SYNTHETIC PATIENTS
Your talk at the Clinical Trials Oncology Forum focused on synthetic patients: What advantages do synthetic patient trials have over trials conducted with physical patients? Traditionally, a randomised clinical trial (RCT) has two study arms: the experimental or active arm, which consists of trial subjects who receive the new treatment, and the control arm, where subjects receive a placebo or standard of care (SoC) treatment regimen. I have discussed replacing the control arm patients with ‘virtual’ or ‘synthetic’ patients derived from existing data repositories. To me, there are certain potential advantages to working with synthetic trial patients in clinical trials. Populating the control arm in a clinical study with patients to be treated with a SoC treatment regimen feels redundant if a large data pool is available with results based on the SoC intervention. In other words, why collect data and expose patients to a clinical study if the data points you need are already out there. The second argument in favour of this approach is, perhaps, more prosaic. By reducing or replacing the number of ‘active’ control patients, there is a potential saving in trial duration and costs. A third benefit worth mentioning is that of using synthetic trial patients in pivotal studies in rare cancer types, where the number of potentially eligible study patients is small and the limited number of available real patients will only be treated in the experimental study arm.
Which applications are needed when working with synthetic patients in trials?
First, a data repository with clinical data to populate the synthetic control arm (SCA) of your trial. Data can be incorporated with external – real-world evidence (RWE) – data from historical clinical trials, electronic medical records, or patient registries. Second, computer algorithms are needed to build patient-level data from the RWE data sets matched one-to-one with the active trial arm.
What are the potential challenges to introducing virtual patient solutions? It is key to minimise the difference between the patient characteristics in the active treatment arm and the RWE-sourced control arm. With correct selection of external data, the control arm can be made robust to known confounders. Obviously, data in the control arm cannot be controlled for unknown confounding variables. Probably, the most favourable external data source are patients from
large, well-conducted RCTs, rich on baseline characteristics similar to those of the target population (see: Thorlund in Clinical Epidemiology 2020). Using techniques like propensity weighting or dynamic borrowing, it is possible to construct an SCA with patient data based on the same geographical setting, and similar across a large spectrum of demographics, plus concordance in disease setting and outcome definitions. My own example is based on the treatment of patients with metastatic Castration Resistant Prostate Cancer (mCRPC) with the SoC taxane drug docetaxel. Docetaxel (intravenous administration of 75 mg/m2 docetaxel every three weeks) has been the SoC treatment modality for this disease population ever since the pivotal Phase III trial by Tannock et al in 2004. After this initial 2004 trial, at least nine other large Phase III RCTs have been completed – with several more studies still ongoing – using the same docetaxel treatment regimen for mCRPC. The total number of patients treated with docetaxel exceeds 4,500 in these pooled studies. These results have the potential to create a large number of suitable synthetic patients for the control arm in the Phase III mCRPC trial that we are planning ourselves. However, back in 2004, when mCRPC patients were treated with docetaxel in the Tannock trial, they had received very little to no prior treatments for their disease condition. In recent years, multiple new drugs have been registered for the mCRPC indication, all of which are given prior to docetaxel, including abiraterone, enzalutamide and apalutamide. Moreover, several novel agents have also reached the market for earlier disease stages. Specifically, for non-metastatic Castration Resistant Prostate Cancer (nmCRPC), these are enzalutamide, darolutamide and apalutamide; and for Metastatic Castration-Sensitive Prostate
“New phenotyping screening tools offer the possibility to improve the selection of viable drug candidates in pre-clinical development. The technologies generate tumour models that capture original heterogeneity and micro-environment”
| 37
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50
