Prostate cancer vaccines: moving therapeutic vaccination forward in the post-Provenge era.

The focus of extensive research in the area of prostate cancer vaccines has led to the approval of the first therapeutic vaccine by the US FDA, sipuleucel-T. As our understanding of immunotherapy has increased, novel approaches have been investigated that have shown considerable promise. As the field has continued to evolve, questions have arisen regarding the potential role of immunotherapy: which populations of patients are most likely to benefit from immunotherapy and how and when should these therapies be administered? In addition, what are the best tools that can be used as surrogates to monitor immune responses to cancer vaccines that truly can give meaningful insight toward improving clinical outcomes? Finally, how can combination approaches be applied to prostate cancer vaccines in terms of both standard of care and experimental therapies? This review will address many of these important concepts with regard to prostate cancer vaccine therapy.

Prostate cancer, except for non-melanoma skin cancer, is the most common cancer in American men. In the USA, an estimated 240,890 cases will be diagnosed in 2011 and 33,720 deaths will occur [1]. The identifi cation of tumor-associated and tumor-specifi c antigens in the 1990s launched an unprecedented enthusiasm to harness the immune system and develop targeted vaccines for the treatment of prostate and other cancers. Most prostate cancer cells and epithelial cells lining the acini and ducts of the prostate gland express prostate-specifi c antigen ( PSA) [2]. In addition to PSA, other unique proteins that can be considered tumor-associated antigens ( TAAs) include prostatic acid phosphatase ( PAP) and prostate-specifi c membrane antigen, both of which are common targets for prostate cancer immunoth erapy. Although many early clinical studies investigating a broad spectrum of cancer vaccines were often disappointing [3,4], a better understanding of basic immunologic principles has led to a variety of techniques for enhancing tumor-specifi c immunity and their potential subsequent translation into improved clinical outcomes. This is exemplifi ed by the US FDA approval of sipuleucel-T ( Provenge™) on April 29, 2010. The vaccine was the fi rst licensed therapeutic cellular immunotherapy and a major milestone in the fi eld of cancer immunotherapy. Importantly, approval of sipuleucel-T was based on a statistically persuasive and clinically meaningful 4.1-month improvement in median overall survival ( OS) in the IMPACT Phase III trial [5]. The improvement in OS associated with sipuleucel-T has been reported to correlate with CD54 upregulation, a measure of the product's potency [6], and the development of antibody titers exceeding 400 at any time against the immunizing antigen P A2024 (a fusion protein) or PAP [5]. Interestingly, although strong T -cell proliferative response to P A2024 and P AP was also observed in patients receiving s ipuleucel-T, no difference or association in survival was documented between patients w ho exhibited T -cell response to either antigen and those w ho did not.
Despite this singular achievement, i t is important to realize that the landscape for Perspective p rostate c ancer therapy has changed considerably since the initial P hase I/I I studies of s ipuleucel-T were conducted when there were few effective F DA-approved treatments for men with m etastatic c astr ate-resistant p rostate c ancer (C RPC). In 2004, d ocetaxel with p rednisone was a pproved based on the results of two large randomized Phase I II studies that showed a 2.4 -month i mprovement in median O S in men with m etastatic C RPC treated with this regimen [7,8]. Although it was not shown to alter the time to disease progression in the majority of patients treated, the approval of s ipuleucel-T has resulted in an additional therapeutic option to further extend O S. However, i t is important to acknowledge that the population for whom this therapy i s indicated is limited to men with m etastatic C RPC w ho are minimally symptomatic with no visceral metastases or asymptomatic b efore receiving d ocetaxel. Furthermore, there are now a number of recently a pproved new drugs for p rostate c ancer that have rapidly altered the c linical setting and potential evaluation of current and f uture therapeutic p rostate c ancer vaccines under development: c abazitaxel (J evtana Injection™; a pproved by the F DA on June 17, 2010), a s econdgeneration t axane, and a bir aterone acetate (Z ytiga™; a pproved by the F DA o n April 28, 2011), a selective C YP-17 inhibitor -both of which are ap proved for use in patients with disease progression after tr eatment with do cetaxel. Unlike si puleucel-T, these compounds have demonstra ted objective cl inical responses in addition to increasing OS .
Consequently, as is the case with every advance in therapeutic options for a specifi c disease, the challenges in product cl inical development to successfully meet the threshold for regulatory approval will in crease in frequency and complexity for pr ostate ca ncer vaccines. Ea rly studies will face increased pressure to genera te solid data that permit mo deling of cl inical end points, estimation of the magnitude and timing of the va ccine effect size and identifi cation of populations to enrich for that will allow generation of an effi cient and informed Phase II I study design. The delayed onset in effectiveness of many vaccines and im munebased therapies further in tensifi es the need to identify and validate biologic or mo lecular surrogates that are reasonably likely to predict cl inical benefi t and outcomes. These issues are further complicated by a cl inical tr eatment background in which licensed agents are now available that result in objective responses based on tumor assessments in the context of an ever-increasing consensus that OS may be the most appropriate primary end point for immunotherapy cl inical trials.

Learning from past failures: identifying populations that should be targeted for therapeutic va ccination
Both cl inical and pr eclinical data suggest that vaccines are more effective in patients with lower tumor burden and less aggressive disease [9,10]. There are several possible reasons for this: activation of an im mune response is more robust than the ra te of tumor growth; in a large tumor, T-cell infi ltration and pe netrance may be insuffi cient for eradication; and im munosuppression in patients heavily pre treated with cy totoxic chemotherapy and radiation diminishes marrow reserves and the likelihood of adequate immune stimulation. In addition, in many cancers, cl inical and me tastatic disease progression has been associated with the induction of multiple im mune-inhibitory pathways that infl uence tu mor-specifi c responses. These in clude the induction of T-cell an ergy as a result of de ndritic cell (DC ) dysfunction and MH C class I do wnregulation, release of immunosuppressive cytokines, T-cell exhaustion and tu mor-specifi c immune suppression by T regulatory cells [11].
On the basis of these considerations, va ccine therapies may lead to better outcomes in adjuvant settings, while tumor burden is still relatively low, rather than in the setting of advanced me tastatic disease, an observation underscored by the recent and prominent failure of several ca ncer vaccines. Tw o Phase II I studies (VI TAL-1 and VI TAL-2) of GV A X ® (Ce ll Ge nesys, In c.), an al logeneic tumor cell va ccine consisting of two pr ostate ca ncer cell lines (LN CaP and PC -3) engineered to express human gr anulocyte macrophage-col ony-stimulating factor (GM-CSF), failed to document a survival benefi t of GVA X or GVA X/docetaxel over standard doc etaxel/pre dnisone in patients with symptomatic met astatic pro state can cer. VIT AL-2 was prematurely terminated because of safety concerns associated with increased deaths (n = 67) in the doc etaxel-GVAX arm compared with the doce taxel-predn isone control arm (n = 47) [12]. Clinical development of GVAX was subsequently abandoned in Octob er 2008 based on the results of a futility analysis of VITAL -1 that indicated the trial had less than a 30% chance of meeting its primary end point of impro vement in OS. S imilar to recent experiences with other immun e-based therapies, the fi nal Kapla n-Meier curves for the two treatm ent arms in VITAL-1 suggested a late favorable effect of GVAX i mmunotherapy on patient survival compared with chemotherapy, with the curve for GVAX p atien ts crossing above the chemotherapy curve at approximately the same time median survival was reached in both t reatm ent arms (21 months) [201]. Importantly, the clinic al design and primary OS end points in these two large Phase III st udies were supported by median survival results from two independent, multicenter Phase II tri als in approximately 115 patients, underscoring the fact that even with solid Phase II dat a, results of Phase III im munotherapy trials can be unpredictable. Indeed, the list of randomized clinic al trials of cancer vaccines with negative results reinforces the treacherous road of cancer immunotherapy clinic al development and the accompanying persistent uncertainty surrounding the ultimate goal of regulatory approval [13]. Although many therapeutic vaccines targeted malignant melanoma (Melaci ne ® , Canvax in, Oncoph age and GM 2-KL H21) and pancreatic (PANVAC™ -VF), renal (Oncoph age and T roVax ® ) and other cancers (MyVax ® , FavId™, GVAX and Therat ope ® ), the failures of randomized clinic al trials have been equally distributed among tumor types and among trials in patients with late (i.e., metast atic) and early disease (adjuvant setting).
An important question that arises then is, how can we predict who mi ght benefi t from a cancer vaccin e? Historically, Gleaso n score had been used as a key prognostic indicator to predict clinic al outcomes for primary prosta te cancer . However, the limitations of this approach led to the development and subsequent Perspective signifi cant use of the Halabi nomogram [14] as a predictive indicator of OS of patients with metast atic CRPC. The Halabi nomogram is based on seven predictors of OS: pr esence of visceral disease, Gleaso n sum, performance status, PSA, l actate dehydr ogenase, alkaline phosph atase and hemoglobin. Derived from the results of clinic al outcomes in six separate C ancer and Leukem ia Group B trials of 1101 patients with metast atic CRPC, this pretrea tment prognostic model predicts OS bas ed on the outcomes of patients receiving chemotherapy or hormon al-based treatm ent for metast atic CRPC a nd does not in clud e untreated patients. This model was used to compare predicted survival with actual observed OS in the Phase II stu dies of GVAX, which together served as the preliminary signal that GVAX m ight be associated with improv ement in survival outcomes [15]. More recently, an additional contemporary prognostic nomogr am has been developed that incorpor ates a more comprehensive characterization of baseline parameters to predict clinic al outcomes in men with metast atic CRPC [ 16]. This nomogr am was based on the outcomes associated with the TAX327 clinic al trial that randomized 1006 men with metastatic CRPC to receive docetaxel or mitoxantrone either weekly or every 3 weeks, each given with predni sone. This multivariate model identifi ed several new independent prognostic factors predictive of clinic al outcomes, including baseline PSA do ubling time (PSADT) and other PSA ki netics. Several additional progno stic markers were also identifi ed, including the presence of liver metastases, the number of metast atic sites (three or more), the presence of pain, the type of chemotherapy and the type of disease progression at baseline (measurable disease by imaging vs PSA bi ochemic al recurrence only). The nomogr am and multivariate model were found to have a bootstrap concordance index of 0.69, suggesting that further validation of this approach may be warranted by additional, prospective clinic al trials.
Although used extensively in previous trials of cytoto xic, radiot herapeutic, surgical and even immune -based interventions, decreasing serum PSA co ncentrations do not seem to be a reliable surrogate end point for prostate cancer -specifi c mortality [17]. Thus, PSADT, or the change in PSA ove r time, has also emerged as a useful predictive marker for assessing disease outcome in patients with prostate cancer, particularly in the setting of biochemica l recurrence -that is, stage D0 disease [18]. In general, an increased PSADT i s associated with a longer time to metastasis and death from prostate cancer, whereas a PSADT o f <3 months is associated with a poorer prognosis and an extremely high risk of disease progression in a variety of treatme nt settings (after defi nitive radioth erapy [RT] or radical prostatectomy, biochemica l recurrence; and androge n deprivation therapy [ADT] af ter RT or r adical prostatectomy). Although PSADT m ay ultimately prove to be a useful predictor of disease outcome in all stages of prostate cancer, there remains a general lack of consensus on its universal use or the exact method of calculation [18]. Despite these limitations, PSADT u sed as a stratifi cation factor in clinica l trials of cancer vaccines may help facilitate detection of a preliminary signal of an agent's activity in early-ph ase studies. Hence, it is highly likely that future t rials of cancer i mmune-b ased therapies will fi nd it increasingly useful to include prospective, planned cohort analyses accounting for Halabi n omogram and PSADT st ratifi cation.

Prostate cancer v accine platforms: the merits & challenges of different therapeutic vaccine approaches
There are a wide variety of platforms and immunoge ns that have been extensively studied in therapeutic vaccine platforms: whole tumor cell vaccines; DC-based platforms; peptide and fusion proteins codelive red with adjuvant s; and viral vectors that serve as delivery vehicles for TAAs. In addition to classic peptide and protein immunoge ns, tumor lysates, tumor mRNA an d DNA have all been used in prostate cancer v accines. PSA, PAP and prostate -specifi c membrane antigen have been the prostate antigens most commonly targeted, but tumor antigens found in a broad spectrum of tumors, including carcinoe mbryonic antigen, LAGE-1, MUC-1/2, NY-ESO-1 a nd telomerase, have also been used. Indeed, a DNA vacc ine targeting PAP has been shown to induce PAP-spec ifi c IFN-γ CD 8 + T-c ell r esponses and CD4 + and CD8 + T-c ell p roliferative responses associated with an incre ase in postvaccinat ion PSADT in men with stage D0 prost ate cancer [ 19].

Autologo us whole tumor cell & tumor lysate v accines
Autologo us whole tumor cell or tumor lysate v accines are derived from the patient's tumor cells in a time-cons uming process that requires a signifi cant amount of tumor tissue. The advantage of autologou s whole tumor cell/lysate va ccines is that they target the patient's TAAs and p resent a broad selection of antigens to the immune system, thereby precluding the need for antigen preselecti on. One disadvantage, however, is that antigens could be diluted by other cellular components. Other disadvantages include th e lack of st andardized measures for vaccine po tency and the diffi culty in evaluating antigen-sp ecifi c immune responses postvaccinatio n because, for the most part, the tumor antigens from these vaccines remain unknown. Despite these challenges, there remains signifi cant interest in pursuing this type of approach because impressive survival outcomes have been described for a tumor lysate vac cine us ing autologous DCs (DCVax -Brai n™; Northwest Biotherape utics, Inc.) for glioblasto ma multiforme, a uniformly lethal and highly aggressive disease [20,21].

Allogeneic tumor cell vaccines
Allogeneic whole tumor cell vaccines that can also be delivered as lysates ar e derived fr om various tumor cell lines and are usually easier to generate. Whol e tumor cells are rendered replicatio n defective by radiation and are frequently combined with nonspecifi c immunostim ulants [22][23][24]. This approach is exemplifi ed by the GVAX platfo rm discussed e arlier in this review that combines two different prostate ca ncer cell lines modifi ed to secrete GM-CSF. The rationale for the putative effi cacy of allogeneic tumor cell vaccines is that tumor-speci fi c antigens contained in the vaccine are often shared in common by different patients. These shared antigens are potentially immunogenic and can enhance the host's ability to generate an effe ctive immune respo nse. Importantly, because they are allogeneic, these vaccines Perspective are able to break tolerance to self-tumor a ntigens and have the advantage of being applicable to a broader number of patients regardless of the availability of bulky, autologous t umor. However, a major disadvantage of whole cell t umor vaccines is that tumor cells, even when foreign, are generally not immunogenic. Thus, these tumor vaccines are frequently coformulated with or engineered to express immunostimul atory proteins such as GM-CSF and h eat shock p rotein to improve immunogenicity [25].

DC vaccines
DCs serve as a bridge between the innate and adaptive immune system and in turn play a criti cal role in the activation of naïve CD4 and CD8 T cells. Hen ce, researchers in cancer immun otherapy have focused a tremendous amount of investigative effort to delineate and understand DC biology, activation, maturation and antigen presentation. Because of the pervasive dysfunction of DCs in vivo as a consequence of the immune dysregulatio n associated with high tumor burdens [26], they have been the platform of choice for d elivery of many tumor cell lines (LNCaP, PC-3) , pepti des, proteins, lysates, mRN As and vi ral vectors expressing TAAs [27,28] . Sipuleucel-T is a prototype of this platform approach: even though it is labele d as an autologous c ellular immunotherapy, its generation uses similar maturation agents and pathways common to DC vaccines [29]. In addition to its documented activity in men with minimally symptomatic CRPC, sipule ucel-T has also been investigated in a randomized trial in men with androgen-dependent prostate can cer [30]. Men with a persistently increased serum PSA after ra dical pros tatectomy received ADT for 3-4 months and were then assessed for time to biochemical fail ure, defi ned as a serum PSA 3.0 ng/m l. Althoug h n o signifi cant difference was seen in time to biochemical fa ilure, there was an increase in PS ADT observed in patients randomized to sipuleucel-T a fter testosterone recovery, indicating that further studies may be warranted in this population to further assess other clinically relevant outcomes. Despite their preliminary activity, production of personalized autologous DC-based vacci nes requires ex vivo stimul ation of peripheral blo od mononuclear cells with cytokines, which can be costly and time consuming . In addition, the number of cells and cellular composition and functional potency may vary signifi cantly because each product is patient-specifi c, making it much harder to characterize and standardize product charac teristics that may correlate with clinically meaningful outcomes.

Peptide-& protein-based vaccines
With the advent and discovery of t umor antigens, pept ideand protein-based immunogen plat forms were among the fi rst approaches to be pursued in cancer immunot herapy. What became evident fairly quickly was that without codelivery of adjuvants to e nhance presentation of vaccine antige ns to the immune system, both peptides and proteins w ere often poorly immunogenic. DNA vaccines, heralded for their simplicity and low production costs, were the least immunogenic of all, requiring large amounts of pro duct, administration by electroporatio n and co-inoculation with plasmids expressing cytokines and CpG motifs to optimize immune responses [31]. Adjuvants are pharmacologic or immunologic agents that modify the effects of other agents while having few, if any, independent direct effects when given alone. They are often included in vaccines to enhance the magnitude, breadth and longevity of a recipient's response to a supplied antigen and to direct the qual ity of the immune response induced by vaccination. Th e goal is to use compounds that achieve a poten t adjuvant effect with minimal reactogenicity, toxicity or lasting effects on the immune system on their own. BCG, keyhole li mpet hemocyanin, Mon tanide ® ISA5 1 and poly-IC-LC are among the most common adjuvants curre ntly used in cancer immunoth erapy platforms in addition to cytokines (IFN-γ, IL-12, I L-15 a nd GM-C SF), immuno stimulato ry complexes (ISOCOMS), costi mulatory m olecules (e.g., ICAM-1, lymphoc yte func tionassoci ated antigen-3 and B7.1 presently used in the PSA-TRICOM vacc ine) and pr oprietary adjuvants (Alhy drogel ® , AP 1903 and AS1 5) that are used to ensure immune recognition of the desired immunogen [32]. Adjuvants can b e classifi ed into three groups: active immunostimulant s that increase immune response to the antigen; carriers that are immunogenic proteins that provide T-cell help; an d vehicle adjuvants that are oil emulsions or liposomes, whic h serve as a matrix for antigens in addition to sti mulating the immune response [33]. A summary of commonly used cancer vaccine immune adjuvants is sh own in TABLE 1.
In keeping with the theme of maximizing immunogenicity of peptide and protein tumor antigens through adjuvant or DC platforms, w e are conducting a prospective, randomized Phase I pilot s tudy of T-cell receptor alternative reading frame protein (TARP) peptides delivered in combination with Montanide ® ISA5 1 VG adjuva nt plu s sargramostim (G M-CSF ) or as a custom-made, pa tient-specifi c autologous pept ide-pulsed DC vaccine in m en with androgen-sensitive stage D0 prosta te disease [34]. These men with PSA biochemical recurr ence and no evidence of metastatic dise ase constitute a population with normal (or near normal) immune function . TARP is express ed by both normal and malignant prostate tissue and is overexpressed i n 95% of prostate cancer specimen s [35-37], making it a good target antigen for therapeutic vaccination. The current vaccine platform consists of two HLA-A*0201 TARP peptide epi topes documen ted in preclinical labo ratory studies in mice and human cell lines to generate cytolytic C D8 T-cell respo nses : TARP27-35 and ep itope-enhanced TARP29-37-9V [38 ]. A total of 1.1 mg of each peptide is delivered per vaccine given ev ery 3 weeks for an ini tial course of fi ve vaccinations with subsequent options for a sixth dose of vaccine at 36 we eks based on T ARP-specifi c im mune response or PSADT criteria; all patients subsequently undergo booster vaccination at w eeks 48 and 96. Primary study end points include safety a nd immunogenicity, with secondary evaluation of the impact of TARP vaccination on P SADT. To date, TARP vaccination has been safe and well-tolerated, with adverse events ( AEs) limited to local injection-site reactions. A preliminary interim analysis performed on the fi rst 26 patients documented that TARP vaccination was associated with a statistically signifi cant slowing in the rate of PSA incr ease as measured by PSADT and slope log (PSA) at 24 and 4 8 weeks in the maj ority of patients, with no Expert Rev. Vaccines 11(3), (2012) Perspective differences between the two study arms detected. Analysis of immunogenicity to TARP WT27-35 and EE29 -37-9V vacci ne epito pes and cro ss-reactivity to native WT29-37 as asses sed by IFN-γ enzyme-lin ked im munosorbent spot (ELISPOT) and TARP tetramer as says is ongoing.

Viral vector vaccines
Poxviruses such as the vaccinia virus h ave been extensively studied and have been shown to induce a strong local response at the site of inoculation. In patients previously inoculated with vaccinia, the qu estion arose as to whether these patients could mount suffi cient immune responses compared with those who were vaccini a naïve. This was resolved by administering higher doses of recombinant vaccinia [39]. H owever, the effi cacy of vaccinia vectors has been limited by the development of neutralizing antibodies to vaccinia after o ne or two injections, resulting in a decline in antigenspecifi c immune responses caused by the vector. It was observed that this phenomenon could be overcome by using a replicationdefe ctive avipox virus suc h as fowlpox as a boo st after a priming vaccination with vaccinia [40][41][42] . Replication-defec tive vectors have the advantage of superior safety because they are unable to replicate in the human body but are able to infect human cells and express their encoded transgenes for 2-3 weeks before cell death. PSA-TRICOM is a po xvir us-based vac cine expressing PSA and three cost imulatory mole cules (B7.1, ICAM-1 and lympho cyte functio n-associate d antigen-3). A small Phase II study in patien ts with metastatic CRPC ev aluated 32 patients treated with PSA-TRICOM on a pr ime-and-boost sc hedule. The median OS for all patient s was 26.6 months compared with a predicted survival of <18 months [43]. This preliminary observation was subsequently confi rmed in a larger, blinded, randomized Phase II study exa mining PROSTVAC-VF (vacci nia-PSA-TRICO M and fow lpox -PSA-TRICOM ) in 125 pat ients with minimally symptomatic metastatic CRPC [4 4]. Althoug h there was no difference seen in the primary end point of progression-free s urvival (PFS) between patie nts who had received PROSTVAC-VF vs co ntrol vector, at 3 years of post-study follow-up, the med ian OS was 25.1 months vs 16.6 months, respectively, an improved survival difference of 8.5 months.
Adenoviruses are also among the select group of viral vectors that have been extensively studied in the delivery of cancer vaccine ant igens, including MART-1 and gp100 i n the treat ment of patie nts with metastatic melanom a [45,46], HPV E7 CD-40 ligan d fusio n prot ein in preclinical animal mo de ls [47], and f ulllength p53 delivered by transduction of autologous DCs wit h an adenov iral vector in patients with extensive small-cell lung cancer receiving s ubsequent chemotherapy [48]. Of interest, the coxsackie and aden ovirus receptor has been shown to mediate adenoviral entry into tumor cells. Unlike cells derived from other malignancies, loss of coxsackie and aden ovirus receptor expression in human and murine prostate ca ncers s eems to be comparatively infrequent and is not associated with the loss of MHC class I expres sion [49]. These observations suggest that cancer vaccines ba sed on modifying whole prostate cancer ce lls using recombinant adenoviral vectors could be feasible without detrimental effects to the target cell or the viral vector itself. Importantly, a Phase I trial of a single dose of an adenoviral/PSA (Ad/PSA) vacci ne in 32 men with measurable metastatic CRPC wa s associate d with an increase in PSA DT in 48%, wher eas 55% survived longer than predicted by the Halabi nomogram [5 0]. Immunologic assessment revealed that anti-PSA antibodie s wer e produced by 34% of patients and anti-PSA T-cell re spons es w ere detected in 68%. Further development is proceeding with this adenoviral vector approach in two parallel Phase II trials in vestig ating three doses of Ad/PSA. Protocol 1 (N CT00583752) ex amines vaccination in men with recurrent prostate cancer after defi n itive treatment who unde rgo randomization to receive 108 pfu of Ad/PSA vacc ine at day s 0, 30 and 60 or delay vaccination until 14 days after the start of ADT, whereas in pr otocol 2 (NCT00583024), men with CRPC receive the v accine monotherap y at day s 0, 30 Perspective and 60 [51]. Preliminary reported results demonstrate that this Ad5 /PSA multidose regi men is associated with stable or decreased PSA and PAP levels in a ma jority of patients. These fi ndings, along with the documented early preliminary activity of PSA-TRICOM, are re ason s for cautious enthusiasm regarding the potential for viral vector vaccines to further advance the fi eld of prostate cancer im munothera py.
The challenges that remain: documenting evidence of cancer vaccine cli nical a ctivity early in developme nt

Challenge 1: clinical trial end points for cancer vaccines & immune-based thera pies
Regular or accelerated regulatory a pproval of oncology drug products and biologics is based on the risk-benefi t evalu ation of end points that demonstrate that the agent provides clinical benefi t as evidenced by a longer or better life or a favorable effect on an established surrogate for a longer or better life [52].
These clinical end points , alone or in combina tion, include OS, disease -free su rviv al (DFS), PFS or time t o prog ression (TTP), overall respo nse ra te and, rarely, other novel end points. Data for regulatory approval are typically generated from randomiz ed studies that examine the activity of the investigational agent vs a comparator control that may be a placebo, standard chemotherapy, or best supportive care. However, regulatory approval for clinical indication s has also been given based on data from single-ar m studies with no comparator group. In a recent review of regulatory actions on 58 indications for oncology and hematology drug products between July 2005 and Decem ber 2007, 37 o f 53 (70%) approved indication s were based on data from randomized studies, whereas the remaining 16 (30%) approved indication s were based on single-arm studies [52]. Importantly, 44 of the 53 approved indication s were based on the results from a single, pivo tal study, highlighting the critical importance of trial design, patient population and selection of primary effi cacy end points to achieve agent approval. Although FDA regulations req uire data from more than one independent study to establish the effi cacy of a new agent, as previously noted, few marketing applications for oncology products are submitted based on more than one study, reinforcing the need for signifi cant sci entifi c and statistical vetting of trial designs under consideration before the huge risk and fi nancial investment in large randomized Phase III studies i s unde rtaken. Ultimately, approval is based on an agent's demonstration of a clinically meaningful superior treatment effect bas ed on the primary end point of a pivotal study and demonstration of a corroborative treatment effect in secondary end points. In addition, the product safety profi le must be acceptable in view of the magnitude of the clinical effi cacy ob served and the disease or condition being treated. OS is the platinum s tandard for demonstrating direct clinical bene fi t bec ause it is an unequivocal outcome measure evaluated on a continuous time scale that allows for precise accuracy of the time of the event. The surrogate end points of DFS, PFS and TTP are not measured continuously and, consequently, the exact day of recurrence or progression cannot be accurately captured in s tudies that use these outcomes.
In addition, assessment of progression in PFS and TTP depends on the f requency and completeness of tumor assessments and selection of target lesions used in evaluation, that is, Response Evaluation Criteria in Solid Tumors (RECIST), in addition to other physical examination, and laboratory and radiologic documentation. Overall response rat e, the only end point that can be reliably assessed in single-arm trials, m ust be durable and of suffi cient clinically meaningful magnitude, although 'clinically meaningful' is still subject to interpretation. Importantly, recent approval of both chemotherapeuti c-and immune-based agents has revealed that improvements in OS m ay not be associ ate d with changes in PFS or TTP, whereas statist ically signifi cant changes in DFS or PFS may or ma y not b e associate d with improvements in OS. Sipuleucel-T is an e xample of the former: improvement in OS in the Phase III IMPACT tri al was not associated with an improvement in PFS [ 5]. In contrast , panitumumab for refr actory colorectal cancer shows minimal but statistically signifi cant changes in PFS with no differen ce observed in OS [53].
On the basi s of published studies of cancer immune-based therapi es recently approved or still un der investigation, it has been suggested that OS may be emerging a s the most appropriate end point in immunotherapy clinical trials [54] for several reasons: objective responses to immune-based therapi es generally take longer relative to chemotherapy; patients receiving immune-based therapi es sometimes demonstrate 'disease progre ssion' by standard RECIST criteria befo re subsequent regression and clinical improvement ; and, as previously noted, improvements in OS a re often not acc ompanied by improvement in PFS o r TTP. This paradoxica l observation to the well-established and widely accepted treatment paradigm t hat agents should initially decrease tumor volume (objective responses) and impact disease progression (PFS, TTP or DFS) sug gests that v accines and immune-based therapi es have distinctly different functional dynamics in vivo. Although OS is an unambiguous outcome that provides direct evidence of c linical benefi t, la rge studies are generally necessary to show modest improvements in OS a nd longer follow -up is necessa ry for OS than for other en d points such as PFS. In addition, th ere may be a delayed effect in the separation of survival curves after treatment randomizat ion. As summarized in TABLE 2, separation in survival curves was seen at ∼4 months with ipilimumab [55], at ∼8-12 months in the trials of sipuleucel-T [3,5], an d not until 14 months in a randomized study of PROSTVAC [44], wherea s differences in OS were seen as early as 3 months wit h abiraterone acetate [ 56] . This means that larger changes in the OS end point after se paration are needed to compensate for no effect on the end point before separation. An additional conundrum with the use of OS as an end point in immunotherapy trials is that separation in survival curves is most likely to occur and show the maximum difference in patients who have the greatest risk of disease progression. However, the patients at greatest risk are in turn likely to have higher tumor burdens and greater immune dysfunction and are less likely to respon d to immunotherapy.
The greatest challenge to the continued development of therapeutic cancer vaccines and i mmune-based therapie s is documenting evidence of preliminary activity early in the clinical  Yes Yes Yes [52,56] Difference in OS at ~3 months † Separation in Kaplan-Meier curve estimates of the probability of OS in groups based on visual review of curves in references cited. ‡ Initial analysis when 552 events had occurred. m CRPC: M etastatic c astrate-resistant prostate c ancer; OS: Overall survival; q3: Every three. development of a product. Although improvement in OS has been establish ed as the sentinel standard for regulatory evaluation with the approval of sipuleucel-T, OS is n ot a clinicall y feasible primary end point for fi rst-in-human Phase I/II trials. Co mpoundin g this challenge is the observation that end points based on objective tumor assessments are unlikely to reliably serve as harbingers of early clinical activi ty in Phase I or II studies because, to date, neither sipuleucel-T nor PROS TVAC studies demo nstrated any impact on T TP or DFS. This is a critic al issue for the entire fi eld of cancer immunotherapy because early positive outcom es are needed to generate suffi cient evide nce of activity to warrant moving forward with the further expense of larger clinical trials of pr olonged duration to capture OS outcomes. Finally, there is a massive lack of validated dis ease surrogates or immunologic correlates that are predictive of clinical outcomes. Th ese challenges underscore the need for truly novel approaches to assessing the clinical activity of therapeutic cancer vaccines.
The fi rst approach is to ensure that patients, who may achieve clini cal benefi t from cancer vaccines in th e form of improved survival, are not prematurely removed from studies for failure to objectively respond or even what may seem to be disease progression on conventional radiologic imaging studies. Currently, no standardized system has been implemented to measure the multiple patterns of response that may be associated with immunotherapy. Fundamentally, it is crit ical to remain cognizant that all assessments of th e impact of therapeutic vaccination on clinic al outcomes and parameters are by their very nature indirect. As shown in FIGURE 1, therapeutic vaccination may or ma y not induce an imm une response, whic h in turn may or may not be ass ociated with an anti-tumor response th at may or may not be ass ociated with measurable tumor regression. Importantly, it is conceivable that induced immune responses may not necessarily b e measurable by the current repertoire of immune assay and molecular diagnostic tools at our disposal. In addition, when an anti-tumor response is fi rst generated, it may potenti ally result in an increase in tumor volume as a result of antigen-specifi c lymp hocyte infi ltrati on of tumors and local infl ammatory changes associated with tumor killing, which in turn causes monitored lesions to Perspective appear larger or caus es the appearance of new lesions related t o previously undetectable micrometastatic disea se [57]. In addition, existing lesions may be slow to regress after immunotherapy as supported by multiple reports of patients taking immune-based therapie s for progressive disease, only to have lesions subsequently regress or even resolve completely [58]. Immune response-relat ed criteria are a novel way of systematically addressing unique pattern s of response that may refl ect the unusual biology of therapeutic vaccination. These cri teria allow the capture of additional response patterns observed with immune therapy beyond those traditionally described by the RECIST [59,60] or WHO criteria [61] that may be associated with favorable OS, including conventi onal responses, delayed responses and durable stable disease. Standardized implementation of immune response-relate d criteria in prospective clinical trials of canc er vaccines and imm une-based therapies is necessary to validate their positive predictive value and clinical correlation wi th OS end points.
Clearly, novel paradigms are needed to assess early and establish defi nitive vaccine activity in the treatment of solid tumo rs. Examination of tumor growth rate and regression rat e constants has been proposed as a possible indicator of therapeutic effi cacy [62]. Using a two-phase mathematical equa tion generated from PSA measurem ents obtain ed before and during therapy in fi ve Phase II studies conduc ted at the National Cancer Institu te in pat ients with metastatic CRPC yielded data on concomitant PSA growth and regressi on rate constants. Growth rate constants correlat ed with survival, except in patients receiving PROSTVAC therapy, where prolonged survival was documented, presumably as a consequence of vaccine-induced immunit y that d eveloped subsequent to vaccine administration. Specifi cally, the PROSTVAC vaccine seemed to have provided marked clinical benefi t that w as not apparent during vaccination but was con sistent with the subsequent development of a benefi cial immune response. Simila r to immune responserela ted criteria, growth rate constants will need to be prospectively evaluated as a mathematical surrogate for survival and should be examined in clinical trials of multi ple immune-based therapy pla tforms in a variety of solid tumors to ensure their robust validation as a potential important new effi cacy end point.
There is also a need to use innovative clinical trial designs t o evaluate new immunotherapy products. Clinical studies with ad aptive sequential designs can provide decision rules for terminating a trial if a predefi ned magnitude of effi cacy or futility is demonstrated. However, given th e general trend of delayed responses to immune-based therapies, we must be careful to avoid designing studies that predispose treatment pla tforms to b e prematurely classifi ed as failures because of a lack of early signal act ivity an d delayed separation of Kaplan-Meier survival cu rves in randomized immunotherapy trials. To that end, altered statistical models describing hazard ratios as a function of time and recognition of differences before and after separation curves may allow improved planning of Phase III trials [63]. Ho wever, even if these approaches are adopted, the hurdle of documenting suffi cient preliminary activity of an immune-based therapy in fi rst-in-human and early Phase I/II studies will r emain. A lthough generally well-tolerated compared w ith chemotherapy and, hence, attractive to patients, vaccine therapies may als o be associated with the induction of an atypical or "inconvenient immune respo nse" that potentiall y inhibits or interferes wi th pre-existing host anti-tumo r immunity. Speci fi cally, poorly immunogenic or irrelevant vaccine antigens that ind uce overabundant quantities o f nonspecifi c or functional ly ineffective memory cells could potentially displace or suppress benefi cial pre-existing memory respon ses and, in turn, accelerate cancer progression [ 13]. It is also important to a cknowledge that vaccine immunogens may in duce tol erogenic rather than c ytolytic anti-tumor respo nses. Cons equently, the development of reliable biomarkers to identify appropriate patient populations and vaccine antigens is pivot al to the prevention and minimization of potential vaccine-specifi c AEs.

Cha llenge 2: lack of correlation of im munologic end points with clinical outcomes
Unlike prophylactic vaccination against infec tious diseases, in which a specifi c immune response directed against the pathogen of interest can be detected that correlates with therapeutic clinical activity (usuall y a pathogen-specifi c antibod y response), identifi cation of immunologic end points that consistently correlate wi th clinical outcomes in clin ical trials of cancer vaccines has been sorely absent. This in part may be attribu ted to a lack of standardization i n immune monitoring or the use of a wide variety of methodologic assays that are specifi c to a vaccine immunogen, platfo rm or un ique immunotherapy approach. Indeed, the literature is replete with examples of clinical studies of therapeutic vaccines in which vaccine-induced immune re sponses are detected by in vitro assays in a sign ifi cant number of patients who ultimately have limited or no objective clinical responses to vaccination [64][65][66]. Furt hermore, the kinetics of immune responses associated with vaccination are by nature indirect and differ signifi cantly from chemotherapies, which act directly to exert antitumor effects. Lack of immunologic correlation may also result from weak immunogens that fail to in duce robust anti-tumor responses or result in the induction of tumor-specifi c responses t hat are not detected by the current repertoire of immunologic monitoring assays commonly used to characterize vaccine activity. These as says include ELISPOT assays for differe nt cytokines (typically IFN-γ), tetramer-based ass ays, int racellular cytokine sta ining by fl ow cytometry, lymphocyte prol iferation a ssays and multiplex assays for serum cytokine and chemokine profi les after v accination. Furthermore, the analysis of single or even multiple assay parameters ex vivo may still fail to suffi ciently refl ect the complex interactions that occur between the immune system and tumors in vivo.
Although historic al approaches to immune monitoring have typically focused on characterizing and measuring the positive induction of cellular or humoral immune responses, an increasing appreciation is emerging for the need to assess and overcome the negative regulatory responses that prevent the immune system from successfully attacking self-tumor antigens. In ad dition to suppressive cytokines such as TGF-β, IL-10 and IL-13 an d membra ne protein s PD-1 and cytotoxic T-lymph ocyte ant igen 4 (CT LA -4), a functionally he terogeneous group of cells has emerged that are now understood to play key roles in dampening the immune response against tu mors that includes T regulatory cells (Tregs), myeloid-derived su ppressor cells a nd tumor-associated macrophag es. There is an emerging opinion that the impact of therapeutic vaccination on both negati ve regulatory a nd positive stimulatory responses needs to be examined for a more comprehensive assessment of vaccine activity that may subsequently allow for better correlation with clinical outcomes [67].
Al though the fundamental need to overcome a nd break selftolerance to tumor an tigens is a universally acknowledged and accepted scientifi c paradigm, characterization of innate immune responses after administration of cancer vaccines is uncommo n. Relative to adaptive immune responses, there is even less consensus within the immunology community about how innate immunity should be routinely characterized and measured. In addition, until recent years, outside of allogeneic tumor vaccines, only a limited number of vaccine platforms have inc luded components or reagents designed to trigger innate immunity at the initial vaccine presentation. Thus , for most of the fi rst 30 years in the fi eld of cancer immunotherapy, effo rts have been focused on producing a clinically effi cacious, functionally specifi c adaptive anti-tumor response in the vacuum of any triggers of innate immunity or interruption of self-tolerance. This may i n part account for the negative outcomes that have been associated with the majority of immunotherapy clinical trials to date. I mportantly, immunologic parameters may not correlate with cli nical outcomes because tumor cells may also continue to evolve and escape in response to naturally occurring selective immune pressures and those induced by vaccination. Despite the t remendous research investment in can cer immunotherapies and vaccin es, the issue of what immune responses to measure, how to measure them and when to measure them in relation to our therapeutic interventions remains the subject of intense and ongoing scientifi c deliberation.

Challenge 3: lack of assay standardizat ion in obtaining immunologic end points
A major focus of the efforts to improve end points for cancer immunotherapy trial s involves addressing the standardiza tion and validation of immunologic assay readouts. As an example, although IFN-γ ELISPOT is one of th e most common immune assays examined in cancer vaccine trials, no consens us has been reached on how to determine whether a positive immune response has been d etected based on ELISPOT assay raw data and criteria can vary widely on how a positive response is defi ned [68]. In fact, the conduct of tw o large-scale international ELISPOT profi ciency panels demonstrated a wide degree of var iability in the detection of CMV and CMV, EBV and Flu peptide p ool r eactivit y using a standard set of predefi ned donor peripheral blood mononucle ar cells, especially in the detection of low responders [69]. Detailed surveys of laboratory practices revealed differences in assay protocol choices and laboratory practices that had marked effects on overall assay outcomes and led to recommendations for the establishment of standard operating procedu res for ELISPOT testing that resul ted in improved performance outcome among sites with the second round of profi ciency testing. The critical importance in addressing these types of issues is highlighted in recommendations from a workshop representing academic, biopharmaceutical, regulatory and government research stakeholders [70] . These recommendations include following accurate , precise and reproduc ible immune assay protocols; the use of functional assays for prim ary immunology readouts in clinical trials; utilizati on of central laboratories for immune monitoring of large, multi-institutional trials ; and standardized testing of multiple phenotypic and functional potency assays specifi c to any cellular product. Furthermore, standardized reporting of assay methodology will permit the development of novel assays for immunologic testing and provide additional insights into immunity in Perspective patients with cancer and how to effectively direct and enhance it to e radicate tumors using immunotherapy.

Cancer immunotherapy plus: the tantalizing promise of combination therapy
Both preclinical animal model a nd human studies su ggest the potential for synergistic effects resulting from the combination of immunotherapy and conventional treatments for prostate cancer such as radiation, ADT and certain chemother apies.

Radiation plus vaccine
Independent of its direct, cytotoxic effects, radiati on therapy has also been documented to have indirect, immune-mediated antitumor effects. Indeed, it has been known that ionizing radiation can reduce tumor growth outside the fi eld of radiation, a phenomenon known as the abscopal effect described by Mole in 1953 [71] and obse rved in a wide variety of tumors, including lymphoma, papillary adenocarcinoma, malignant melanoma and hepatocellular carcinoma [72][73][74][75]. Studies in mice confi rm that the abscopal effect is, in part, immune-mediated, and that T cells are required to med iate distant tumor inhibition induced by radiation [76]. These radiation effects are presumably related to infl ammatory and maturation signals that are generated by dying cells [77][78][79][80], which can result in the activation of dendritic and tumorspecifi c T cells and m arkedly increas e the permeability of solid tumors to these effector populations [81,82] . Furthermore, radiation therapy can alter the tumor cell phenotype with respect to MHC class I expression and m olecules such as FAS (CD95, APO-1), thus potentiating anti-tumor immune effects [83 -86]. Synergism of radiation therapy in combinat ion with therapeutic vaccination has been observed in human clinical trials. This was obs erved in a randomized Phase II trial in which patie nts wi th prostate cancer were randomiz ed to rec eive radiation therapy with vaccin e vs radiation therapy alone. Increases in PSA-specifi c T c ells occurred in 13 of 17 patients in the combination arm. No increase was observed in the RT-only arm [87]. Furthermore , patients receiving the combination therapy were able to mount immune responses to additional tumor antigens found on the tumor that were not contained in the vaccine. Future studies shoul d be designed to try to determine how immunotherapy can be optimized to take advantage of these basic biologic observations in patients undergoing radiation therapy for high-ri sk localized prostate cancer with a focus on clinic al outcomes as a primar y end point. In addition, investigation of these therapies may also prove useful in the adjuvant setting for patients who have completed either loc al radiation or surgical therapy and whose immune systems are less likely to have dysfuncti onal impairment and a greater ability to optimally respond to vaccination.

ADT plus vaccine
Another ther apeutic a pproach that has been noted to potentiate immune responses is ADT [88,89]. Androgen ablatio n decreases CD4 + T-cell tolerance to a pr ostat e cancer-associated an tigen in an autochthonous mouse prostate cancer model. In this model, a doptively transferred CD4 + T cells responded to spec ifi c vaccination after androgen abl ation, but not in intact tumorbearing mice [ 83]. Other murine studies have demonstrat ed that thymic regeneration has been r eported in aging mice 4 weeks before castration, with an increase in MHC class II expre ssion in thy mocytes co mpared with that in young adults [90]. Immune effects of ADT have been described in sev eral human studies. In one report, T-cell infi ltration of CD4 + ly mphocytes was observed 1-3 weeks after ADT [91]. In anothe r study, pat ients receiving neoadjuvant ADT for prostate ca ncer demonst rated an increase in tumor-asso ciated an tibody respo nses [92]. Finally, a vaccine trial consisting of pat ients before prostatectomy receiving sipuleucel-T vaccine after 3 mo nths of ADT s howed a trend in time to P SA progression compared with p atients who did not receive vaccine [93 ].

Chemotherapy plus vaccine
There are mounting data challenging the dogma that chemotherapy agents are immunosuppressive and thus cannot be combined with vaccine therapy. An alternative perspective is that chemotherapy may induce anticancer immune responses as a co nsequence of immunogenic cell stress and death and increase tumor cell susceptibility to lysis and/or induce indirect im munostimulato ry effe cts [94]. Both in vitro and in vivo studi es ha ve demonstrat ed that certain chemotherapy agents can enhance immune responses through upregulation of MHC class I exp ression and by incr easing immune stimulatory c ytokines and decreasing regulatory T cells [95,96]. In the clinic, a randomized Phase II s tudy was performed to deter mine the immune effects of docetaxel combined with a poxvi ral-PSA vaccine compared w ith vaccine al one in patients with m etastati c CRPC. Both arms were able to mou nt increases in PSA-specifi c T-cell responses to the vaccine. In addition, patients w ere administered corticosteroids as premedication with docetaxel in this study. The pul se use of steroids did not seem to induce an immune suppressive eff ect [97]. However, chronic daily oral steroid administration that is common to many prostate cancer regimens needs t o be eval uated because the immunosuppressive effects of these agents on both cell-mediated and humoral i mmuni ty are well known and likely to interfere with the generation of effective vaccine-induced memory responses . It is important to note that patients who have previously received num erous chemotherapy agents may have impaired marrow reserves that may blunt their ability to mount a potent immune response.
In preclinical animal studies inves tig ating the therapeutic effect of an anti-HER2/neu antibody, triggeri ng an d activation of innate immunity initiated by antibody treatment was necessary [98]. Ho wever, administration of chemotherapy in this model diminished this antibody-i nitiated immunity and in turn resulted in earlier relapse or diminished resistance to tumor re-challenge. In contrast, admin istration of immunotherapy further augmented activation and resulted in increased tumor eradication and resistance to challenge related to an infl ux of both innate and adaptive immune cells into the tumor microenvironment. Clearly, the timing, dosing and scheduling of chemotherapy and immune-based therapies may signi fi cantly impact the activity of subsequent interventions. Hence, Expert Rev. Vaccines 11(3), (2012) Perspective careful consideration should be given to the design of clinical trials investigating th ese types of combination approaches.

Cytokines & immunomodulatory monoclonal anti bodies plus vaccine
Monoclonal antibodies su ch as ipilimumab targeting the human C TLA-4 monoclonal antibody have been tested in patients with prostate cancer. The primary fun ction of CTLA-4 is to provide a regulator y checkpoint by downregulating the immune respon se over time, thus preventing autoimmunity. The administration of anti-CTLA-4 antibody has been sh own t o sustain and potentiate immune responses [99][100][101][102]. This has been observed in a Phase I study of ipilimumab with the GVAX prostate cancer vaccine in p atien ts with m etastat ic prostate cancer. A cor relation be tween imm une-related AEs and immune res ponse was noted in this study [103]. Similarly, in the pivotal Phase III study that led to the appro val of ipilimumab by the FDA in March 2011 for its use in patient s with metastatic malignant melanoma, ip ilimumabrelated autoimmune phe nomena also seemed to correlate with treatment response [55]. The most common AEs associated with ipilimumab th erapy are related to its immunologic mechanism of acti on leading to T-cell activation and proliferati on, with subsequent immune-mediated AEs that most com monly involve th e colon (enterocolitis), liver (hepati tis), ski n (dermatitis, includin g toxic epidermal n ecrolysis), endocrine system (adrenal insufficiency, hypopituitarism and hypothyroidis m) and nervous system (neuropathy). These autoimmune AEs are often of suffi cient sever ity as to warrant their management with the initiation of high-dose corticosteroids (1-2 mg/kg/day of prednisone or equivalent) and disc ontinuation of ipilimumab. The observation that a utoimmune phenomena are associated with its benefi cial clinical effects implies that brea king tolerance to self-antigens likely plays a criti cal role in inducing anti-tumor effects that are associa ted with positive, objective clinical outcomes, for example, im proved survival.
Ipilimumab is also being studied i n prostate cancer in several clinica l setting s: as neoadjuvant therapy with leuprolid e acetate before prostate ctomy (NCT01194271) ; in combination with ADT for metastatic, castrate-sensi tive dis ease (NCT0137738 9); a nd in Phase III cli nical studies of pati ents w ith metastatic CRPC who are chemothera py naïve (N CT010 57810) and postdocetaxel ch emotherapy (NCT00 86161 4). Hence, there is a co mpelling scientifi c rationale to consider combining therapeutic approaches that have direct anti-tumor activity (chemoth erapy a nd radiation) with approaches that have indirect anti-tumor activity throug h their e ffects on the immune system (therapeutic vaccines, monoclonal antibodies and other immune-based therapies) (FIGURE 1) . Specifi cally, chemother apy or RT, in addition to reducing tumor burden, could potentially enhance both cellular and humoral immunity as a result of ti ssue necrosis and the release of multiple tumor antigens. In turn, immunotherapy could lead to further enhancement of tumor-specifi c responses through v accination. The additional use of monoclonal antibodies that block negative regulators of immunity such as anti-CTLA-4 and anti-PD1 could aug ment traffi cking of i nnate and adaptive immune cells into the tumor microenvironment. This multipronged, multiplatform approa ch will allow a more comprehensive and coordinated antitumor attack because engaging both innate and adaptive immunity seems to have a role in increasing the anti-tumor activi ty of monoclonal a ntibodies such as anti-HER2/neu that have already be en de monstrated to be highly effi cacious [98 ].

Expert commentary
Scientifi c advances and successes are frequently accompanied by parallel challenges and pitfalls when the euphoria of success is inevitably tempered by the reality check that accompanies practical application. The fi eld of immunooncology is poised to facili tate a fundamental paradigm shift in the appr oach to cancer treatment through continued gains in our understanding of how to optimize manipulation of the human immune system. It is critically important to acknowledge that therapeutic cancer vaccines comprise a diverse group of complex biologic agents with distinctive kinetics and clinical profi les from standard ch emotherapy. Thus, novel methodologies, predictive biomarkers, immunologic and clinical end points and a shift in the standard clinical development paradigm for chemotherapy drugs will be required to continue to move this evolving fi eld forward [104]. Indeed, the inherent biologic variability of anti-tumor immune responses (whethe r de novo or vaccine induced) and th eir delayed kinetics relative to chemotherapy make identifi cation of baseline attributes that may predict immunotherapy activity a key scientifi c priority. However, it is important to acknowledge that the predictive utility of baseline factors may vary based on the patient population or the specifi c immunotherapy platform being studied.
Investigation of multimodality approaches in combin ation with cancer vaccines in the hopes of ac hieving even better clinical outcomes should be vetted whenever possible in well-designed preclinical animal s tudies to unde rsc ore the scientifi c rationale for carrying out human trials and to ensure these approaches document potential mechanistic synergy rather than antagonism or excess toxicity. With the approval of sipuleucel-T, patients are adminis tered vaccine before initiating chemothe rapy. Ho wever, the fi eld of prostate cancer is changing with t he recent approval of abiraterone acetate and the promis ing data from the MDV-3100 studi es. Currently, abira terone is approved as second-lin e therapy fo llowing docetaxel in patients with metastatic androgenindependent pr ostate cancer. However, emerging clinical data suggest it may gain approval for use pre docetaxel. The question then a rise s as to where immunotherapy would fi t in the treatment paradigm sequence. It ma y be appropriate to still consider initiating immunotherapy before these other treatments or p ossibly even in combination with abiraterone in an attempt to furth er delay the need for systemic chemotherapy. However, it will be important that clinical researchers investigate t he timing and scheduling of immunotherapy with these newer approved agents to gain insight in to how to maximize the clinical benefi t and take advantag e of the currently available therapeutic repertoire. Newer compounds are being explored in the clinic that can also affect immune checkpoints in addition to anti-CTLA-4. Another molecular imm une t arget is PD1, wh ich, through its interaction with B7-H1, leads to Perspective the inhibition of T-cel l function. Anti-tumor respons es have been obse rved that are associated with the blockade of this interaction by an antibody targ eting PD1. Although additional studies a re required, the anti-PD1 antibody preliminarily se ems t o have a better safety profi le compared with anti-CTLA-4 blockade.
Indeed, a ta ntali zing (and potentially infi nite) spectrum of multiple different combinations of vaccine platforms and adoptive cel l therapies, cytokines, chemotherapeutic agents, radiation therapies and checkpoint inhibitors of negative regulation is already possible. However, our goal as a scientifi c discipline should be to gain addit ional fundamental insight from every product under study at every stage of product development and avoid assuming that we defi nitively know anything about what we are really doing when it comes to infl uencing and manipulating the human immune system. Importantly, evaluation of candidate vaccines and immune-based therapy platforms sho uld entail comprehensive examination of innate and adaptive immunity, alterations in host negative regulatory and suppressive mechanisms in response to treatment and characterization of responses in combination with specifi c classes of agents to more rapidly identify synergies, interferences and toxicities.

Five-year view
The development of vaccine platforms that deliver mor e potent danger signals at the time of initial antigen presentation is required to allow triggering of innate immunity and the generation of high-avidity adaptive responses that are functionally potent and prolonged to overcome self-tolerance and reject tumor an tigens. Better vaccines will result in the development of more rapid immune responses that have a greater likelihood of objectively impacting clinical outcomes earlier rather t han later. These approaches will be augmented by the coadministration of agents (sequen tially or concomitantly) designed to dampen immunosuppressive negative regulation and remove the brakes off host tumor-specifi c responses. Immunoge ns and their respective pl atforms will need to promote cross-priming and epitope spreadin g to maximize the breadth of anti-tumor T-cell responses, thereb y minimizi ng the risk of immune evasion and escape. These ambitious scientifi c and clinical approaches will require n ew regulatory paradigms for approval and unprecedented cooperation among academic, biopharma and government stakehold ers with differing intellectual property and fi scal priorities to continue to move the fi eld forward. The present state of the science and current achievements w ere a long time coming. However, with emerging insights into molecular immunity, we are on the cusp of making personalized cancer medicine a more cost-effect ive and practical reali ty.

Financial & competin g interests discl osure
The authors have no relevant affi liations or fi nancial involvement with any organization or entity with a fi nancial interest in or fi nancial confl ict with the subject matter or mater ials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.

Key issues
• Population enrichment and stratifi cation should be incorporated into well-designed early-phas e trials of vaccines and immune -based therapies. Randomization should be used aggressively to be able to detect defi nitive, clinically meaningful signals of activity early in clinical development. • Im mune-base d therapies should incorp orate combination strategies to overc ome self-tolerance to tumor antigens by addressing negative immune regulatio n systemically and in the tumor microenvironment. • Standardization of assays for primary immunologic end points used to characterize responses to therapeutic vaccination is a critical priority th at will allow better identifi cation of valid assays, end points and correlation with clinical outcomes. • Developmental pa radigms that will allow standardized assessment and characterization of the potency and immunogenicity of personalized cancer therapy are needed.

References
Papers of special note have been highlighted as: • of interest