Chemotherapy-Drug-dose-optimization

---

Introduction

Cancer involves uncontrolled cell growth, forming tumors that disrupt normal tissue.

• Benign tumors have limited growth.
• Malignant tumors spread.

They indicate a breakdown in the body’s control mechanisms, causing unregulated cell growth (neoplastic diseases).

Cancer treatment primarily involves:

• Surgery – common for benign tumors and early malignant cancers, but less effective at late stages.
• Radiotherapy – targets localized tumors but may damage surrounding tissues.
• Chemotherapy – effective against widespread cancer cells.

Modern treatment often combines these approaches.

---

Problem Statement

Chemotherapy’s effectiveness is often limited by:

1. Drug toxicity
2. Drug resistance

We need an optimization scheme that:

• Identifies optimal dosage and schedule.
• Maximizes therapeutic effect.
• Minimizes side effects.

This study develops a multi-objective optimization scheme for single-drug chemotherapy using GAMS.

---

Mathematical Formulation

Mathematical models in cancer treatment are:

• Cost-effective compared to lab or clinical trials.
• Able to bypass ethical limits by simulating scenarios not possible in humans.
• Useful for exploring complex biological systems.

Tumor–Immune–Drug Model

We consider three interacting populations: normal cells $N(t)$, tumor cells $T(t)$, immune cells $I(t)$, and a control/drug variable $u(t)$ (drug concentration or effect at the tumor site).

System of ODEs

Normal cells
(1)

Tumor cells

(2)

Immune cells

(3)

Parameters:
$r_N,r_T$ (growth), $K_N,K_T$ (carrying capacities), $\alpha_N,\alpha_T,\alpha_I$ (drug toxicity),
$\beta$ (immune killing), $s$ (baseline recruitment), $p,q$ (tumor-driven stimulation), $d_I$ (immune death).

---

Objectives

Minimize tumor burden over the horizon $[0,,\mathcal{T}]$

(4)

Minimize drug usage/exposure

(5)

---

State Constraint (patient safety)

Maintain normal cells above a threshold:

(6)

---

Drug Kinetics and Bounds

(Optional explicit drug dynamics)

(7)

Dosing bounds (admissible controls):

(8)

---

Compact State-Space Form

Let $x = [,N,;T,;I,;u,]^\top$ (or omit $u$ if using $u(t)$ directly), then

(9)

Initial conditions:

(10)

Scalarized (weighted-sum) single-objective (if desired):

(11)

Multi-objective formulation:

(12)

Results and Discussion

Using GAMS-based optimization (example outcome):

• Initial tumor count: 0.25
• Final tumor count: 0.082

Key findings:

• Significant tumor reduction.
• Achieved with minimal drug dose.
• Reduced adverse effects compared to higher doses.

---

Conclusion

Optimal Dosages and Schedules via GAMS

• GAMS identifies effective dosages and schedules.
• Enhances treatment efficacy while reducing side effects.

Efficiency of Mathematical Programming

• Faster and more effective than trial-and-error.
• Streamlines decision-making in treatment planning.

Potential Impact

• Improved outcomes and quality of life via tailored, model-guided therapy.

---

Future Work

• Extend to multi-drug regimens (combination therapy).
• Use Genetic Algorithms or regimen modification heuristics for complex, nonconvex schedules.
• Incorporate richer pharmacokinetics/pharmacodynamics and toxicity constraints.

---

Anti‑Cancer Drugs and Side Effects

Drug	Bone Marrow	Kidney	Nausea/Vomiting	Heart	Peripheral Nerves
Adriamycin	+++	–	++	++	–
Epirubicin	+++	–	++	+	–
Cisplatinum	++	+++	+++	–	+++

Notes:

• + signs represent severity of side effects.
• In multi-drug schedules, risks add up.
• Avoid pairings with excessive cumulative risk (e.g., Adriamycin + Epirubicin) via explicit constraints.