Normal and pathological dynamics of platelets in humans

Gabriel P. Langlois, Morgan Craig, A. R. Humphries, Michael C. Mackey, Joseph M. Mahaffy, Jacques Bélair, Thibault Moulin, Sean R. Sinclair, Liangliang WangJournal of Mathematical Biology 2017 · DOI · arXiv

Abstract

This paper presents a sophisticated mathematical framework for understanding how our bodies regulate platelet production and why this system sometimes breaks down into wild oscillations. By integrating the dynamics of the hormone thrombopoietin (TPO) with the life cycle of megakaryocytes (platelet precursor cells), the authors identify TPO receptor dysfunction as a primary driver of cyclic thrombocytopenia.

Introduction

Platelets are essential for blood clotting. In healthy humans, their levels are kept remarkably stable through a feedback loop involving the hormone thrombopoietin (TPO). However, patients with cyclic thrombocytopenia (CT) experience massive, periodic swings in platelet counts—ranging from dangerously low (risk of hemorrhage) to abnormally high (risk of clots).

Previous mathematical models of this process were often too simplified, neglecting the complex TPO-mediated feedback or the distinct stages of megakaryocyte development. The authors sought to build a "physiologically realistic" model that could not only simulate healthy states but also pinpoint which specific biological parameters must change to trigger the pathological oscillations seen in CT patients.

Modeling the Platelet Production System

The authors employed a system of integro-differential equations with constant delays to capture the age-structured dynamics of platelet production.

Schematic representation of the platelet production model, illustrating the feedback loops between thrombopoietin (TPO), megakaryocytes, and platelets.

Key features of this model include:

  • Age-Structured Modeling: Megakaryocyte development is divided into a mitotic phase (cell division, $\tau_m$) and an endomitotic phase (growth in volume/ploidy without dividing, $\tau_e$).
  • TPO Feedback Loop: TPO ($T$) stimulates production, but it is also cleared from the blood by binding to receptors on the very platelets ($P$) and megakaryocytes ($M_e$) it helps create. This creates a "sink" that acts as a crucial governor on the system.
  • Stability Analysis: The authors used linearization and characteristic equations to find Hopf bifurcations—the mathematical "tipping points" where a steady system begins to oscillate.

Methods and Data

The study utilized Approximate Bayesian Computation (ABC) to fit the model to a wide array of existing clinical data, including:

  • Healthy Subjects: Data from studies of healthy humans receiving Romiplostim (a TPO mimetic) to establish baseline parameters.
  • CT Patients: 15 published datasets of patients with cyclic thrombocytopenia. Four of these datasets contained concurrent measurements of both platelet counts and TPO levels over time, which were particularly valuable for validating the model's phase predictions.

Key Insights and Results

The authors found that the healthy hematopoietic system is incredibly robust; varying parameters by 10–20% was insufficient to cause disease-like oscillations. Instead, the model suggests that CT requires a "triple threat" of profound physiological shifts:

  1. Failure of TPO Clearance: A near-total failure of TPO receptor clearance (reduced to an average of 0.13% of normal values).
  2. Increased Platelet Destruction: A massive increase in the rate of platelet destruction (over 2100% increase), likely due to an autoimmune response.
  3. Maturation Delays: A moderate increase in the duration of megakaryocyte maturation.

The model successfully replicated the oscillatory patterns of all 15 CT patients. It correctly predicted that in CT, TPO levels should peak just as platelet levels are hitting their nadir (lowest point), a phenomenon confirmed by clinical observations. Interestingly, the oscillations are not caused by the primary pair of eigenvalues, but a second pair that crosses the imaginary axis, explaining the 20–40 day periods observed clinically.

Discussion and Limitations

While the model provides a powerful diagnostic framework, it has certain limitations. Due to a lack of human data, the authors assumed the hematopoietic stem cell (HSC) population ($Q^*$) remains at a steady state. In reality, HSC oscillations might interact with this system. Furthermore, the model is deterministic and does not account for biological "noise."

While the model identifies that TPO receptor clearance fails, it cannot explain why (e.g., genetic mutation vs. antibody interference). The authors suggest further clinical investigation into TPO receptor density in CT patients is needed to validate these findings and potentially guide future treatments.

Summary updated

BibTeX
@article{langlois2017normal,
  author = {Gabriel P. Langlois and Morgan Craig and A. R. Humphries and Michael C. Mackey and Joseph M. Mahaffy and Jacques Bélair and Thibault Moulin and Sean R. Sinclair and Liangliang Wang},
  title = {{Normal and pathological dynamics of platelets in humans}},
  journal = {Journal of Mathematical Biology},
  year = {2017},
  doi = {10.1007/s00285-017-1125-6},
}
Figure