(510g) Data Analytics and Optimization for Minimization of Chemotherapeutic Toxicity

Data
Analytics and Optimization for Minimization of Chemotherapeutic Toxicity

Katherine Schmidt¹,
Alex D’Aloia², Purnima M. Kodate³ and Kirti M. Yenkie²*

¹Department of
Mathematics, Rowan University, Glassboro, NJ, USA

²Department of
Chemical Engineering, Rowan University, Glassboro, NJ, USA

³Department of
Pathology, Government Medical College, Nagpur, India

Cancer is a
leading cause of death worldwide. Approximately 1 in 285 US children will be
diagnosed with cancer before their twentieth birthday [1]. Leukemia,
particularly ALL (Acute Lymphoblastic Leukemia) and AML (Acute Myeloid
Leukemia), are the most common type of childhood cancers. Typical treatment
consists of a long-term chemotherapy regimen, usually incorporating atleast one
anthracycline, which has major limitations including the risk of cardiotoxicity
[2], [3] and
myelosuppression [4], [5]. From 2007 to
2013, the 5-year survival rate of children diagnosed with ALL was 88%.

Despite the high
survival rate, most children will not live a normal life following the harsh
treatment undergone by their bodies. Two-thirds of childhood cancer survivors
will face at least one chronic health condition, while one-fourth will face a
life-threatening toxic effect from the treatment afterwards in life [1]. Thus, it is of utmost
importance to detect cancers at an early stage and implement suitable treatment
strategies to avoid the disease progression as well as minimize any other
health complications resulting from incorrect diagnosis and inefficient
treatment.

One possible way of
detecting the cancer early and identification of its correct sub-type is
through the use of biomarkers. Some biomarkers are also capable of indicating additional
toxic effects to the patient while undergoing cancer treatment. Especially, cardiac biomarkers
are substances released in the blood when the heart is damaged or stressed [6]. These include proteins,
enzymes, and hormones. Markers such as troponin proteins, natriuretic peptides,
and myeloperoxidase have been studied as potential predictors of incidence and
severity of heart failure in cancer patients [7]. In this work we
have identified that myeloperoxidase is the main biomarker associated with
leukemia subtype identification through the use of machine learning methods. Other variables
studied include age, gender, periodic acid-Schiff, and total leukocyte count.

In addition to the
early cancer diagnosis there is a need for appropriate treatment
strategies that can simultaneously balance treatment efficacy and residual patient
toxicity; herein lies an optimization problem. To this end, a systems engineering
based mathematical model for cancer proliferation, tumor
degradation, immunotoxic and cardiotoxic effects was developed and validated to
facilitate the identification of clinical biomarkers responsible for long-term
toxicity. Utilizing the information from the mathematical model, an objective function
was formulated to minimize tumor size while maintaining the leukocyte counts
and cardiovascular characteristics to the desired values. This resulted in an
optimal control problem where the decision variable was the dosing values of
the chemotherapeutic drug and its schedule for achieving the desired treatment
objective. This problem was solved using two different methods: the maximum
principle and discretized nonlinear programming [8], [9].
The dosing profiles predicted by both these method were then substituted in the
model to check for treatment efficacy and the residual toxic effects. Our
overall aim is to develop patient-specific treatment policies which can ensure a
healthier life with no comorbidities for cancer survivors using principles of data
analysis, mathematical modeling, cancer biology, optimization and control.

Keywords:
Leukemia, chemotherapy, systems engineering, mathematical modeling, optimization

References:

[1]  American
Cancer Society, Surveillance Research, “American Cancer Society (ACS) Special
section: Cancer in children and Adolescents,” 2014.

[2]  M. Florescu, M. Cinteza, and D. Vinereanu,
“Chemotherapy-induced Cardiotoxicity,” Maedica (Buchar), vol. 8, no. 1,
pp. 59–67, Mar. 2013.

[3]  S. E. Lipshultz and M. J. Adams, “Cardiotoxicity
After Childhood Cancer: Beginning With the End in Mind,” JCO, vol. 28,
no. 8, pp. 1276–1281, Mar. 2010.

[4]  L. E. Friberg, A. Henningsson, H. Maas, L.
Nguyen, and M. O. Karlsson, “Model of chemotherapy-induced myelosuppression
with parameter consistency across drugs,” J. Clin. Oncol., vol. 20, no.
24, pp. 4713–4721, Dec. 2002.

[5]  A. L. Quartino, L. E. Friberg, and M. O.
Karlsson, “A simultaneous analysis of the time-course of leukocytes and
neutrophils following docetaxel administration using a semi-mechanistic
myelosuppression model,” Invest New Drugs, vol. 30, no. 2, pp. 833–845,
Apr. 2012.

[6]  J. Mair, A. Jaffe, F. Apple, and B. Lindahl,
“Cardiac Biomarkers,” Dis Markers, vol. 2015, 2015.

[7]  E. Christenson and R. H. Christenson, “The Role
of Cardiac Biomarkers in the Diagnosis and Management of Patients Presenting
with Suspected Acute Coronary Syndrome,” Ann Lab Med, vol. 33, no. 5,
pp. 309–318, Sep. 2013.

[8]  U. Diwekar, Introduction to Applied
Optimization. Springer Science & Business Media, 2013.

[9]  K. M. Yenkie and U. M. Diwekar, “Comparison of
different methods for predicting customized drug dosage in superovulation stage
of in-vitro fertilization,” Computers & Chemical Engineering, vol.
71, pp. 708–714, 2014.