Aspetti metodologici e statistici degli studi clinici in oncologia

Transcript

1 Roma, SSFA, November 11, 2013 Aspetti metodologici e statistici degli studi clinici in oncologia Paolo Bruzzi Epidemiologia Clinica IRCCS AUO San Martino-IST Istituto Nazionale per la Ricerca sul Cancro Genova - Italy Biostatistica Formule matematiche Equazioni complesse Concetti astrusi Metodologia Strumenti critici 1

2 Biostatistica Formule matematiche Equazioni Complesse Concetti astrusi Metodologia Strumenti critici Nota Bene Lettura di articoli scientifici Scrittura di articoli scientifici Referaggio di articoli scientifici Disegno di studi clinici Referaggio di grant applications STESSI STRUMENTI CRITICI 2

3 Coordinate per valutare/interpretare i risultati di uno studio Validita interna: Validita esterna: Coordinate per valutare/interpretare i risultati di uno studio Validita interna: Validita statistica, assenza di bias/errore sistematico SONO VERI? Validita esterna: Possibilita di estrapolare/generalizzare i risultati Applicabilita COSA ME NE FACCIO? 3

4 Coordinate per valutare/interpretare i risultati di uno studio Validita interna: Validita statistica, assenza di bias/errore sistematico Risultati = Verita +/- caso Lettura di articolo: Ricerca di bias, errori (checklist) Validita Interna Identificazione dello/gli scopo/i dello studio Criteri di selezione espliciti Assegnazione casuale e imprevedibile dei trattamenti Rilevazione coerente/unbiased dell endpoint Piano statistico (appropriato) 4

5 Validita interna Nota Bene: L onere della prova e carico di chi presenta lo studio (bias fino a prova contraria) - Dettagliata presentazione di tutta la metodologia, l organizzazione e la realizzazione dello studio - Discussione esplicita di tutti i possibili bias e degli accorgimenti per evitarli/ridurne le conseguenze INFORMAZIONE MANCANTE = BIAS PRESUNTO Validita Esterna = Generalizzabilita? (termine ambiguo) 5

6 Validita Esterna come utilizzare i risultati dello studio? - Per confermare ipotesi biologiche (studi osservazionali es. prognostici) - Per fare altri studi (es. studi di fase II) - Per dimostrare un principio (proof of principle) - Per sviluppare linee guida - Per decidere il trattamento al letto del paziente Chi legge un articolo deve capire la chiave di lettura dello studio = validita esterna Validita Esterna VALIDITA INTERNA! Disegno di studio (Contrasto) Criteri di selezione/caratteristiche dei pazienti Tipologia dei centri partecipanti Protocollo terapeutico- follow-up Endpoint Compliance Contaminazione Precisione delle stime Analisi intention to treat 6

7 Validita interna Protocollo di ricerca Identificazione dello scopo(i) primario(i) Procedure di randomizzazione Scelta endpoint/ mascheramento Piano statistico Persi al follow-up/non valutati Analisi intention to treat Validita esterna VALIDITA INTERNA! Disegno di studio Criteri selezione /Caratteristiche dei pazienti Tipologia dei centri part.ti Protocolli clinici Endpoint Compliance Precisione delle stime Analisi intention to treat Contenuti: Piano Statistico - Endpoint primario e secondari - Indici riassuntivi - Piano di analisi statistica (contrasti, test, analisi per sottogruppi) - Livelli di significativita - Potenza- Sample Size - Interim analyses 7

8 Piano Statistico - Endpoint primario e secondari - Indici riassuntivi - Piano di analisi statistica (contrasti, test, analisi per sottogruppi) - Livelli di significativita - Potenza- Sample Size - Interim analyses Piano Statistico - Endpoint primario e secondari - Indici riassuntivi - Piano di analisi statistica (contrasti, test, analisi per sottogruppi) - Livelli di significativita - Potenza- Sample Size - Interim analyses 8

9 Analisi per sottogruppi Test appropriato Test di interazione Modelli Multivariati Molteplicita Analisi Intermedie Per superiorita Analisi sequenziali (group sequential analyses) Alfa-spending functions Per futility Conditional power Bayesian monitoring 9

10 Contenuti: Piano Statistico - Endpoint primario e secondari - Indici riassuntivi - Piano di analisi statistica (contrasti, test, analisi per sottogruppi) - Livelli di significativita - Potenza- Sample Size - Interim analyses Standard process of development of an anticancer drug ( 60s) Preclinical studies Phase I studies Phase II trial(s) Phase III trial(s) = Randomised Clinical Trial(s) of adequate size 10

11 Conventional Phase I Trials AIM -> Maximum Tolerated Dose Patients -> Refractory to standard treatments (often in different tumors) Design -> Adaptive - Dose increases based on observed toxicity Conventional Phase II Trials AIM -> Response rate worth phase III trials Patients -> Evaluable (i.e. metastatic) Pretrated/Untreated Design -> Uncontrolled Objective response: very specific 11

12 Development of a new cytotoxic drug Phase I Pre-treated pts MTD Phase II Tumour A Metastatic Pts High response rate Tumour B Metastatic pts Low response rate Phase III Metastatic Pts Survival (PFS) Phase III Radically Resected Pts Survival (RFS) Conventional Phase III Trials AIM -> Efficacy/Effectiveness - OS Patients -> Stage Specific Efficacy -> Ideal pts Effectiveness -> Future targets Design -> Classical RCT s (ITT, Double Blind, etc.) 12

13 Conventional Phase III Trials Statistics -> Predefined (Stopping rules for toxicity or efficacy based on predetermined interim analyses) Inclusion Criteria -> Fixed Size -> Huge Similar stiffness in other research areas? NO! (Guided) Trial and Error! Biology Engineering Chemistry Early results and new external knowledge are continuously used to modify the experiments 13

14 Contenuti: Piano Statistico - Endpoint primario e secondari - Indici riassuntivi - Piano di analisi statistica (contrasti, test, analisi per sottogruppi) - Livelli di significativita - Potenza- Sample Size - Interim analyses Per salvaguardare il tasso di errore! Protezione da risultati falsamente positivi: Metodologica: Randomizzazione, Intention to treat, (doppio cieco) Statistica: Piano statistico predeterminato Protezione da risultati falsamente negativi Dimensioni P value! 14

15 Why little (or no) flexibility was allowed in the design of Phase III trials? Frequentist statistical philosophy (see below) - Control of false positive rate (alfa error) No chance of replicating trials If positive, another RCT unethical If negative, replication too expensive Inadequate knowlege of cancer biology Inadequate knowledge of cancer biology Consequences Aspecific drugs (cytotoxic) Unselected patients (Non-small cell lung cancer, Colo-rectal cancer, Soft-tissue sarcoma, etc.) Large trials aimed at small/moderate effects Little chance to learn from early results 15

16 Last 2 decades Dramatic breakthroughs in Genetics Molecular biology Cancer biology (Pathways, Angiogenesis, etc) Molecular techniques and engineering Pharmacology Last 2 decades Cancer molecular prognostic and predictive factors Cancer Targets Drugs with molecular targets (e.g. HER2) Drugs with biological targets (angiogenesis) (Pharmacogenomics) 16

17 Consequences New drugs -> New methods (Phase I-II) New Patients-> New methods (Phase II-III) Prognostic subgroups { Personalized Patients with the target Medicine Evaluable patients New Statistical Methods Perspectives a) Progress in biology/technology b) Personalised Medicine c) Statistical Innovations Radical changes in the way cancer trials are designed and analysed 17

18 Major Changes Sequence/Separation of phases Design of phase I-II-III trials Statistics Rigid separation of Phases Phase I -> MTD -> Dose increases in different groups of patients Phase II -> Activity -> Uncontrolled trial Phase III -> Efficacy -> RCT 18

19 Phase I Trials with modern drugs AIM : MTD? Dose-Response? Dose/Toxicity? Minimal effective dose? Association with chemotherapy Phase I-II trials Phase I-II trials of targeted therapies Study design (for dose-activity)? Randomization? (e.g. different doses) Patients? Duration of treatment? Single agent or + CTX? Endpoint? 19

20 Modern Phase II-III Trials? New Methodology? Modern Phase II-III Trials Challenges: - (Activity Endpoints) - Efficacy Endpoints (RFS-PFS?) - Interim Analyses - Selection of Patients -> Rarity 20

21 Rarity: New problem in cancer research Rare Tumors Rare Cancer Conditions Rare Tumors Rare histologies in frequent sites (e.g. squamous gastric cancer) Rare sites (e.g. uveal melanoma) Both (e.g. astrocytomas, most sarcomas) 21

22 Rare Cancer Conditions Rare tumors Rare presentations (e.g. skin metastases) Prognostic Subgroups Molecular Variants (of common tumors) Molecular Variants All cancers are going to become rare Early Breast cancer Her2+, Postmenopausal, ER+ In BRCA+ women Triple negative with low proliferation and N-. 22

23 Lung ADK: <40% senza driver mutation Best Available Evidence in rare cancers/cancer conditions Often no information/evidence from RCTs focused on the question of interest Studies of questionable validity Indirect(ly pertinent) evidence Subgroup analyses of low power 23

24 Causes of the problem SMALL NUMBERS = LIMITED EVIDENCE Misunderstanding of statistics Conventional statistical reasoning Conventional Statistical Rules A study must have an adequate size Unjustified Implication If an adequate size cannot be attained, no methodological ties Small size Poor quality 24

25 Study protocol Poor Quality? (Classified as Phase II trials) No Randomised controls Lack of planned comparisons with historical controls Primary endpoint: Objective response No statistical plan First take-home message The organization of a trial of small size (according to conventional statistical rules) requires more care in Protocol preparation Study design/methodology Statistical design Addressing Clinical & Organizational issues than a standard size trial 25

26 Causes of the problem Misunderstanding of methodology Conventional statistical reasoning Conventional Statistical Reasoning To reject H0, only evidence collected within one or more trials aimed at falsifying it can be used -> LARGE SAMPLE SIZE No use of External evidence Evidence in favor of 26

27 Example Question: Efficacy of radiochemotherapy in a tumor type very rare in a site (e.g. squamous histology in stomach c.) External evidence: RX+CTX is effective in squamous cancers in more common sites Evidence in favor of..: The response rate in a phase II trial is very high (e.g. 6/10) Does this information affect. - the sample size of the phase III trial aimed to assess RT+CTX in squamous gastric c.? - the analysis of its results (p value)? 27

28 Squamous gastric cancer Planning a trial of RT+CTX Analysing its results (p value) Squamous gastric cancer Planning a trial of RT+CTX Herbal therapy Analysing its results (p value) 28

29 Squamous gastric cancer Planning a trial of RT+CTX Herbal therapy Analysing its results (p value) Conventional (frequentist) statistical reasoning Exclusive reliance on experimental evidence Large Trials (Regardless of the available knowledge) 29

30 Large Trials Implication in rare tumors: Generic Selection criteria (All STS s + Stage + treatment line) - Appropriate for chemotherapy trials - Possibly inappropriate for trials of Targeted Therapies What can be done? 30

31 Methodological recommendations for clinical studies in rare cancers: a European consensus position paper P. G. Casali1,*, P. Bruzzi2, J. Bogaerts3, J.-Y. Blay4, on behalf of the Rare Cancers Europe (RCE) Consensus Panel - Annals of Oncology Adv. Acc. Publ. October 1, 2014 alternative ways to conceive study design, analysis of data and combination of results would be exceedingly important. It is possible that some innovative solutions may imply a price to pay in terms of a higher uncertainty. Recent statistical developments (<10 yrs) in rare cancers Bayesian Statistics Adaptive trials New types of evidence summaries Surrogate endpoints 31

32 Differences between Conventional (Frequentist) and Bayesian Statistics Meaning of probability Use of prior evidence Conventional P Probability of the observed difference (if the experimental therapy does not work) Bayesian Probability Probability that the experimental therapy works/doesn t work 32

33 Differences between Conventional and Bayesian Approaches Meaning of probability Use of prior evidence Conventional P Probability of the observed difference (if the experimental therapy does not work) Bayesian Probability Probability that the experimental therapy works/doesn t work (given observed difference and prior knowledge) 33

34 Conventional (frequentist) statistical reasoning Experimental evidence Bayesian statistical reasoning Experimental evidence + Previous Knowledge Proposed (Bayesian) methodology Prior information probability distribution of the likely effect of the experimental treatment + Trial results (if necessary and possible) = Posterior Probability distribution of the likely effect of the experimental treatment (range of plausible effects) 34

35 Example Mortality Tumor X Nil vs A 15% vs 10% N=2000 P = H0 Rejected: A is effective in X Example Mortality Tumor X Nil vs A 15% vs 10% N=2000 P = Tumor Y Nil vs A 15% vs 7.5% N= 240 P=0.066 H0 not rejected: A not shown effective in y 35

36 Prior Information: X and Y are BRAF+ Mortality Tumor X Nil vs A 15% vs 10% N=2000 P = Tumor Y Nil vs A 15% vs 7.5% N= 240 P=0.066 Prior Information: X and Y are BRAF+ A = Anti BRAF Mortality Tumor X Nil vs A 15% vs 10% N=2000 P = Tumor Y Nil vs A 15% vs 7.5% N= 240 P=0.066 INTERPRETATION? 36

37 Interpretation of the two trials CONVENTIONAL Tumor X: P = Tumor Y : P= Efficacy of treatment A Proven in X NOT PROVEN in Y Interpretation of the two trials BAYESIAN (Posterior) Probability that treatment A significantly (HR<0.8) lowers mortality in tumor X: >90% in tumor Y: >90% 37

38 Prior Evidence and Scientific Evidence Prior evidence is a crucial component in the interpretation of any finding (e.g. X-ray) Less direct evidence is required for decision when prior evidence is taken into account Bayesian statistics allows to conjugate prior evidence with trial results Disadvantages of Bayesian Statistics It is (felt as) Subjective Arbitrary Amenable to manipulations (pharma companies?) 38

39 Advantages of Bayesian Statistics Reflects human reasoning ( common sense ) It is focused on estimates of effect Provides a conceptual framework for medical decision making IT IS TRANSPARENT Recent developments (<10 yrs) in rare cancers Bayesian Statistics Adaptive trials New types of evidence summaries Surrogate endpoints 39

40 Systematic Reviews in rare cancers Need to use information from studies less than 100% VALID less than 100% PERTINENT TO THE QUESTION OF INTEREST, Weighted on the basis of their quality and pertinence Sources of evidence - (Randomised Trials) - Biological & Preclinical Studies - Case-reports - Uncontrolled studies - Studies with surrogate endpoints - Studies on the same cancer in different stages - Studies on cancers in different sites with similar features (molecular changes)? 40

41 If a new drug Note - with a well identified molecular target - which is present in subgroups of different tumors - shows strong clinical effects in one of these (usually the most frequent) it may become no longer ETHICALLY acceptable to start a Randomised trial in the other tumors CML -> Large RCT GLEEVEC GIST -> Large uncontrolled trial Other rare indications -> Case Series (dermatofibrosarcoma protuberans, plexiform neurofibromas, chordomas) 41

42 New paths to drug use Large RCT in a frequent cancer with the target - Proof of principle Toxicity Uncontrolled (but formal) trial(s) in other (rare) cancers with the target Off label use in individual cases with the target New paths to drug use Large RCT in a frequent cancer with the target - Proof of principle Toxicity Uncontrolled (but formal) trial(s) in other cancers with the target Off label use in individual cases with the target 42

43 Proposal Tan SB, Dear KB, Bruzzi P, Machin D. Strategy for randomised clinical trials in rare cancers. BMJ Jul 5;327(7405):47-9. Each piece of information (study) has to be used, weighted according to its: Precision Quality Pertinence (relevance to the study question) Differences between the present and the proposed approach Present : Rational but informal integration of the available knowledge (NCCN 2A) Proposed Formal, explicit and quantitative integration of the available knowledge Verifiable quantitative methods Sensitivity analyses Focus on summary effect estimates 43

44 Example: new approach New drugs in Pediatric melanoma Available Evidence: 1. RCT s in adults (completed, positive, indirect) + 2. Uncontrolled trials in children (ongoing) Pertinent but invalid Assembling evidence in rare cancers Need to develop and validate new (metaanalytic) approaches to summarize prior information in rare tumors Requirements Explicit Quantitative Reproducible 44

45 Recent developments (<10 yrs) in rare cancers Bayesian Statistics Adaptive trials New types of evidence summaries Surrogate endpoints Conventional Trials Methodology Rigid separation of PHASES of development Phase I -> MTD -> Dose increases in different groups of patients Phase II -> Activity -> Uncontrolled trials Phase III -> Efficacy -> RCT s 45

46 Conventional Trials Methodology Within each phase/trial One fixed primary aim/endpoint Fixed eligibility criteria Fixed treatment protocol Prespecified type/number/timing of analyses Prespecified sample size Within each phase/trial Adaptive trials One primary aim/endpoint Fixed eligibility criteria Fixed treatment protocol Prespecified type/number/timing of analyses Prespecified sample size 46

47 FDA s Guidance applies to A&WC studies A&WC Adequate and well-controlled effectiveness studies intended to provide substantial evidence of effectiveness required by law to support a conclusion that a drug is effective = Pivotal trials: BOTH PHASE II & III (Exploratory Studies)? 47

48 Adaptive design clinical trial FDA s Definition: a study that includes a prospectively planned opportunity for modification of one or more specified aspects of the study design and hypotheses based on analysis of data (usually interim data) from subjects in the study Why adaptative designs are so attractive? Early endpoints Molecular Endpoints Response rate It is possible to obtain, during the trial, crucial information to improve some of its design features 48

49 Most conventional trials have an adaptative component Stopping rules based on interim analyses: Toxicity Rejection of null-hypothesis Futility Examples of well-understood adaptive designs Adaptation of eligibility criteria. (per facilitare il reclutamento) Adaptations to maintain study power. (es. Se bassa incidenza, aumenti n. pz o durata f.-up) Adaptations for stopping early. (Interim analyses x Futility, Efficacy) 49

50 Less well-understood types of adaptations Adaptations for Dose Selection Studies. Adaptive Randomization Based on Relative Treatment Group Responses Adaptation of Sample Size Based on Interim-Effect Size Estimates Adaptation of Patient Population Based on Treatment- Effect Estimates Adaptation for Endpoint Selection Based on Interim Estimate of Treatment Effect Adaptation of Multiple-Study Design Features in a Single Study Adaptations in Non-Inferiority Studies Less well-understood types of adaptations Adaptations for Dose Selection Studies In studi di fase II-III, puo essere scelto - il dosaggio con il miglior rapporto tossicita / attivita - il regime piu tollerato - l associazione + attiva o - tossica 50

51 Less well-understood types of adaptations Adaptations for Dose Selection Studies. Adaptation of Patient Population Based on Treatment-Effect Estimates Viene selezionato il/i sottogruppo/i dove si osserva l effetto piu marcato del trattamento sperimentale Seamless Phase II-III Trials 51

52 Less well-understood types of adaptations Adaptations for Dose Selection Studies. Adaptation of Patient Population Based on Treatment-Effect Estimates Adaptation for Endpoint Selection Based on Interim Estimate of Treatment Effect Sulla base dei risultati di attivita si sceglie lo specifico target del trattamento (es. funzionalita vs sintomi, difficilmente utilizzabile in Oncologia) Less well-understood types of adaptations Adaptations for Dose Selection Studies. Adaptation of Patient Population Based on Treatment- Effect Estimates Adaptation for Endpoint Selection Based on Interim Estimate of Treatment Effect Adaptation of Multiple-Study Design Features in a Single Study Adaptations of Superiority/non-Inferiority margin Sulla base della tossicita osservata, si ridetermina il margine di non inferiorita o di superiorita 52

53 I-SPY2 Phase II trial in neoadjuv. setting (>3cm BC) Primary Endpoint: pcr Adaptive design Primary Aim: Compare the efficacy of new drugs in addition to standard CTX vs CTX Seconday Aim: Test, validate and qualify biomarkers of activity of new drugs GOAL: Identify subgroup-specific treatments I-SPY2 Main features Bayesian design Pre-trial-> predictive (posterior) probability of efficacy High predictive P: Regimens graduated with their corresponding biomarker Low predictive P: Regimens dropped Biomarkers: Standard (Eligibility) Qualifying Exploratory 53

54 I-SPY2 Main features Adaptations: Bayesian Adaptative randomization based on accumulating data in biomarkers subgroups Futility analyses to drop regimens uneffective in all subgroups Sequential monitoring to graduate a drug+biomarker pair when adequate predictive P is achieved -> phase III trial Informatics Portal to integrate and interpret huge amounts of complex and disparate data Concerns associated with adaptative design 1. Bias Associated with the Multiplicity of Options (False Positive results) 2. Difficulty in Interpreting Results When a Treatment Effect is Shown (Overestimate of effect size) 3. Operational Bias 54

55 Concerns associated with adaptative design - Blinded interim analyses - Unblinded interim analyses Blinded interim analyses Adaptations of eligibility criteria. Adaptations to maintain study power. based on blinded interim analyses of aggregate data. No bias No need to be (but better if) planned in advance 55

56 Adaptations based on unblinded analyses for stopping early for dose selection studies of patient subgroups based on treatment-effect estimates for end-point selection based on interim estimates of treatment effect Bias if not addressed in study design Requirements of adaptative trials Adaptations based on unblinded analyses - Planned in advance (study design-protocolstatistical plan-sops) - Statistical adjustments (larger sample size stricter p values Bayesian approach) - Specialised statistical support 56

57 Unplanned adaptations based on unblinded analyses Undermine the statistical validity of the study Compromise the possibility to interpret its results ALSO IN EXPLORATORY STUDIES Adaptive studies Bayesian Statistics 1. To monitor study results for early stop 2. To use validated surrogate endpoints 3. To extrapolate results across cancers of different sites bearing the same target 57

58 Adaptive trials Strong preclinical evidence of activity Adaptive trials Strong preclinical evidence of activity Phase IA: Safety 58

59 Adaptive trials Strong preclinical evidence of activity Phase IA: Safety Phase I-II Dose-response Randomised Comparison Biological endopint Stop if no activity Unselected pts Adaptive trials Strong preclinical evidence of activity Phase IA: Safety Phase I-II Dose-response Phase IIB Randomised Comparison Biological endopint Stop if no activity Unselected pts Surrogate endpoint (to select patients) 59

60 Adaptive trials Strong preclinical evidence of activity Phase IA: Safety Phase I-II Dose-response Phase IIB Randomised Comparison Biological endopint Stop if no activity Unselected pts Surrogate endpoint (sel. Pts) Efficacy (OS, PFS, RFS) Adaptive trials Strong preclinical evidence of activity Phase IA: Safety Phase I-II Dose-response Phase IIB Randomised Comparison Biological endopint Stop if no activity Unselected pts Surrogate endpoint (sel. Pts) Stop if no difference with control arm (futility) or toxicity Efficacy (OS, PFS, RFS) 60

61 Adaptive trials Strong preclinical evidence of activity Phase IA: Safety Phase I-II Dose-response Phase IIB Randomised Comparison Biological endopint Stop if no activity Unselected pts Surrogate endpoint (sel. Pts) Stop random if suff. evid. of efficacy Stop if no difference with control arm (futility) or toxicity Efficacy (OS, PFS, RFS) RCT -> EBM in Oncology Golden Age Rigid protocols Drugs Doses Cycles Modifications for toxicity or progression/relapse Generic Selection Criteria Site (e.g. Stomach) Histology (ADK vs Lymphoma) Stage (early vs late) 61

62 RCT -> EBM in Oncology Golden Age Large and Simple Clinical Trials - Systematic Reviews Meta-analyses Clinical Guidelines/Recommendations Generic Flexibility in pt management not considered Ideologia Sottostante: Equita Prospettiva di sanita pubblica (es. vaccinazioni) Piccoli effetti x grosse popolazioni = grandi benefici Terapie semplici, eguali x tutti (trombolisi, Tamoxifen) Pazienti non selezionati 62

63 Evidence Based Medicine Cookbook Medicine? New generation of efficacy trials in rare cancers Uncontrolled efficacy (phase II-III) trials of high quality Randomized activity (Phase II) trials followed by uncontrolled efficacy trials (with historical controls) RCT s with surrogate endpoints Adaptive, Bayesian, activity/efficacy RCT s 63

64 Previous Conclusion Planning of new trials with innovative design requires more care in.. Study Design Statistical Plan Protocol preparation Organization than a conventional trials New Conclusion Planning of new trials with innovative design requires more care in.. Study Design Statistical Plan Protocol preparation Organization than a conventional trials 64

65 Phase I-II trials of new therapies Critical issues 1. Endpoint 2. Design 3. Selection of patients 65

66 Endpoint= Most critical choice in the design of a phase II trial preclinical studies studies in other cancers studies with similar treatments feasibility Wrong endpoint effective treatment discarded Endpoints of Phase I-II trials Activity endpoints vs Surrogate (clinical) endpoints 66

67 Endpoints in phase II trials Phase II trials, if possible, should use Activity endpoints Stronger rationale for phase III trials Smaller size Negative Phase III trials can be interpreted Activity Endpoints = Biological Endpoints Activity of Targeted Agents = Biological Activity - Markers on tumor tissue or circ. tum. cells, - Markers in Plasma/serum, - Metabolic imaging, - etc. NEOADJUVANT PHASE II TRIALS 67

68 Designs Biopsy Window of opportunity study Experimental treatment (short term) Surgery / evaluation of activity Window of opportunity studies = Phase I-II studies Poorly Understood biologic effects: Exploratory Biologic-Molecular studies Expected mechanism of action against the proposed target= hypothesis driven studies Preliminary markers of tumor resistance or sensitivity: (-omics, tumor markers, host factors) To define a dose that effectively inhibits the drug target (MED) = dose-response studies. 68

69 Endpoints in Phase II (I-II) trials 2. Clinical Endpoints Objective response Time to progression/relapse Clinical Benefit (QoL, improvement in symtoms, combined endpoints) 2) Designs in Phase II (I-II) trials 1. Single-arm designs 2. Non-comparative randomised trials 3. Randomised comparative Phase II trials Necessary for non specific endpoints Sample Size < than in Phase III trials Various designs (e.g. random discontinuation) 69

70 Randomised Phase II (I-II) trials Reduced Sample Size Large Effect (esp. for biological endpoints) Relaxed significance level 0.10 (-23%) 0.20 (- 47%) Stopping rules for evidence of futility (clearly inactive treatment) 3) Selection of patients for Phase II trials Bearing the (presumed) target Increased power Only if strong biologic rationale Risk if wrong target 70

71 Phase II (I-II) trials Selection of patients Bearing the (presumed) target All patients Identification a posteriori of markers of susceptibility (targets) Much larger sample size (than needed if true target) Phase II (I-II) trials Selection of patients With the (presumed) target All patients Enriched populations (unbalanced sampling of pts with the target) 71