DrugResistanceGenotypes

Drug resistance in success/failure models

With the simple success/failure models in which drug treatment either clears all infections in the host or does nothing (the treatSimple option in the health system / intervention component) drug resistance can be included by reducing the probability that treatment is effective. This does not allow for the effects of drug pressure on the level of resistance.

Multiple genotypes

With the options for modeling [explicit pharmacokinetics and pharmacodynamics] (ModelDrug), multiple drug resistance genotypes can be simulated (though OpenMalaria does not currently simulate mutation or genetic exchange so it cannot be used to model the origins of new resistance genotypes/phenotypes). Application this functionality requires that multiple genotypes should be declared (with each assigned initial frequencies in the parasite population and relative fitness in the absence of drug pressure); the corresponding phenotypes (i.e. pharmacodynamics) specified, and the outcomes monitored (as below):

Declaration of multiple parasite genotypes

Genotypes are specified in the parasiteGenetics element, which in turn is part of the drug element.

An example:

  <parasiteGenetics samplingMode="tracking">
    <locus name="mdr">
      <allele name="sensitive" initialFrequency="0.5" fitness="1.00"/>
      <allele name="resistant" initialFrequency="0.45" fitness="0.99"/>
      <allele name="highly resistant" initialFrequency="0.05" fitness="0.95"/>
    </locus>
  </parasiteGenetics>

The "initialFrequency" values must sum to one (if they don't, an error message is displayed requesting the sum of the initial probabilities of alleles equals 1). The "fitness" multiplies the within-host multiplication rate of the parasites, so that in the absence of drug pressure, genotypes with a high fitness outcompete the resistant genotypes (can the fitness of sensitives be >1)?

Assignment of drug resistance phenotypes

Each set of PD parameters is labelled a "phenotype". When there are multiple phenotypes, each phenotype must specify how genotypes are mapped to the phenotype via "restriction" elements, i.e. the "restriction" specifies to which allele of which locus the specified pharmacodynamics apply. The "restriction" elements are part of the "drug" specification:

An example:

<drug abbrev="CQ">
  <PD>
    <!-- three alleles match those described above -->
    <phenotype name="sensitive">
      <restriction onLocus="mdr" toAllele="sensitive"/>
      <max_killing_rate>3.45</max_killing_rate>
      <IC50>0.02</IC50>
      <slope>1.6</slope>
    </phenotype>
    <phenotype name="resistant">
      <restriction onLocus="mdr" toAllele="resistant"/>
      <max_killing_rate>3.45</max_killing_rate>
      <IC50>0.04</IC50>
      <slope>1.6</slope>
    </phenotype>
    <phenotype name="highly resistant">
      <restriction onLocus="mdr" toAllele="highly resistant"/>
      <max_killing_rate>3.45</max_killing_rate>
      <IC50>0.08</IC50>
      <slope>1.6</slope>
    </phenotype>
  </PD>
  <PK>
    <negligible_concentration>0.00036</negligible_concentration>
    <half_life>30.006</half_life>
    <vol_dist>300</vol_dist>
  </PK>
</drug>

Monitoring of the dynamics of drug resistance

To monitor the effects on parasite gene frequencies and parasite load, three different survey measures are available (in each case these are recorded by age-group):

id	measure	description
69	nInfectByGenotype	The number of human hosts with an infection (patent or not), for each genotype, at the time the survey is taken.
70	nPatentByGenotype	The number of human hosts whose total (blood-stage) parasite density, for each genotype, is above the detection threshold
71	logDensByGenotype	The number of hosts positive for each genotype and the sum of the logarithm of the corresponding parasite densities

Records in the output.txt file appear for each genotype, which is identified by an integer value (multiple of 1,000,000) added to the age group number (and to the number identifying the cohort, if present) in the 2nd column of each record. From version 34, to decode these genotype identifiers, OpenMalaria should be run from the command line with the --print-genotypes option. A decode table then appears on the screen. For example, if the XML specification of the parasite genetics is as follows:

  <parasiteGenetics samplingMode="tracking">
    <locus name="mdr">
      <allele name="sensitive" initialFrequency="0.5" fitness="1.00"/>
      <allele name="resistant" initialFrequency="0.5" fitness="1.00"/>
    </locus>
  </parasiteGenetics>

The decode information will read:

Genotype	mdr	init freq	fitness
0	sensitive	0.500	1.000
1000000	resistant	0.500	1.000

There will be corresponding lines in 'output.txt' for each of the specified measures 69-71, and each age-group, for each of these genotypes:

A more complex example with multiple loci:

  <parasiteGenetics samplingMode="tracking">
    <locus name="mdr">
      <allele name="sensitive" initialFrequency="0.5" fitness="1.00"/>
      <allele name="resistant" initialFrequency="0.45" fitness="0.99"/>
      <allele name="highly resistant" initialFrequency="0.05" fitness="0.95"/>
    </locus>
    <locus name="l2">
      <allele name="a1" initialFrequency="0.5" fitness="1.00"/>
      <allele name="a12" initialFrequency="0.45" fitness="0.99"/>
      <allele name="a123" initialFrequency="0.05" fitness="0.95"/>
    </locus>
  </parasiteGenetics>

The decode information will then read:

Genotype	mdr	l2	init freq	fitness
0	sensitive	a1	0.250	1.000
1	sensitive	a12	0.225	0.990
2	sensitive	a123	0.025	0.950
3	resistant	a1	0.225	0.990
4	resistant	a12	0.203	0.980
5	resistant	a123	0.023	0.941
6	highly resistant	a1	0.025	0.950
7	highly resistant	a12	0.023	0.941
8	highly resistant	a123	0.003	0.902