Exploring Response of Biomarkers in a Clinical Trial

Posted on Jul 25, 2016


Biomarkers are widely used in clinical research and their presence as primary endpoints in clinical trials is now accepted almost without question. What is a biomarker? According to FDA, a biomarker is a characteristics that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions. Molecular, histologic, radiographic, or physiologic characteristics are types of biomarkers. A familiar example is the use of blood glucose levels as a biomarker to measure the effectiveness of a diabetes medication.

This study looks at the way in which biomarkers were used in one study to uncover which biomarkers change as a result of medication, and which biomarkers can be used to predict adverse effects.  In a Phase I clinical trial, 39 healthy volunteers enrolled in a study and were asked to take a pill daily. Measurements of 20 different biomarkers were collected before the subjects started taking the pill (week 0) and on a weekly basis thereafter (from week 1 to 8). The values of some biomarkers might change as a result of the treatment and thus reflect the effectiveness of the pill. 12 volunteers had an adverse event while on study. Adverse event is any unfavorable and unintended sign, symptom, or disease temporarily associated with the use of the pill, whether or not considered related to the pill.  The change of some biomarkers might show certain patterns before the event and therefore help to predict the occurrence of the later event. This prediction could be useful in deciding whether and when a volunteer should stop taking pills during the course of treatment. To sum up, two questions that are of interest: 1) What biomarkers are changing as a result of the treatment? 2) What biomarkers might work as indicator of the adverse event?


What biomarkers are changing as a result of the treatment?

When comparing measurements of all biomarkers during the treatment (week=1-8) with the corresponding measurements before taking the pills (week=0) in the line chart below, it seems that the values of biomarker M1, M2, M3 and M4 generally increase during the course of treatment.   There appears to have positive linear dependence between values of M1, M2, M3, M4 and length of treatment. The rest of the 20 biomarkers instead show no significant difference between weeks. As such, M1, M2, M3 and M4 might have shown response to the treatment.



To further check the change of M1, M2, M3 and M4, the standard deviation of those biomarker values were plotted in the line chart below. There is large standard deviation for M1, M2, M3 and M4, which would potentially make the between week difference of those biomarkers not significant. However, those measurements could be further normalized by taking the ratio of Marker Value during treatment (Week = 1-8) and Marker Value as control (Week=0). There are a few Marker Values at week=0 equal to 0. As such, both the numerator and denominator of the normalized marker value is added by 1, and the new normalized metric named Fold is,

Fold = (Marker.Value(Week=1-10) +1) / (Marker.Value(Week=0) +1).


The line chart of Fold vs Week is shown below. As compared to the chart of Marker Value vs Week above, the standard deviation for M1 and M2 have become much smaller when tracking the Fold change during the treatment. However, M3 and M4 still have relatively large standard deviations, which might be caused by some large outliers. Based on the Fold change during the treatment, M1 and M2 might work as better biomarkers in response to the treatment since they have less variance between volunteers.


What biomarkers might work as indicator of the adverse event?

To study which ones of the M1, M2, M3 and M4 biomarkers’ changes are COnrelated to adverse events, the volunteers were grouped by whether they had an adverse event or not during the course of treatment. 12 of them who had an event were grouped together and labelled with Event=1 and the left 27 were grouped with Event=0. The two groups were compared in the line chart below.


In the chart for M1, the two groups have relatively large deviation at Week=6. A large deviation between groups is also shown in the chart for M3. Among the 12 volunteers who had an adverse event, 7 of them had the event at Week 7 which is one week after the large deviance occurred for M1 and M3. In the chart for M4, the largest gap between the two line charts is at Week 3, which is one week before 4 volunteers experienced an adverse event. There is 1 volunteer who had the event at Week 6. In chart for M3, the two groups also had large differences at Week 5, which is one week before the event occurred for that volunteer. As such, the between group difference one week before the adverse event might be a potential predictor for the upcoming event.

Since the adverse effect occurred at Week=4, 6, 7, the box-and-whisker plots below drill down and further analyze the between group difference one week before the event. The distribution of Marker Fold for M1 at Week 3 and 6, and that for M3 at Week 3, 5 and 6 appear to be different between groups. Since the sample size from group with event and group without event are not equal, Welch’s two sample t-test is applied to examine those differences. However, none of the 12 pairwise differences demonstrated in the chart below are statistically significant at alpha=0.05 level. It’s worth noting that 12 of those 24 distributions shown here have some outliers beyond the reach of whiskers. For instance, in the group without event at Week 6 for M1, there are three outliers above the upper whisker of the box.  The sample size for each group is between 12 and 27.  Therefore the influence of even one outlier to the group mean might be significant.  If those outliers are removed, the sample mean from this group will be smaller and its difference with the sample mean of group with event will get bigger. Then the two sample t-test might have to accept the hypothesis of difference of mean between groups.


Concluding Remarks

M1,M2,M3,M4 might work as biomarkers that respond to the treatment.

M1 and M2 might be used as predictors for the adverse event.

There are some outliers that need to be further understood.

About Author

Ricky Yue

As a data enthusiast, Ricky loves to think the real life issues in a quantitative way. He likes to talk about probability and alternative. He’s proud of his Bayesian skepticism based on years of scientific training. He was...
View all posts by Ricky Yue >

Leave a Comment

No comments found.

View Posts by Categories

Our Recent Popular Posts

View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI