Tumor Growth Instability and Its Implications for Chemotherapy

Bibliographic Reference

Castorina, P., Carcò, D., Guiot, C., & Deisboeck, T. S. (2009). Tumor growth instability and its implications for chemotherapy. Cancer Research, 69(21), 8507–8515. https://doi.org/10.1158/0008-5472.CAN-09-0653

Core Argument

Tumor clonal heterogeneity — modeled here as a two-population system in which a small, faster-replicating subclone emerges from an initially monoclonal tumor — produces growth instability that requires modification of the standard Norton-Simon late intensity chemotherapy schedule. The required drug dose depends strongly on the balance of the two populations and on their distinct growth rates, not merely on total tumor size.

Methods

Mathematical modeling of two-population Gompertzian growth dynamics. The authors model a tumor where, at a delay time t_delta, a secondary cell population N2 emerges that proliferates at a distinct (faster) rate than the parent population N1. Both populations follow the Gompertz law (GL), with different asymptotic saturation values (N1_infinity != N2_infinity) producing different specific growth rates. Three scenarios are analyzed: (1) instability onset before the start of chemotherapy, (2) instability onset after the start of chemotherapy, and (3) differential drug effectiveness against the two clones (different regression rates omega1 != omega2). Model predictions for survival probability are compared with published clinical data on ER+ versus ER- breast cancer patients from Colleoni et al. (2008).

Key Findings

• In the two-population Gompertzian model, the emergence of a faster-replicating subclone modifies the linear Norton-Simon late intensity drug schedule: both the slope and the initial dose depend on the ratio of the two populations and their specific growth rates. Greater differences between subclone growth rates produce larger deviations from the single-population schedule.
• If instability precedes treatment onset, the required drug concentration C(t) differs substantially from the single-population calculation. The ratio D = C/C_g increases by 20% to 70% depending on the ratio of the growth-rate parameters K_g1 and K_g2.
• If instability follows treatment onset, the drug concentration ratio D_II deviates from the value of 1 that holds under the single-population assumption, but changes by less than 10% for the tested parameter set.
• When the two subclones respond differently to the same chemotherapeutic agent, no pair of constant regression rates (omega1, omega2) can simultaneously satisfy the equations for both populations — unless the populations are identical. A time-dependent reduction rate omega2(t) is mathematically required.
• The model predicted differential survival probabilities for ER+ versus ER- breast cancer patients that matched clinical data from Colleoni et al. (2008), with fitted parameters alpha1 approximately 0.071/year and beta approximately 0.08/year (linear survival model) or beta approximately 0.13/year (exponential survival model).

Concepts Introduced or Used

• Tumor growth instability: A small tumor cell fraction N2 emerging at time t_delta that exhibits a higher proliferation rate than the parent strain N1. Distinct from genomic instability — this is a kinetic/phenotypic instability arising from clonal diversity.
• clonal-evolution — the emergence of a faster-growing subclone from a parent population represents a clonal evolutionary transition
• clonal-expansion — N2 expands from a small fraction to sometimes dominate the final tumor mass
• intratumor-heterogeneity — explicitly discussed as “cancer’s inherent clonal diversity” and “clonal heterogeneity” that results in nonhomogeneous growth patterns
• therapy-resistance — differential drug responsiveness between subclones (omega1 != omega2) drives the need for schedule modification
• Gompertz law (GL): logarithmic specific growth rate, alpha(t) = K_g * ln(N_infinity / N), used as the underlying growth model for both populations
• Norton-Simon late intensity schedule: the standard regimen where drug concentration increases linearly with time, C(t) = A + B(t - t*), derived from the requirement that tumor cell number decreases exponentially with a constant rate
• Norton-Simon hypothesis: therapy results in a rate of regression proportional to the growth rate of an unperturbed neoplasm of that size

Entities Referenced

• Cancer types: breast cancer (primary and metastatic), specifically ER+ (ER/PgR present) and ER- (ER/PgR absent) subtypes
• Genes/receptors: estrogen receptor (ER), progesterone receptor (PgR)
• Clinical trials referenced: CALGB 97-41 (dose-dense chemotherapy trial led by Cancer and Leukemia Group B)
• Growth models: Gompertz law (Gompertzian growth), Universal growth law (West et al., 2001)
• Therapeutic approaches: Norton-Simon late intensity schedule, dose-dense chemotherapy, cycle-nonspecific chemotherapy
• Clinical data source: Colleoni et al. (2008), Annals of Oncology, on preoperative chemotherapy response by receptor status

Limitations (as stated by authors)

• The model is simplified to two-population dynamics; realistic tumors are polyclonal and the method would need extension to n populations.
• The analysis assumes constant omega1 for simplicity when computing omega2(t); full optimization of parameters was not performed.
• The survival probability analysis uses a trial-and-error estimation of beta rather than rigorous fitting.
• The survival probability model does not reproduce the saturation of patient survival probability observed at long time intervals in clinical data.
• The model assumes both subpopulations follow the Gompertz law with the same functional form and the same K_g parameter in the simplest case.
• The approach does not account for interspecific competition between subclones — competition is encoded only indirectly through the ratio K_g1/K_g2 and boundary conditions.

Relevance to Clonal Evolution

This paper makes the case that even a single clonal branching event — the emergence of one faster-growing subclone — changes optimal chemotherapy dosing in a mathematically nontrivial way. It connects intratumor-heterogeneity (clonal diversity) directly to a concrete therapeutic consequence (modified drug scheduling), providing a quantitative bridge between clonal-evolution dynamics and treatment strategy that the standard Norton-Simon framework, with its assumption of single-population Gompertzian growth, cannot capture.