The Spring School challenge is over. You can watch the solution webinar and download the solution material below.
Download the challenge solution material (runs and slides)
Download the webinar Q&A
Challenge presentation
Put your new modeling and simulation know-how in practice and try to win the prize!
This challenge is part of the 2021 Lixoft Spring School and only those who have registered for the Spring School can participate. The goal is to develop a model for a phase I data set and use the model to plan the phase IIa clinical trial. Try to find the best model and best phase IIa trial design!
The data set has been designed to be of progressively increasing difficulty. Don’t hesitate to start!
Download the challenge starting material
Background information
Check the introductory slides: DOWNLOAD
The (hypothetical) drug is a small molecule developed to treat hypertension. It is acting as a brain aminopeptidase A (APA) inhibitor. Inhibition of brain APA, which converts Ang II into Ang III, has been proposed as an anti-hypertensive treatment.
It is believed that transporters play an important role during absorption. The metabolism of the drug leads to the formation of a metabolite through hydroxylation, probably via cytochromes CYP3A4 and CYP2D6. A first-pass effect is uncertain. The drug is the likely effector of the effect, but an effect of the metabolite is not excluded at this point.
This is the data from a phase I, double-blind, placebo-controlled, ascending single-dose study with iv or oral administration in healthy subjects. The drug and metabolite concentrations have been recorded, as well as the systolic blood pressure (PD).
CYP2D6 polymorphism in the population may lead to high variability in the drug clearance with fast and slow metabolizer. Therefore the CYP2D6 genotypes have been investigated and are recorded in the data set as “wild type”, “heterozygous mutated” and “homozygous mutated”.
Dataset description
The challenge_dataset.csv dataset contains the drug concentration (DVID=”1_Drug”, in nM), metabolite concentration (DVID=”2_Metabolite”, in nM) and blood pressure (DVID=”3_BP”, in mmHg) for 56 individuals split in 6 groups of single ascending doses and one placebo group. The meaning of the columns is the following:
- ID: subject identifier
- TIME: nominal time (in h)
- AMT: dose amount (in nmol)
- ADMID: 1 for IV dose and 2 for oral doses
- DV: drug concentration (in nM), metabolite concentration (in nM) and blood pressure (in mmHg)
- DVID: 1_Drug for drug concentration, 2_Metabolite for metabolite concentration, 3_BP for blood pressure
- CENS: 1 if observation below the LOQ (LOQ=0.5 nM for drug and metabolite), 0 otherwise
- LIMIT: lower limit for the BLQ data (0 for all)
- ROUTE: “oral” or “iv”
- DOSEGR: dose group 1, 5, 15, 25, or 35 (in umol)
- SEX: “F” for female, “M” for male
- CYP2D6: CYP2D6 genotype wild type, mutated heterozygous, or mutated homozygous
- AGE: age (in years)
- WT: weight (in kg)
- AST: serum aspartate aminotransferase (IU/L)
- ALT: serum alanine transaminase (IU/L)
- ALBUM: serum albumin (g/L)
- BILI: serum bilirubin (mg/L)
The cov_phaseII.csv file contains the same covariates as above, extracted from previous phase II/III trials for the same indication. They fulfill the inclusion/exclusion criteria we need for phase IIa. It is useful only for the Simulx part of the challenge.
Instructions
(and more tips further down!)
Monolix
Find the best (in terms of BICc via linearization) joint model for the drug concentration, metabolite concentration and blood pressure data (full dataset including the BLQ data). Ensure that your model is not over-parameterized by checking that all RSE values via linearization can be calculated.
Simulx
The phase IIa clinical trial will be a parallel trial comparing a single treatment arm with oral multi-doses to a placebo arm. The covariates of eligible individuals are provided in the challenge material. These subjects have hypertension and are on average older than the healthy individuals of the phase I.
The “outcome” for each individual is the change in the 24-hour mean systolic blood pressure from baseline. In practice the blood pressure is measured once at the treatment start and every 4 hours over 24h at treatment steady-state. The measures taken over the 24h interval are baseline corrected and averaged.
-
- Step 1: find the oral multi-dose regimen that leads to the largest change in the 24-hour mean systolic blood pressure from baseline, while ensuring that at least 99% of the individuals have Cmax(drug) < 450 nM and Cmax(metabolite) < 120 nM at steady-state. The available tablet strengths are 1000 nmol and 5000 nmol. Several tablets can be combined to obtain greater amounts.
OD (once per day), BID (twice per day) and TID (three times a day) dosing regimens can be considered.
- Step 1: find the oral multi-dose regimen that leads to the largest change in the 24-hour mean systolic blood pressure from baseline, while ensuring that at least 99% of the individuals have Cmax(drug) < 450 nM and Cmax(metabolite) < 120 nM at steady-state. The available tablet strengths are 1000 nmol and 5000 nmol. Several tablets can be combined to obtain greater amounts.
-
- Step 2: find the minimal number of individuals per arm that need to be recruited in a parallel phase IIa trial comparing the change in the 24-hour mean systolic blood pressure from baseline in the treatment group (select the dosing regimen found in step 1) versus the placebo group. The trial is considered successful if the difference of the arithmetic means of the two arms is significantly larger than 10 mmHg.
Winner
Only joint models for the 3 types of observations and for which all RSE via linearization have been estimated will be considered. The compliant models will then be ranked according to the BICc calculated via linearization. Participants having provided models with equivalently low BICc will be separated based on their proposal for phase IIa. At equivalently low BICcs, the biological relevance of the model and covariate effects will also be taken into account.
Prize
The winner will receive a $250 Amazon gift card and the challenge trophy!
Practical information to send your results
Send your best Monolix run as a zip and your two simulx projects as a zip by April 16th at 11pm PDT per email to springschool@lixoft.com. By sending your results you accept your name to be made public if you win. The object of the email should be “Challenge Firstname Lastname”.
Monolix zip:
- Name the zip Monolix_Firstname_Lastname.zip
- Include the .csv data set, .txt model file, .mlxtran monolix file, and result folder of your best run. Name your monolix project as Firstname_Lastname.mlxtran. Include only your best run (one run per participant). See the Tips section for more detailed guidance.
Make sure that:
- the “standard errors” and “likelihood” task have run using the “use linearization method” option
- the RSE values are displayed for all parameters in the tab Results > POP.PARAM
- the results are reloaded when you open the .mlxtran file in Monolix.
Simulx zip:
- Name the zip Simulx_Firstname_Lastname.zip
- Include the two .smlx files (step 1 an step 2) and the two result folders. Name your simulx projects as Simulx1_Firstname_Lastname.smlx (step 1) and Simulx2_Firstname_Lastname.smlx (step 2). See the Tips section for more detailed guidance.
Make sure that:
- the results are reloaded when you open the .smlx files in Simulx.
Tips
You can find resources and answers to your questions on https://lixoft.com/lixoft-university/ (list of all our videos) and on the documentation website of each application: http://mlxtran.lixoft.com/, http://monolix.lixoft.com/, http://simulx.lixoft.com/ and https://sycomore.lixoft.com/.
Do not forget to check the Spring School material: https://lixoft.com/lixoft-university/spring-school-2021/.
Monolix
- Tag the ‘CENS’ column as CENSORING and ‘LIMIT’ column as LIMIT, in order to take into account the full dataset, including the BLQ data. If you ignore the BLQ data, your BICc cannot be compared with runs that do include the BLQ.
- Proceed stepwise:
- Work on the drug data only first, then on a joint model for drug and metabolite and finally add the blood pressure. The full model runs in 15 min on a standard laptop.
- Add covariates before investigating correlation between random effects.
- Keep the option “use linearization method” ticked all time, as the BICc calculated by linearization will be used to rank the received models.
- For your last run, go to Settings > Project settings and tick “Save the data set and the model beside the project”. Do a “save as” and give Firstname_Lastname.mlxtran as name. Now all necessary files are in the same folder. Zip together the data set, .txt model file, Firstname_Lastname.mlxtran and Firstname_Lastname result folder.
Simulx
- Import your best Monolix run into Simulx
- Create a treatment element with multiple oral doses and use the Exploration tab to find the order of magnitude of doses fulfilling the safety criteria
- Create a covariate element using the covariate table cov_phaseII.csv provided in the challenge material. They represent individuals eligible for the phase IIa.
- Use the Simulation tab to find the best dosing regimen.
- For the two runs you want to share, go to Settings > Project settings and tick “Save the user files in the results folder”. Do as “save as” and give Simulx1_Firstname_Lastname.smlx and Simulx2_Firstname_Lastname.smlx as names. Now all necessary files are in the simulx result folder. Zip together Simulx1_Firstname_Lastname.smlx, Simulx2_Firstname_Lastname.smlx, Simulx1_Firstname_Lastname result folder and Simulx2_Firstname_Lastname result folder.
GOOD LUCK !
Spring School challenge