Innovating Administration of Chemotherapy for Better Patient Outcomes

What is the closed-loop drug delivery system you are developing for chemotherapy? 

We are designing a system that is able to keep the concentration of a drug near its target concentration in the blood (i.e., its therapeutic window). Our initial work focused on chemotherapies, where it is critical to keep the concentration above a certain level to have efficacy, but below a certain level to avoid toxicity. 

We call the system the Closed-Loop Automated Drug Infusion Regulator (CLAUDIA), which is designed to allow for the personalization of drug dosing for patients receiving chemotherapy via an intravenous infusion. 

How does it work? 

With CLAUDIA, we first sample blood from the patient, and then CLAUDIA uses a sensor that can detect the concentration of drug in the blood of the patient. That concentration is put into an algorithm that knows what the desired concentration should be for any given drug. The algorithm then decides if more or less drug should be administered to the patient by the infusion pump and changes the infusion rate to the patient accordingly. 

What makes chemotherapy a good fit for closed-loop delivery? 

Patients are currently dosed on a body surface area (BSA) basis, which leads to widespread pharmacokinetic variability. BSA is estimated using an equation from 1916, over 100 years ago. This equation calculates body surface area using a power law equation that uses a patient’s body weight and height as inputs, and the constants in the equation were fit to data from just nine people.

Even if these equations estimated BSA perfectly, dosing chemotherapy using BSA ignores many additional factors that differ between patients with the same BSA, which results in significant changes in the concentration of drug in the patient’s blood. Given that chemotherapies are highly toxic and have a narrow therapeutic window, patients thus either experience significant toxicities or lack of efficacy. Thus, implementing a closed-loop system for people getting a chemotherapy infusion could have significant benefit for patients. 

"Implementing a closed-loop system for people getting a chemotherapy infusion could have significant benefit for patients."

How does it compare or differ from other continuous monitoring technologies?

The artificial pancreas for diabetes treatment is similar to CLAUDIA in that they are both closed-loop drug delivery systems. Both sense a molecule’s concentration and use an algorithm to change the infusion rate of a drug to keep the target molecule’s concentration within its therapeutic window. With the artificial pancreas, glucose is detected and the amount of insulin given to the patient changes, while with CLAUDIA the drug of interest is both measured and administered to the patient. 

The major difference between CLAUDIA and other closed-loop systems is the sensor. We are using high-performance liquid chromatography-mass spectrometry (HPLC-MS) as our sensor, which has been used and trusted by the FDA  for many years in clinical trials to understand the pharmacokinetics of different drugs. HPLC-MS can measure the concentration of multiple drugs with minor changes to the method with the system, and thus CLAUDIA has the potential to be used to control the concentration of multiple different drugs that are given to the patient either individually or in combination. That is different from continuous glucose monitoring, where the sensor was designed just for glucose. 

What is the algorithm you’re using? 

We used an adaptive, proportional-integral-derivative (PID) controller in the initial work published in Med. A PID controller is a common control algorithm used in several settings, including in some closed-loop artificial pancreas systems. One of the challenges with our system that makes the control problem difficult is the fact that there is widespread pharmacokinetic variability observed clinically. This led us to need to implement the adaptive component of our control algorithm. 

What does that mean? 

Pharmacokinetics is the study of how the concentration of drug changes in the patient’s body when administered to the patient. You can have a patient who is completely deficient in the enzyme that metabolizes a chemotherapy like 5-fluorouracil, and the concentration would greatly exceed the target window leading to toxicities or even death. 

On the other hand, another patient may be a rapid metabolizer of the drug and thus they would thus remove the drug quickly resulting in the concentration of drug being below the therapeutic window. This large difference in the pharmacokinetics of the same drug in different patients is referred to as pharmacokinetic variability. 

"The biggest potential benefit for patients who are already receiving chemotherapies or another drug would be improving therapeutic outcomes while decreasing toxicities."

How did you account for that widespread variability? 

We used a PID controller, but we had to add an adaptive tuning algorithm to take into account some of that widespread variability that is observed clinically. The issue of the widespread variability is that there are some drugs where one set of controller settings would not work for all patients. We addressed this issue by applying an algorithm that changes the controller settings based upon the concentration data that is being collected in real time. 

That allowed us to successfully control the concentration of a drug over a clinically relevant range of pharmacokinetic conditions. We are doing additional work to develop a next generation control algorithm that outperforms our previous algorithm. 

What would be the next logical therapies to expand into? 

We initially focused on applying CLAUDIA to one chemotherapy called 5-fluorouracil. The next step would be applying CLAUDIA to additional small-molecule chemotherapies. We could then expand CLAUDIA into antibiotics, biologics and anesthesia. Any drug that has a narrow therapeutic window and is given by a constant infusion could potentially benefit by being controlled by CLAUDIA.  

Currently, much of the system is done by hand. What could be eventually automated? 

All the different steps were designed with automation in mind. In the initial paper, all the main components were done manually, from the drawing of blood from the animal models to sample preparation, liquid-liquid extraction steps, centrifugation, filtering steps, etc. 

That sample then goes onto the HPLC-MS, which then gives the concentration reading. Then we put that into the control algorithm and adjust the infusion rate. There are ways to automate all those steps using commercially available pieces, as we discuss in the Med paper. 

How does this fit into the existing infrastructure of clinical care and clinical trials? 

The goal is for this to be seamlessly integrated into a normal infusion care setting. The doctor would set the desired concentration of drug into CLAUDIA, the patient would be hooked up to lines as for their normal infusion and CLAUDIA would take care of everything. In terms of clinical trials, CLAUDIA could allow for different drugs to be dosed to reach a certain concentration instead of being dosed on a BSA-basis, which may result in improved rates of success in clinical trials. 

What does this change for patients? 

The biggest potential benefit for patients who are already receiving chemotherapies or another drug would be improving therapeutic outcomes while decreasing toxicities. That’s the hope for personalizing dosing for patients. 

Some patients are being underdosed, which can result in the patient not responding to chemotherapy. It may not be the drugs itself that results in a patient not improving when getting chemotherapy, but it may be the case that the drug levels aren’t high enough to have a therapeutic benefit to the patient. 

The opposite is potentially true as well. If patients have too high a concentration, you could bring them down into the therapeutic window with CLAUDIA. For example, we know that there are genetic polymorphisms for the enzyme that metabolize a drug like 5-fluorouracil and it’s believed that 1% of the population has an enzyme deficiency, which can result in significant toxicities and even death. It’s so serious that a drug that serves as an overdose antidote was approved by the FDA. CLAUDIA could help mitigate or prevent an overdose in these patients by decreasing the infusion rate and bringing the concentration to the therapeutic window. 

What are the next steps? 

The first will be developing and validating the fully automated system. We also need to talk with the FDA to delineate the regulatory needed to bring CLAUDIA to the clinic. We then will take the necessary next steps to bring CLAUDIA to the clinic, where it has the potential to improve the lives of patients.

In This Article