Modeling multimutation and drug resistance: analysis of some case studies
 Mitra Shojania Feizabadi^{1}Email author
DOI: 10.1186/s129760170052y
© The Author(s). 2017
Received: 2 April 2016
Accepted: 8 March 2017
Published: 21 March 2017
Abstract
Background
Druginduced resistance is one the major obstacles that may lead to therapeutic failure during cancer treatment. Different genetic alterations occur when tumor cells divide. Among new generations of tumor cells, some may express intrinsic resistance to a specific chemotherapeutic agent. Also, some tumor cells may carry a gene that can develop resistance induced by the therapeutic drug. The methods by which the therapeutic approaches need to be revised in the occurrence of drug induced resistance is still being explored. Previously, we introduced a model that expresses only intrinsic drug resistance in a conjoint normaltumor cell setting. The focus of this work is to expand our previously reported model to include terms that can express both intrinsic drug resistance and druginduced resistance. Additionally, we assess the response of the cell population as a function of time under different treatment strategies and discuss the outcomes.
Methods
The model introduced is expressed in the format of coupled differential equations which describe the growth pattern of the cells. The dynamic of the cell populations is simulated under different treatment cases. All computational simulations were executed using Mathematica v7.0.
Results
The outcome of the simulations clearly demonstrates that while some therapeutic strategies can overcome or control the intrinsic drug resistance, they may not be effective, and are even to some extent damaging, if the administered drug creates resistance by itself.
Conclusion
In the present study, the evolution of the cells in a conjoint setting, when the system expresses both intrinsic and induced resistance, is mathematically modeled. Followed by a set of computer simulations, the different growing patterns that can be created based on choices of therapy were examined. The model can still be improved by considering other factors including, but not limited to, the nature of the cancer growth, the level of toxicity that the body can tolerate, or the strength of the patient’s immune system.
Keywords
Cancer modeling Conjoint cell growth Chemotherapy Drug resistance MutationBackground
The mechanism under which different types of cancers progress is complex. This progression depends on many factors including, but not limited to, the growth rate of cells, mutual interaction of cancer cells with surrounding normal cells, the way in which they are affected by immune system, and their response to anticancer treatment strategies. It also depends upon the mutations that may happen during cell division, leading into the inapplicability of chemotherapeutic treatments.
The most common therapeutic approach to reduce the population of cancer cells and control their progression is chemotherapy. However, on many occasions, the success of this treatment is barred as a result of ineffectiveness of the drug used, which is known as drug resistance [1]. Some of the factors that can contribute to the creation of drug resistance are related to the drug delivery defects, insufficient drug activation at the tumor site, or the resistance that results from genetic mutation of the tumor cells [2, 3]. Given the importance of implementing the best therapeutic strategy, detecting any preexisting drug resistance, or any resistance induced by the drug during a course of chemotherapy, is important as it provides insight into the way in which the chemotherapeutic approach needs to be modified [4].
Cancer cells constantly divide into new generations of cancerous cells. Some of the newly born cells may contain mutated genes that express resistance to anticancer drugs, while the rest can still be susceptible to the therapy. This type of drug resistance is known as intrinsic resistance. A variety of approaches are being implemented to overcome the development of intrinsic drug resistance, including the uses of very high doses of chemotherapy, or utilizing combination therapies [5]. However, the success of chemotherapeutic treatments is less probable in cases when the implemented drug induces drug resistance. For example, some works show that in the case of drug induced resistance, it takes only few days for a tumor to regrow after the appearance of the resistance [6]. In some types of cancers, such as nonsmall cell lung cancer (NSCLC), a high percentage of patients develop drug resistance after longterm drug administration. In NSCLC, this kind of drug resistance is associated with a new set of mutations (T790M) created among wild cancerous tumor cells exposed to the specific chemotherapeutic agent. In such cancers, while the drugresponsive tumor cells shrink, the mutated tumor cells become resistant to the therapy; therefore, they grow and form a new tumor that no longer is responsive to the treatment [7].
An effective adjustment to a therapeutic approach is tied to a close monitoring of the evolution of untreated and treated normal and cancer cells, and the detection of any sign of resistance that may occur during the term of therapy. Modeling the evolution of the system has attracted more attention as it can provide insight into the progression of the diseases under a specific treatment; therefore, it is considered a parallel tool for tailoring the therapeutic approaches.
The various models expressed the growth of untreated normal and tumor cells with their possible interactions [8–12]. These models were then expanded to evaluate the dynamic of the system when different therapeutic approaches, including chemotherapy, virotherapy, radiation, and immunotherapy, were utilized. Also, some models have mathematically and numerically examined different cases in which some level of drug resistance exists in the system [13–23]. The evolution of different types of cancer cells in a multiresistance setting under chemotherapeutic treatment is not well considered in previous researches.
The aim of the present study is to construct a new model to improve biological reliability, and to enable the evaluation of the conditions in which the special chemotherapeutic drug can contribute to the treatment of cancer or cause more damage during the treatment by including drug resistance.
This paper focuses on simulating the evolution of different types of cancer cells in a multiresistance setting under chemotherapeutic treatment.
Methods
Conjoint core model in an intrinsic chemoresistance setting
where N(t), T(t), T_{R}(t) are respectively the total number of normal cells, drugresponsive tumor cells, and drugresistant tumor cells with the unit of cells. Also, K_{N}, K_{T}, K_{R} are the carrying capacity of normal cells and two types of tumor cells with the unit of cells. The per capita growth rate for the drugresponsive tumor cells, drugresistant tumor cells, and normal cells are expressed by r_{T,} r_{R}, r_{N} with the unit of time^{−1}. The T* is the critical size of the collection of tumor cells with the unit of cells. The second term in equation 1c represents the interaction between tumor and normal cells. In this term κ has the units of time^{−1}. The drugresponsive tumor cells become intrinsically resistant tumor cells with a mutation rate of τ (time^{−1}). The last term in eq. 1a and 1c represents the interaction of normal and drugresponsive tumor cells with chemotherapeutic drugs. These cells die due to drug toxicity. The response function to the chemotherapeutic drug can be structured as a_{i}(1e^{MC}), where M is associated to the drug pharmacokinetics and known as the drug efficiency coefficient with the unit of m^{2}.mg^{−1}, and C represents the amount of the drug at the tumor site (mg. m^{−2}). The coefficient a_{i} when i = N, T with the unit of time^{−1} expresses the rate of chemotherapyinduced death [18, 24].
To achieve a more complete picture of the evolution of the cells in a drug resistance setting, the current model is modified below to include those types of drug resistance created as a result of interaction with chemotherapeutic agents.
Conjoint core model and druginduced resistance and simulations
Extended model
A group of tumor cells with specific mutated genes may develop resistance to chemotherapeutic agents as they interact with the drug. To introduce this type of the drug resistance in our model, three groups of tumor cells were considered. As explained before, the first group are tumor cells that are responsive to the drug and grow under the logistic law, and their population decreases as they interact with the drug. The drugsensitive tumor cells create a new generation of tumor cells as they divide. We assume that the newly born tumor cells can be placed in one of the following three groups. The first group includes those that are still responsive to the administered drug, and are known as wild tumor cells, T. The second group is those tumor cells that are still responsive to the drug, but carry a mutated gene that causes drug resistance as they interact with the introduced drug. These tumor cells are placed in the category of mutated tumor cells, T_{M}. The third group of tumor cells is those that are not responsive to the drug and intrinsically resist the administered drug. This group is identified by T_{R}. All of these tumor cells are assumed to grow under the logistic law. The term τ_{1}T(t) in equations 1a and 1b expresses the transition of wild tumor cells to resistant tumor cells. The newly introduced term τ_{2}T(t) in equations 2a and 2c represents the transition of wild tumor cells to mutated tumor cells. Also, the toxic effect of the administered drug, which leads to the reduction in populations of cells, has been expressed by a_{T}(1e^{MC})T on wild tumor cells as well [24]. The interaction of the drug with the mutated tumor cells partially kills them and partially turns them into drugresistant tumor cells. The toxic effect of the drug which leads to the reduction of the population of mutated tumor cells has been expressed as a_{TM} (1e^{MC})T_{M}, where a_{TM} is the killing rate of mutated tumor cells induced by the first administered drug. Also, we considered that the mutated tumor cells also follow the logistic growth. Therefore, K _{ M } and r _{ M } are the carrying capacity of the mutated tumor cells and per capita growth rate of this group of tumor cells, respectively. The term that expresses the conversion of mutated tumor cells to drugresistant in equations 2b and 2c has been expressed by τ_{M → R}(1 − e^{− MC})T_{M}. In this term τ_{M → R} with the unit of time^{−1} expresses the conversion rate of mutated tumor cells to resistant tumor cells due to interaction with the drug.
Furthermore, to evaluate cases that undergo combination therapy, a second chemotherapeutic drug can be added to the treatment, in the way that this second drug can be effective on drugresistant tumor cells and can be mathematically introduced as \( {\mathrm{a}}_{\mathrm{T}\mathrm{R}}\left(1{\mathrm{e}}^{{\mathrm{MC}}_2}\right){\mathrm{T}}_{\mathrm{R}} \) (equation 2b). In the following equations, a_{T}, and a _{ TM }, are the death rate induced by the first administered chemotherapeutic drug, while a_{TR} is the death rate of the tumor resistance cells induced by this second drug. In addition, the concentration of the second drug is introduced by C_{2}.
Numerical simulations and choice of parameters
Parameters used in simulations in different therapeutic cases introduced above
Resistance Detected  Growth Parameters  Specifications of Drug I  τ _{2}  τ _{ M → R }  τ _{1}  Specifications of Drug II 

DrugInduced  K_{T} = K_{N} = K_{R} = 10^{6} (cells) r_{T} = r_{r} = r_{M} = 0.25 r_{n} = 0.5 day^{−1} κ = 0124 day^{−1} T^{*} = 3*10^{5} cells  Constant Drug C = 0.2 (mg.m^{2}) a _{T} = 0.15 (day^{−1}) t(start) = 50 days  10^{−3}/day Mutation starts at t = 0 day  10^{−4}/day Conversion starts at t = 50 days  0 (No intrinsic resistance)  0 (No combination therapy) 
DrugInduced  Same as above  Constant Drug C = 0.2 (mg.m^{2}) a_{T} = 0.15 (day^{−1}) t(start) = 50 days  10^{−3}/day Mutation starts at t = 0 day  10^{−4}/day Conversion starts at t = 150 days  0 (No intrinsic resistance)  0 (No combination therapy) 
DrugInduced  Same as above  Decaying Drug C = 0.2exp(0.001 t) (mg.m^{2}) a_{T} = 0.15 (day^{−1}) t(start) = 50 days  10^{−3}/day Mutation starts at t = 0 day  10^{−4}/day Conversion starts at t = 50 days  0 (No intrinsic resistance)  0 (No combination therapy) 
DrugInduced and Intrinsic  Same as above  Decaying Drug C = 0.2exp(0.001 t) (mg.m^{2}) a_{T} = 0.15 (day^{−1}) t(start) = 50 days  10^{−3}/day Mutation starts at t = 0 day  10^{−4}/day Conversion starts at t = 50 days  10^{−4}/day  C_{2} = 0.6 (mg.m^{2}) a_{TR} = 0.15(day1) t(start) = 50 days 
The set of equations introduced above (2a2d) shows the population dynamic of four variables (T, T_{R}, T_{M}, N). The terms of these equations and associated parameters describe the growth of each introduced population of the cells, or the way that they are affected as a result of a) the existing dependency among populations, b) transition among populations, and c) the interaction with the drug.
The values of the parameters associated to the growth of tumor cells are tumorspecific. Also, previously reported measurements show that the obtained values from participants in clinical trials may also be different from those obtained from in vivo experiments [24]. The current work has no concentration on a specific type of tumor. However, the values of the parameters we have chosen are in the range with those reported by other studies. For parameters with no specific reported values, ad hoc values have been chosen. The goal is to analyze the behavior of the system under these specifically chosen cases. Below, the choice of parameters has been explained with further details.
Two parameters that describe the growth are the growth rate and the carrying capacity. Other studies have reported a value of less than one for the growth rate, and a carrying capacity between 10^{5} cells and 10^{9} cells [15, 25]. However, it should be noted that the value of the carrying capacity is organsspecific [26]. To be in this range, the adapted values in this study are K_{M} = K_{T} = K_{N} = K_{r} = 10^{6} (cells), r_{T} = r_{R} = r_{M} =0.25 (day^{−1}), r_{N} = 0.5 (day^{−1}). To choose the value of T*, we referred to the reported study of Demicheli et al. [27], which explains that more information is available on the initial and last stages of the tumor growth than on tumor growth in the intermediate phase. The study explains that the growth pattern in the intermediate phase is very complex and tumor specific. The critical tumor cell size of human colon carcinoma cell line "LoVo," where the pattern of growth deviated from the Gomertizian, was measured to be in the intermediate phase of growth when the tumor cell population was approximately 10^{3} cells [27]. Relying on the findings of this study, and since the carrying capacity (the last phase of growth) in the current work is set to be 10^{6} cell, T* has been chosen to be 10^{5} cells, to be placed in the intermediate growth phase.
The parameters related to the chemotherapeutic agents are: the death rate of the tumor cells induced by the drug, the drug concentration and, accordingly, toxicity coefficient, and the drug pharmacokinetics parameters. The induced death rate for drug sensitive tumor cells a_{TR} and a _{ T } (0.15 day^{−1}) are considered to be equal, and the order of the value of these parameters is consistent with those reported by previous studies [15]. Also, the two parameters are associated to the chemotherapeutic agent: "M", the pharmacokinetics parameters with a value of 1, and C, the drug concentration or toxicity coefficient [24]. We first evaluated the cases when the drug concentration remains constant (C = 0.2 mg.m^{−2}) in the tumor site. Then, the dynamic of the cell populations was evaluated in cases in which the amount of the chemotherapeutic agent went under decay in the tumor site. This decay was considered to be exponential with the drug decaying rate of 10^{−3} day^{−1}. The value for the decaying parameter is an ad hoc value and is case based.
The value of the parameters related to the transition of the cells from one subpopulation to another are also ad hoc values. These transitions include those from wild tumor cells to resistant tumor cells; wild tumor cells to mutated tumor cells that can be converted to resistant tumor cells as they interact with the drug; and mutated tumor cells to resistant tumor cells. The method of measurement of the rates of these mutations and transitions is being explored. Therefore, the evaluation of the dynamic of the system is limited to the case study with some ad hoc values. The selected values for these parameters are reflected in Table 1.
Under this general method of choosing parameters, different values are considered for the parameters that are linked to the therapeutic approaches. The outcomes of simulations under these choices are discussed below.
Results and Discussion
Numerical simulations under different therapeutic approaches
Constant drug and druginduced resistance
In our work, a Gompertzian model is considered for the growth of cells with the chosen value for the carrying capacity equal to10^{6} cells. Therefore, the critical point of the growth, when the cells enter a slower phase of growth, is around 2*10^{5} cells. This critical point is considered as a point when tumor population becomes clinically detectable [28]. In such a case, around t = 375 days, their populations can be detectable. However, they immediately grow and become the dominant population at the end of the simulation time, t = 500 days. Figure 2b evaluates the clinically observed cases when the druginduced resistant tumor cells mutate as the wild tumor cells interact with the administered drug for a long time. Distinct from Fig. 2a, in this simulation, it is assumed that the transformation of the mutated tumor cells to resistant cells starts at the later time of t = 150 days. The shrinkage of drugresponsive tumor cells can be seen. In this case, the population of drugresistant tumor cells is not yet dominant at the end of the simulation.
Decaying drug and druginduced resistance
In Fig. 2c, the therapy with anticancer drug starts at t = 50 days. However, the amount of the drug will not stay constant and decreases exponentially over time, with a decay factor of 10^{−3} day^{−1}. At t = 450 days, the drugresistant tumor cells become detectable, and at the end of the simulation time, they are not dominant tumor cells. The population of drugresponsive tumor cells is still in the declining phase as they interact with the drug.
In the current case, as the simulation shows, some wild and mutated tumor cells still exist in the system. As they are responsive to the drug, the treatment can be continued. The drugresistant tumor cells were created with a delay, as compared with the previous case when the amount of the drug was constant, and the growth of drugresistant tumor cells occurred earlier.
It should be noted that in some cases, detecting drug resistance can be an advantage, as the therapeutic approach can be altered accordingly. By switching to another drug or by utilizing a combination therapy, there can be a path to a more successful treatment. This, certainly, is a therapeutic choice, depending on many health factors of the given patient.
Intrinsic and druginduced resistance
The simulation 2 d evaluates the case when the system expresses an intrinsic drug resistance, in addition to an induced drug resistance.
One of the therapeutic approaches to controlling intrinsic drug resistance is the use of a higher dosage of a drug in early stages of the cancer progression. Under the simulation conditions, it is assumed that two types of drugs are implemented, one with a higher dosage that is toxic to drugresistant cells. The amount of this drug is considered to stay constant over time. The second drug is toxic to tumorresponsive cells. The amount of this drug decays exponentially over time. The result of this simulation shows that this approach is effective in overcoming drugresistant tumor cells. By the end of the simulation time, they die out of the system and the remaining tumor cells are responsive to the drug. In such a case, tailoring the therapy to achieve the best outcome is again a factor of concern. As there are still some drugresponsive tumor cells in the system, the treatment with the second drug can be continued. However, since the drugresistant tumor cells have died out of the system, the treatment with the first drug can be terminated. The continuation of the treatment with the second drug, which is effective on drug responsive cells, raises the possibility of the creation of another drugresistant tumor cell. It is possible that periodic treatments could be a more successful therapeutic approach to control both the progression of cancer as well as the existing resistance.
Conclusions
Cancer drug resistance, which is an obstacle to successful treatment outcomes, is not limited to intrinsic resistance. The success of the treatment becomes more unpredictable if the introduced drug induced some resistance. The utilization of different chemotherapeutic drugs, combination therapy, and periodic therapy are some protocols that are currently implemented. The use of modeling and computer simulations enhance our understanding of the evolution patterns that may occur during treatments. In the present study, the evolution of the cells in a conjoint setting, when the system expresses both intrinsic and induced resistance, is mathematically modeled. Followed by a set of computer simulations, the different growing patterns that can be created based on choices of therapy were examined. The model can still be improved by considering the nature of the cancer growth: for example, it would be more realistic to include such things as the blood supply, the three dimensionality of the growth, and other important biological variables.
Some open concerns include whether the mutations happen at a constant rate or if the rate can be affected by the drug and other surrounding conditions. Furthermore, in all of these cases, the level of toxicity that the body can tolerate, along with other factors such the strength of the patient’s immune system, play important roles in making decisions with regards to impactful therapy.
Abbreviations
 NSCLC:

Nonsmall cell lung cancer
Declarations
Acknowledgement
None
Funding
Not Applicable (None)
Availability of data and materials
The computer codes/datasets during and/or analyzed during the current study available from the corresponding author on reasonable request.
Competing interests
The author declares that she has no competing interests.
Consent for publication
Not Applicable
Ethical approval and consent to participate
Not Applicable
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Authors’ Affiliations
References
 Holohan C, Van Schaeybroeck S, Longley DB, Johnston PG. Cancer drug resistance: an evolving paradigm. Nat Rev Cancer. 2013;13:714–26.View ArticlePubMedGoogle Scholar
 Gottesman MM. Mechanisms of cancer drug resistance. Annu Rev Med. 2002;53:615–27.View ArticlePubMedGoogle Scholar
 Goldie JH, Coldman AJ: Extensions of the random mutation model of drug resistance. In: Goldie JH, Coldmaan AJ (eds.) Drug resistance in cancer: mechanisms and models. New York: Cambridge University Press; 1998:148194.
 Foo J, Franziska M. Evolution of acquired resistance to anticancer therapy. J Theor Biol. 2014;355:10–20.View ArticlePubMedGoogle Scholar
 Giaccone G, Ponedo HM. Drug resistance. Oncologist. 1996;1(1&2):82–7.PubMedGoogle Scholar
 Menchón SA. The effect of intrinsic and acquired resistances on chemotherapy effectiveness. Acta Biotheor. 2015;63(2):113–27.View ArticlePubMedGoogle Scholar
 Juchum M, Gunther M, Laufer SA. Fighting cancer drug resistance: Opportunities and challenges for mutationspecific EGFR inhibitors. Drug Resist Updat. 2015;20:12–28.View ArticlePubMedGoogle Scholar
 Witten TM. Modeling cellular aging and tumorigenic transformation. Math Comput Simul. 1982;24:572–84.View ArticleGoogle Scholar
 Witten TM. Population models of cellular aging: Theoretical and numerical issues. In: Vichnevetsky R, Stepleman RS, editors. Advances in computer methods for partial differential equations VI: proceesings of the sixth IMACS international symposium on Computer methods for partial differential equations. New Brunswick. New Jersey: IMACS, Department of Computer Science, Rutgers University; 1987.Google Scholar
 Feizabadi MS, Volk C, Hirschbeck S. A twocompartment model interacting with dynamic drugs. Appl Math Lett. 2009;22:1205–9.View ArticleGoogle Scholar
 Feizabadi MS, Carbonara J. Twocompartment model interacting with proliferating regulatory factor. Appl Math Lett. 2010;23:30–3.View ArticleGoogle Scholar
 Feizabadi MS, Witten TM. Chemotherapy in cojoint agingtumor systems: some simple models for addressing coupled agingcancer dynamics. Theor Biol Med Model. 2010;7:21.View ArticlePubMedPubMed CentralGoogle Scholar
 Bajzer B, Carr T, Josic K, Russell SJ, Dingli D. Modeling of cancer virotherapy with recombinant measles viruses. J Theor Biol. 2008;252:109–22.View ArticlePubMedGoogle Scholar
 Dingli D, Cascino MD, Josic K, Russell SJ, Bajzer Z. Mathematical modeling of cancer radiovirotherapy. Math Biosci. 2006;199:80–103.View ArticleGoogle Scholar
 de Pillis LG, Gu W, Radunskaya AE. Mixed immunotherapy and chemotherapy of tumors: modeling, applications and biological interpretations. J Theor Biol. 2006;238:841–62.View ArticlePubMedGoogle Scholar
 Kirschner D, Panetta JC. Modeling immunotherapy of the tumorimmune interaction. J Math Biol. 1998;37(3):235–52.View ArticlePubMedGoogle Scholar
 Wu JT, Kirn DH, Wein LM. Analysis of a threeway race between tumor growth, a replicationcompetent virus and an immune response. Bull Math Biol. 2004;66:605–625.
 Sameen S, Barbuti B, Milazzo P, Cerone A, DelRe M, Danesi R. Mathematical modeling of drug resistance due to KRAS mutation in colorectal cancer. J Theor Biol. 2016;389:263–73.View ArticlePubMedGoogle Scholar
 Tomasetti C, Levy D. An elementary approach to modeling drug resistance in cancer. Math Biosci Eng. 2010;7(4):905–18.View ArticlePubMedPubMed CentralGoogle Scholar
 Lavi O, Gottesman MM, Levy D. The dynamics of drug resistance: A mathematical perspective. Drug Resist Updat. 2012;15:90–7.View ArticlePubMedPubMed CentralGoogle Scholar
 Panetta JC. A mathematical model of drug resistance: heterogeneous tumors. Math Biosci. 1998;147:41–61.View ArticlePubMedGoogle Scholar
 Feizabadi MS, Witten TM. Modeling the effects of a simple immune system and immunodeficiency on the dynamics of conjointly growing tumor and normal cells. Int J Biol Sci. 2011;7(6):700–7.View ArticlePubMedPubMed CentralGoogle Scholar
 Feizabadi MS, Witten TM. Modeling drug resistance in a conjoint normaltumor setting. Theor Biol Med Model. 2015;12:3.View ArticlePubMedPubMed CentralGoogle Scholar
 de Pillis LG, Radunskaya A. The dynamics of an optimally controlled tumor model: A case study. Math Comput Model. 2003;37:1221–44.View ArticleGoogle Scholar
 de Pillis LG, Savage H, Radunskaya AE. Mathematical model of colorectal cancer with monoclonal antibody treatments. Br J Med Med Res. 2014;4(16):3101–31.View ArticleGoogle Scholar
 Kuang Y, Nagy JD, Elser JJ. Biological stoichiometry of tumor dynamics: Mathematical models and analysis. Discrete Contin Dyn Syst Ser B. 2004;4(1):221–40.Google Scholar
 Demicheli R, Foroni R, Ingrosso A, Pratesi G, Soranzo C, Tortoreto M. An exponentialGompertz description of LoVo cell tumor growth from in vivo and in vitro data. Cancer Res. 1989;49(23):6543–6.PubMedGoogle Scholar
 Talley J, Frankum B, Currow D. Essentials of Internal Medicine. 3rd ed. London: Elsevier; 2015.