Jaroslav Flegr á Pavel ZaÂboj á SÏteÏpaÂnka VanÏaÂcÏovaÂCorrelation between aerobic and anaerobic resistance
to metronidazole in trichomonads: application
of a new computer program for permutation tests
Received: 14 October 1997 / Accepted: 2 December 1997
Abstract An indispensable step of any comparative of the trait re¯ects simply a random process of evolu-
study is testing the concordance between the distribution tionary history of the taxon cladogenesis (Archie 1989).
of phenetic traits and the evolutionary history of the taxa
Several approaches are being used for testing of
under study. Here we present a computer program concordance between the distribution of the trait and the
TREEPT which can perform these tasks on the basis of a evolutionary history of a taxon, depending on the type
permutation test. The use of the program was demon- of data (character set/distance matrices) available for the
strated on trichomonad drug sensitivity data. The pro- studied trait. If the trait is described by character data, a
gram can also perform permutation tests analogous to cladistic analysis can be performed with forced tree to-
the t-test, analysis of variance (ANOVA) and correlation pology (re¯ecting the previously known cladogenesis of
analysis and is available at http://www.natur.cuni.cz/~ the taxon). The consistency index provided by common
¯egr/PROGRAMS/TREEPT.ZIP and http://www.karlin. cladistic programs can be used as a simple measure of
the degree of concordance between the distribution of
the trait and the phylogeny. The null model can be tested
by a permutation-tail probability test (Moore et al.
1994). If the trait is described by distance data, Mantel
tests can be used to test one or more hypotheses (inde-
pendent variables represented as matrices) against an
The testing of concordance between the distribution of a observed pattern (dependent matrix) using (partial) re-
particular trait and the evolutionary history of a taxon is gression or correlation (Thorpe 1996).
a principal task of many comparative studies (Harvey
The latter method is more universal because any
and Pagel 1991). The distribution of a trait among character data can be transformed to distance matrices.
representatives of the taxon can re¯ect either the distri- However, before analysis the distances should be cor-
bution of a common function and, therefore, of a rected for dierences in the rates of evolution in dierent
common selective pressure pattern (the species subjected branches of the phylogram. Moreover, no integrated
to the same selective pressure have the same trait) or the public-domain software exists for Mantel tests or for
evolutionary history of the taxon (the phylogenetically other important types of permutation tests.
relative species share the traits). The existence of a sta-
Recently we developed the program TREEPT for
tistically signi®cant association between the distribution various types of permutation tests, including those for
of the trait and the position of the species (or strains) analysis of concordance between the distribution of
within the genealogic tree indicates that the distribution traits and a phylogeny. The program can analyze the
qualitative and quantitative character data as well as the
distance matrices. The phylogenetic tree can be entered
in the usual bracketed format. The average distance
between sister OTUs (operational taxonomic units, i.e.,
Department of Algebra, Faculty of Mathematics and Physics,
sister strains or sister branches of the tree) is calculated
Charles University, Ke Karlovu 3, Prague 121 16, Czech Republic
(or read from a distance matrix) and used as a measure
of concordance, which is tested in a one-sided or two-
Correspondence to: Dr. J. Flegr, Department of Parasitology,
sided permutation-tail test (Manly 1991). The program
Faculty of Science, Charles University, VinicÏna 7, Prague 128 44,
can generate all possible permutations of terminal
branches of the tree or, alternatively, the number of trees
to be generated can be user-de®ned. Usually, 5000 ran-
Table 1 Susceptibility of Trichomonas vaginalis strains to metronidazolea
a Data represent geometric mean MLC values for metronidazole (in lg) as determined by in vitro microtiter plate assay (Tachezy et al.
1993) under aerobic and anaerobic conditions
bData obtained from MuÈller et al. (1988)
dom trees provide a stable estimation of the P value and The MLC values were obtained both under aerobic and
can be generated by an ordinary personal computer anaerobic conditions using a standard microplate assay
(Tachezy et al. 1993; Table 1). We used these data to test
The use of the program can be demonstrated on the correlation between aerobic and anaerobic MLC
trichomonad drug-susceptibility data. Some strains of values. The resistance data showed a non-normal dis-
parasitic protozoan Trichomonas vaginalis show either tribution. Therefore, we used a nonparametric Spear-
aerobic or anaerobic resistance to the main antitricho- man correlation test. The results [Spearman R 0.674,
monad drug metronidazole. The mechanisms of these t(9) 2.74] suggested that a signi®cant (P 0.022)
two types of resistance seem to be dierent (CÏerkasov- correlation existed between the aerobic and anaerobic
ova et al. 1988). However, the clinical isolates of resistance of the strain. The same results provided a
T. vaginalis isolated from patients refractory to treat- permutation test of correlation implemented in the
ment with metronidazole that have thus far been tested TREEPT program (P 0.01). Both tests work well
have consistently displayed the aerobic type of resis- under the condition that the data for dierent strains
tance. Therefore, routine susceptibility assays are per- represent independent observations. Because of the ex-
Preliminary istence of phylogenetic relationships between dierent
laboratory results suggest a possible correlation between strains, this condition can be violated. Therefore, the
the values for aerobic and anaerobic susceptibility possible concordance between the drug susceptibility
measured in in vitro tests (Lossick et al. 1986). Unpub- and the strain phylogeny should be tested ®rst. The
lished drug-susceptibility data available at our depart- phylogenetic tree generated for 11 strains of tricho-
ment include values for the minimal lethal concentration monads using DNA-®ngerprinting data by the neighbor-
(MLC) of metronidazole for 11 strains of T. vaginalis.
joining method (Saitou and Nei 1987) is shown in Fig. 1.
The concordance between the position of the strain
within the tree and the aerobic or anaerobic MLC value
was estimated by a permutation test, typing:
TREEPT ±n 5000 1X92 4X7 3X35 2X01 107X2 3X1251000 2X14 2X5 1X56 3X85
TREEPT ±n 5000 1X4 61X7 1X1 0X78 2X45 1X6716X0
respectively. The results suggest a signi®cant concor-
dance between aerobic (P 0.014) and anaerobic
(P 0.0023) drug resistance and the strain phylogeny.
Fig. 1 Phylogenetic tree created for 11 strains of Trichomonas
Therefore, the results of the correlation tests could be
vaginalis. The geographic origins of the strains are shown in
positively biased and could not be used with presently
parentheses. The numbers show the OTU-based jackkni®ng values
(in percent), which re¯ect the statistical support for the existence of
available data as a proof of correlation between the two
The program TREEPT can also perform the per- CÏerkasovova A, CÏerkasov J, Kulda J (1988) Resistance of tricho-
mutation tests in a manner analogous to that used for
monads to metronidazole. Acta Univ Carol Biol 30: 485±
the t-test, analysis of variance (ANOVA), and correla-
Harvey PH, Pagel MD (1991) The comparative method in evolu-
tion analysis. As shown by Adams and Anthony (1996),
tionary biology. Oxford University, Oxford
the permutation tests can be used for non-normally Lossick JG, MuÈller M, Gorrell TE (1986) In vitro drug suscepti-
distributed data and are generally more powerful than
bility and doses of metronidazole required for cure in cases of
analogous nonparametric tests. (The program TREEPT
refractory vaginal trichomoniasis. J Infect Dis 153: 948±955
is available at http://www.karlin.m.cuni.cz/$zaboj/ Manly BFJ (1991) Randomization and Monte Carlo methods in
Moore J, Freehling M, Gotelli NJ (1994) Altered behavior in two
species of blattid cockroaches infected with Moniliformis mon-
iliformis (Acanthocephala). J Parasitol 80: 220±223
Acknowledgements We wish to express our thanks to Dr J. Kulda
MuÈller M, Lossick JG, Gorrell TE (1988) In vitro susceptibility of
for providing drug-susceptibility data. This work was supported by
Trichomonas vaginalis to metronidazole and treatment outcome
grant 206/95/0638 from the Grant Agency of the Czech Republic.
in vaginal trichomonasis. Sex Transm Dis 15: 17±24
Saitou N, Nei M (1987) The neighbor-joining method: a new
method for reconstructing phylogenetic trees. Mol Biol Evol 4:
Tachezy J, Kulda J, Tomkova E (1993) Aerobic resistance of
Trichomonas vaginalis to metronidazole induced in vitro. Pa-
Adams DC, Anthony CD (1996) Using randomization techniques
to analyse behavioural data. Anim Behav 51: 733±738
Thorpe RS (1996) The use of DNA divergence to help determine
Archie JV (1989) A randomization test for phylogenetic informa-
the correlates of evolution of morphological characters. Evo-
tion in systematic data. Syst Zool 38: 239±252
CATARACT SURGERY PLEASE MARK ANY OF THE FOLLOWING THAT APPLY Is the Patient taking Flomax or Cardura? If yes, does the Patient have kidney disease? MEDICATIONS BEFORE SURGERY If you currently take anticoagulants or any other blood thinning medications, continue taking blood thinners and aspirin as prescribed by your primary care physician. If you are taking aspirin for a
Supplemental Checklist for: Pharmacies MONTEREY BAY AREA GREEN BUSINESS PROGRAM Supplemental Checklist: Pharmacies The following measures are intended to supplement those in the Minimum Measures and Office/Retailchecklist. In addition to completing the Minimum Measures and Office/Retail checklist, this checklistmust also be completed prior to certification. Remember, the program offers
Table 1 Susceptibility of Trichomonas vaginalis strains to metronidazolea
a Data represent geometric mean MLC values for metronidazole (in lg) as determined by in vitro microtiter plate assay (Tachezy et al.