cell % of Cell % of vs.
(approx) 5.25 L number Baseline number Baseline placebo
man (x109) C (x109) D
(x109)
T-Cells - CD3 17.0%** 5.36 2.9 54% 3.6 67% 24%
B-Cells - CD19 2.5%** 0.79 0.6 76% 0.67 85% 11%
NK cells - CD56 4.5%** 1.42 0.9 71% 1.03 80% 13%
Monocytes - 10.0% 3.15 2.2 70% 2.37 75% 8%
CD14
other 0.8 0.87
TNC 7.4 8.53
Ω ᴪ
from Figure 4, above from figure 5, above
C = A – B , D = C + (2/3 * B)
* This TNC # triangulates with the number in Table 19 on 2nd page of this write-up for TNC in product parameters
administered in safety database (9.831x109 in 3 doses). Small difference due to different data sets.
** lymphocytes total 24% of baseline WBC; triangulates with # in Table 8 on p3 of this write-up showing patient
baseline characteristics (1.425/6 cells per µl)

The only source for leukocyte counts during manufacture is the redacted 2007 CMC document. After these data
were submitted Dendreon ceased the monitoring of these parameters. The 2010 CMC Review has been redacted
to such an extent, it is not clear whether these data were re-examined during the final FDA review prior to
approval.

On p31 of the Briefing Document supplied by Dendreon to the 2007 review committee47 the volume of blood
undergoing leukapheresis is stated as “a standard 1.5 to 2.0 L blood volume leukapheresis procedure to collect
PBMCs”. This is incorrect. The “L” should not be there. The actual volume of blood was 1.5-2 times the entire

47
March 2007 CTGT Advisory Committee Meeting, Sipuleucel-T Briefing Document
http://www.fda.gov/ohrms/dockets/ac/07/briefing/2007-4291B1_01.pdf
body blood volume, equating to 8-10.5 L for these patients. This has been confirmed with investigators and is
correctly written without the “L” in the Journal of Clinical Oncology paper on the first PhIII study (9901) by
Small et al.48

These two features of the 2007 briefing documents made it very difficult to see in publicly-available
information the magnitude of the cell numbers being lost from Provenge and, to an even larger magnitude,
placebo patients.

Data on patient cell counts through the course of treatment were also not made publicly available, so no
assessment can be made of the ability of these elderly cancer patients to replace the lost cells. There were no
cases of leucopenia or lymphocytopenia reported in AE tables, suggesting that shifts in relative subtypes of
leukocytes (e.g. naïve vs memory T cells, T-cell receptor diversity, APC diversity, etc), and their
potency/competency, rather than absolute number, would be needed to see which of these might account for a
survival impact due to cell loss.

48
Small et al, Placebo-Controlled Phase III Trial of Immunologic Therapy with Sipuleucel-T (APC8015) in Patients with Metastatic,
Asymptomatic Hormone Refractory Prostate Cancer. J Clin Oncol (2006) 24:3089-3094
Appendix C

Summary of Phase 3 Clinical Studies

Study # Year of Primary Study Study Design Population Product, Dosage, # of Subjects
Patient Endpoints Route of
enrollment Administration,
and Schedule
D9901 Jan 2000 - TTP Placebo- Asymptomatic Sipuleucel-T or 127
~Oct 2001 OS was not a controlled, metastatic CRPC placebo, (82 Sipuleucel-T:
pre-specified double-blind, with a minimum of 45 Placebo)
endpoint multi-center, All Gleason 3x106 CD54+ cells/
randomized scores dose, i.v. at Weeks 19 clinical study
(2:1) 0, 2, & 4 centers
no cancer pain
D9902A May 2000 - TTP Placebo- Asymptomatic Sipuleucel-T or 98
~Apr 2002 (OS revised controlled, metastatic CRPC placebo, (65 Sipuleucel-T:
secondary double-blind, with a minimum of 33 Placebo)
(completion endpoint multi-center, All Gleason 3x106 CD54+ cells/
May 2005 – 3 following randomized scores dose, i.v. at Weeks
yr survival analysis of (2:1) 0, 2, & 4
announced D9901) No cancer pain
July 2005)
D9902B July 2003 OS Placebo- Asymptomatic or Sipuleucel-T or 512
40% enrolled Elevated to controlled, minimally placebo, (341 Sipuleucel-T:
IMPACT by primary double-blind, symptomatic with a minimum of 171 Placebo)
Nov 2005 endpoint in Nov multi-center, metastatic CRPC 20x106 CD54+
Gleason 2005 randomized cells/ dose, i.v. at 71 study locations
removed and (2:1) Minimum 40% Weeks 0, 2, & 4
minimal pain Gleason ≤7
allowed
Oct 2007 Absence/minimal
cancer pain

 May 2005 first time DNDN announced survival benefit (that would be presented at ASCO)
 August 2003 docetaxel demonstrated survival advantage
Appendix D –Q&A

How could Immunodepletion not cause an increase in infections?

(1) these people have cancer already, and die before infection gets them (2) innate immune system is the body‟s
first line of defense in defense against other organisms. The cells most involved in innate immunity (and those
first recruited to sites of infection), namely mast cells, eosinophils, basophils and the phagocytic cells including
macrophages and neutrophils are barely affected by the leukapheresis. Although dendritic cells are also
involved in innate immunity and through the process of antigen presentation they serve as a link between the
innate and adaptive immune systems.
Pathogens that evade the innate immune response are generally rare (often due to successful vaccination
programs that make use of the adaptive immune response memory), such as tuberculosis, salmonella typhi
(typoid), polio, smallpox, measeles, mumps, rubella and Bacillus anthracis (anthrax). Others, such as
staphylococcus and streptococcus are likely to have been encountered previously and be represented in the
memory T cell compartment. It seems unlikely if these pathogens had not been encountered previously that
these would be encountered for the first time by a study subject during the 6-18 month timeframe of the trial).

I heard leukapheresis is a treatment for cancer, why would they do this if it is bad for cancer patients?

Leukapheresis is only a treatment in cancers of the blood cells, never solid tumors. In the case of hematological
malignancies such as acute leukemias, there are white blood cell counts high enough to cause hemostasis and
"sludging" in the capillaries. This can effect retinal vasculature leading to vision changes, pulmonary
vasculature leading to shortness of breath from decreased efficiency in oxygen exchange, as well as other organ
systems such as the brain which would become clinically apparent with neurological deterioration of a patient
from cerebrovascular compromise. Leukapheresis aids in reducing WBC numbers and reducing this risk.