K. Waisser et al. / Eur. J. Med. Chem. 35 (2000) 733–741
737
In the case of the selectivity of the sth effect, the ‘ideal’
drug is represented by the unit vector Us, the components
of which are given by:
The norm of activity vector g(A) is a measure of the
overall antimycobacterial potency of the compound. Com-
plex criterion S0 is proposed for a comparison of the
evaluated compound with the ideal broad-spectrum drug,
i.e. compound possessing equal MICs against all strains
taken into consideration. Cosine coefficient k0 is a mea-
sure of relative similarity to the ideal drug as the
influence of the overall potency is eliminated. This
aspect, of course, is of greater importance in the search
for selectively acting compounds. The more the cosine
coefficient is close to the value of 1, the more the profile
of evaluated compound relatively agrees with that of the
ideal drug. As follows from table V, the values of cosine
coefficient k0 (0.963–1.000) justify the conclusion that
the compounds represent broad-spectrum antimycobacte-
rial agents. Detailed inspection of table V reveals that, in
general, higher values of S0 are found in compounds 3
and 4, while values of k0 closer to 1 are found in 2. This
is consistent with the fact that the replacement of one or
both oxo groups by thioxo groups yielded the compounds
3 and 4 with greater activity against M. tuberculosis and
M. kansasii but with lesser activity against M. avium than
that of the compounds 2.
for i = s,
͌
n − 1 / n n − 1
͑
͒
͑
͒
Usi =
Ά
for i 7 s, i = 1,
, n.
͌
− 1/ n n − 1
͑
͒
(5)
It is noteworthy that all but the sth effect are considered
to be undesired and the negative sign is attributed to
them. The formula for calculating the complex criteria Ss
(complex selectivities) can be written as:
n
1
Ss =
nA −
A ,
i
ͩ ͪ
s
͚
i=1
͌
n n − 1
͑
͒
where s = 1,
, n.
(6)
Quantitative structure–activity relationships were analy-
sed by the Free-Wilson method [14] in the Fujita-Ban
modification [15].
For the n-component vector of regression coefficients
a, interpreted as contributions of respective fragment to
the analysed activities, analogous measures, i.e. complex
criterion s, the norm of the vector g(a) and cosine
coefficient k(a, U), can be introduced:
For quantitative analysis of the structure–antimycobac-
terial activity relationships we used the method of Free-
Wilson [14] as modified by Fujita-Ban [15]. In order to
describe the influence of replacement of one or both oxo
groups by thioxo group(s) and substitution on the phenyl
ring, the unsubstituted on the phenyl ring dione 2a was
chosen as the reference compound for all the analyses.
Since no data for 2-thioxo-1,3-benzoxazine-4(3H)-ones
were available, sulfur atoms occur six times at position 2
and 12 times at position 4 of the basis set. Table VI shows
the derived correlation equations.
s = a, U = g a k a, U .
(7)
͑
͒
͑ ͒ ͑
͒
5. Results and discussion
The antimycobacterial activities of compounds 2, 3, 4
and isoniazid against Mycobacterium tuberculosis My
331/88, M. avium My 330/88, M. kansasii My 235/80 and
M. kansasii 6509/96 (clinical isolate resistant to iso-
niazid) are shown in table IV. In general, the synthesized
compounds possess in vitro activities against all myco-
bacterial strains tested, better than or comparable to that
of isoniazid. We cannot compare the data presented here
with the previously published results of testing of com-
pounds 2a–e against M. tuberculosis H37Rv and M. kan-
sasii PKG 8 [9] as they were obtained for different
strains.
Evaluation of broad-spectrum antimycobacterial pro-
files of individual compounds was performed by the
procedure described previously [16]. Activities, expressed
as log (1/MIC), for strains of Mycobacterium tuberculo-
sis, M. kansasii and M. avium obtained from the CNCTC
were used for the calculation. The clinical isolate of
M. kansasii was omitted from these calculations as it
exhibits the same (to compounds 2) or even better
susceptibility (to compounds 3 and 4) than the strain of
M. kansasii from the CNCTC. The results of calculations
are summarized in table V.
These equations confirm that each fragment has a
constant, independent and additive contribution to the
activity of the whole molecule against M. tuberculosis
My 331/88, clinical isolate of M. kansassii 6509/96 and
M. kansasii My 235/80, in the last case at least for data
after 14 days. Analyses of data for M. avium My 330/88
and M. kansasii My 235/80 after 21 days resulted in
non-significant correlations.
In all cases of the significant correlations, the only
significant positive contributions were those of the sulfur
atom in position 4. It is also observed that the contribu-
tions of substituents on the phenyl ring are non-significant
in comparison to the unsubstituted phenyl, except for
M. tuberculosis data after 14 days. The contribution of
the sulfur atom in position 2 is also non-significant in
comparison to the oxygen atom. The antimycobacterial
activity of 3-phenyl-2-thioxo-1,3-benzoxazine-4(3H)-ones