Sensors and Actuators B 193 (2014) 715–722

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

journal homepage: www.elsevier.com/locate/snb

Novel phage-piezoelectric sensor for rapid drug susceptibility testing

of Mycobacterium tuberculosis
Xianwen Mi a,b , Fengjiao He a,∗ , Meiyu Xiang a , Yan Lian a , Songlin Yi c
a
State Key Laboratory of Chemo/Biosensing and Chemometrics, College of Chemistry and Chemical Engineering, Hunan University, Changsha 410082, China
b
Huaihua Medical College, Huaihua 418000, Hunan, China
c
Hunan Institute of Tuberculosis Control, Changsha 410006, China

a r t i c l e i n f o a b s t r a c t

Article history: Novel phage-piezoelectric sensor was constructed for rapid drug susceptibility testing based on the pro-
Received 4 October 2013 posed method of phage amplified multichannel series piezoelectric quartz crystal sensor (PA-MSPQC) for
Received in revised form the detection of Mycobacterium tuberculosis and criterion of resistant strains by CLSI. Under the condition
19 November 2013
of anti-TB drugs, the resistant M. tuberculosis survived, but the sensitive strains can be killed. According to
Accepted 20 November 2013
Available online 11 December 2013
the criterion of resistant strains by CLSI, the 105 cfu of strains with frequency shift less than 121 Hz were
resistant to rifampin, isoniazid and ethambutol, 92 Hz to streptomycin. The sensitivity and specificity of
557 strains drug susceptibility test by proposed methods are 91% and 93%, respectively, compared with
Keywords:
Mycobacterium tuberculosis the mycobacteria growth indicator tube (MGIT) method. But the turnaround time of proposed method is
Mycobacterium smegmatis no more than 35 h, far less than that of MGIT 960 methods. It will find application in clinical test.
Phage D29 © 2013 Elsevier B.V. All rights reserved.
Drug susceptibility testing

1. Introduction polymorphism (PCR-SSCP) technology [6,7], LCD array technology

with DNA sequencing method [8] and molecular beacons meth-
The reasons for the pandemic emergency of TB are AIDS ods [9,10]. These methods were rapid and sensitive [11–15]. They
epidemic and the appearance of multi-resistance strains. WHO are specially suits the susceptibility testing of mycobacteria with
recommended the directly observed therapy, short-course strat- slow growth rate. However, their molecular mechanisms in anti-
egy (DOTS) measurement for controlling TB in the whole world. mycobacterial agents have not been elucidated and it is difficult to
It is critical to DOTS strategy that the Mycobacterium tuberculosis design a rapid diagnostic test to detect all resistant tubercle bacilli,
(M. tuberculosis) are rapidly detected and patients are rationally and then to develop more effective therapeutics method for treat-
administrated based on drug susceptibility testing [1]. ing multidrug-resistant M. tuberculosis [16–18]. The sensitivity and
So far, there are two main susceptibility testing methods of specificity of the molecular biology method are affected by the state
M. tuberculosis, including molecular biology method and pheno- of sample [7], and inhibitors and the amount of bacteria. These
type analysis method. The principle of the detection method for will easily result in false-negative results. Meanwhile, non-specific
drug susceptibility testing based on molecular biology is that: the amplification products of polluting bacterial DNA decreased detec-
encoding gene mutation of the target molecule aimed by the drug tion specificity.
will leads to the lost sites of drug action, and resulted in resistant The drug-resistant detection method of M. tuberculosis based
stains to drugs. The resistant strains can be detected by deter- on phenotypes analysis include agar proportion method, mini-
mining the presence of drug-resistance mutation with molecular mum inhibitory concentration (MIC) method [19,20], automated
biology method. The molecular biology method for drug sus- mycobacterial culture systems and phage culture method. Most of
ceptibility testing includes polymerase chain reaction–restriction their detection time were limited by generation time of M. tuber-
fragment length polymorphism (PCR-RFLP) method [2,3], nested culosis(18 h). Among them, agar proportion method is prescript
PCR amplification [4], denaturing high-performance liquid chro- standard method for susceptibility of M. tuberculosis by CLSI, but
matography (DHPLC) method [5], PCR-single strand conformation it is time-consuming (28 days), and complicated to operation, and
the test results are easily affected by artifacts. The MIC method is
based on the lowest concentration of an antimicrobial to inhibit the
∗ Corresponding author. Tel.: +86 137 87799232/+86 731 88272269; visible growth of a microorganism (28 days). Automated mycobac-
fax: +86 731 88808630/+86 731 88055818. terial culture systems such as BACTEC 460 TB, BACTEC MGIT 960
E-mail address: [email protected] (F. He). and ESP, showed good sensitivity and shortened the detection time

0925-4005/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.snb.2013.11.062
716 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722

to 10 days for detection of M. tuberculosis in clinical specimens [21]. 108 cfu/mL of M. smegmatis suspension was made for use according
This cannot meet the clinical requirement for diagnosis. Phage cul- to McFarland’s turbidity.
ture method was based on the fact that the phage D29 can specifi-
cally infect the viable M. tuberculosis and Mycobacterium smegmatis, 2.1.2. Preparation of clinical bacterium suspension
and the detection of resistant M. tuberculosis was converted to the The clinical isolates of positive M. tuberculosis were inoculated
detection of M. smegmatis by solid agar culture. So detection time on the L-J slants, incubated at 37 ◦ C. After 28 days, single colonies
in phage culture method was shorted when compared with agar were scraped from slants by sterile ring vaccination and placed into
proportion method and MIC method [19,20]. The detection speed a sterile porcelain mortar. M. tuberculosis suspension of 0.5 McFar-
of M. smegmatis is much faster than that of M. tuberculosis. The land’s (1.5 × 108 cfu/mL) unit was made for susceptibility testing.
luciferase reporter phage method can improve the sensitivity of
the method. But the luciferase reporter phage (phAE142) requires a 2.1.3. Preparation of phage solution
special propagating strain of M. smegmatis (mc2 4502) as host cells Phage D29 was streaked on the M7H9 plate containing M. smeg-
[22–24]. But this method was limited by solid agar culture [25–28], matis. After 24 h, the plaques were collected with M7H9 medium.
time-consuming and complicated to operation. The phage solution was 10-fold dilution series. Then, 1.0 mL of
Recently, Dickert fabricated the mass sensitive sensor combing the different titer phage solution was mixed with M. smegma-
bio-mimetic imprinted surfaces with quartz crystal microbalances tis solution, and incubated 24 h at 37 ◦ C. The titer of phage D29
for directly detection of the yeast and bacterial cells [29]. The was calculated according to the plaque number on different M7H9
multi-channel series piezoelectric quartz crystal(MSPQC) sensor plates and dilution. The concentration of phage D29 was diluted to
technology has been applied to detecting microorganism based on 107 pfu/mL before use.
the changes of electrical parameters in the medium during the
growth process [30,31]. Ren developed a new method for rapid
2.2. Detection system
detection and antibiotic susceptibility testing of M. tuberculosis,
and the turnaround time of M. tuberculosis detection has been
2.2.1. MSPQC system
shortened to 7 days [32,33]. However, it is difficult to further
The MSPQC system was self-designed product in our lab [30].
shorten the turnaround time due to long generation time (18 h)
The block diagram of the MSPQC system is shown in Fig. 1. This
for M. tuberculosis. In addition, the PQC sensor for the detection
system consists of three major components. Part I is an array con-
of microorganism is lack of specificity. Our previous work pro-
sisting of 32 quartz crystals sensing growth signal of microbal.
posed a phage amplified–MSPQC method for the detection of M.
Label 1 is detection cells, and label 2 is piezoelectric quartz crys-
tuberculosis, which shortened the turnaround time to 30 h [34], and
tal. Part II is a signal processing system of the sensors; Part III is a
improved the specificity of the MSPQC method.
data processing system, which sets experimental parameters and
In this paper, based on criterion of resistant strains of M.
records the experimental data.
tuberculosis proposed by CLSI and our previous work, a novel
This system is sensitive to the change of electrical parameters in
phage-piezoelectric sensor for rapid drug susceptibility testing of
solution [37]; the value of frequency shift can be expressed as Eq.
M. tuberculosis was proposed. In this method, the turnaround time
(1):
was shortened to 30 h. As the phage D29 can only specifically infect
the viable M. tuberculosis and M. smegmatis, it was more special than F02 Cq (42 F02 Cs2 Y + 4F0 Cs G − YG2 )
MSPQC method. The drug susceptibility testing of 557 M. tubercu- F =  2 × G (1)
losis isolates were detected by proposed method and mycobacteria 42 F02 Cs (C0 + Cs ) − 2F0 C0 YG + G2
growth indicator tube (MGIT) method at the same time. The results
showed that there was no significant difference between two meth- • F is the frequency shift (F0 is the fundamental frequency of the
ods. But the detection time was shorted to 35 h, far less 110 h of
piezoelectric crystal);
MGIT method. • C0 and Cp are static capacitance and motional capacitance of the
crystal, respectively;
2. Experimental section • G and Cs are the conductance and capacitance of solution, respec-
tively. G = , Cs = ε + Cp , ( is the cell constant. X the specific
2.1. Reagents and isolates conductivity. ε the solution permittivity. Cp is the parasitic capac-
itance between the leading wires of the electrodes).
M. tuberculosis (H37 Ra) was bought from National Institute for • Y is a parameter related to phase shift of the oscillator.
the Control of Pharmaceutical and Biological Production. M. smeg-
matis (ATCC607, dry powder) and phage D29 (dry powder) were Eq. (1) could be simplified as Eq. (2) under the experimental
purchased from Institute of Microbiology Chinese Academy of Sci- condition [34]:
ences. M. tuberculosis clinical isolates were collected from the
Hunan Institute of Tuberculosis Control. F = −kG = −2.9G (2)
Anti-tuberculosis drugs were purchased from Sigma. The con- This is the basic principle of MSPQC analysis.
centrations of stock and working solution were prepared according MGIT 960 was produced by Becton & Dickinson Corporation.
to the CLSI procedure. The concentration of stock solution of iso-
niazid, rifampicin, streptomycin and ethambutol were 2, 40, 40,
2.2.2. Experimental procedure
100 ␮g/mL, respectively, 20 times higher than that of the working
A volume of 1.0 mL M. tuberculosis positive clinical isolate
solution [20,35,36].
suspension was individually transferred to five reaction tubes
containing 0.1 ␮g/mL isoniazid, 2.0 ␮g/mL rifampicine, 2.0 ␮g/mL
2.1.1. The preparation of M. smegmatis suspension streptomycin, 5 ␮g/mL ethambutol and a medium blank tube,
First, M. smegmatis was inoculated on L-J slants and incubated at respectively. The mixed solutions were incubated for 48 h. A vol-
37 ◦ C. After 40 h, pure lawns were scraped from slants by sterile ring ume of 0.2 mL 107 pfu/mL phage D29 solution was added to reaction
vaccination and placed into a sterile porcelain mortar, ground and tubes and infected viable M. tuberculosis for 1 h. FAS solution with
mixed with the detection medium in the porcelain mortar. Finally, final concentration of 10 m mol/L was added to reaction tubes and
 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722 717

Fig. 1. Block diagram of a multichannel series piezoelectric quartz crystal sensor system.

reacted for 5 min, 10 mL of the M7H9 medium was then added for the progeny phage D29 was released by the creaking host cell M.
susceptibility testing. tuberculosis in detection medium. The progeny phage D29 under-
A volume of 1.0 mL sample solution was added to different went a rapid cycle process of infection, replication and lysis in M.
detection cells containing 4.0 mL of detection medium, 0.5 mL of smegmstis. The cycles of phage D29 constitutes the biological chain
108 cfu/mL M. smegmatis solution, and then placed into the MSPQC amplification reaction, and causes complete inhibition of M. smeg-
system for detection. For comparison, 1.0 mL of sample solution matis growth. The number of phage D29 cycles to completely inhibit
was inoculated in a growth indicator tube of MGIT 960, and the the growth of the host cell was linear to the negative logarithm of
results were obtained after inoculated for 6 weeks in the MGIT 960 M. tuberculosis in the samples. So, the growth state of M. smegma-
system. tis is related to concentration of M. tuberculosis in the sample. The
All experiments described here were operated in Class II biolog- conductive ingredients in M7H9 medium were consumed by M.
ical safety cabinets, and all experiment wastes and utensils were smegmatis for growth, which resulted in the conductivity change
washed after autoclaving. of the medium. This change was monitored by phage-piezoelectric
sensor. The response curve of M. tuberculosis at different concen-
tration in M7H9 medium by phage-piezoelectric sensor was shown
3. Results and discussions in Fig. 2.
The phage-piezoelectric sensor was rapid, sensitivity, secu-
3.1. Principle of phage-piezoelectric method for detection of drug rity and economy. The turnaround time was greatly improved by
susceptibility testing of M. tuberculosis using MSPQC and the transformation of detection of M. tuber-
culosis into M. smegmatis and, in which phage D29 acted as a
3.1.1. The principle of the phage-piezoelectric sensor for detection bridge.
of M. tuberculosis
The turnaround time is mainly depends on bacterium gen-
eration time. The turnaround time of M. tuberculosis by the 3.1.2. The construction of phage-piezoelectric sensor for rapid
piezoelectric sensor method is 7 days because of its 18 h generation drug susceptibility testing
time [33]. To solve this problem, the phage-piezoelectric sensor The principle of the phage-piezoelectric sensor for the detec-
is constructed for the rapid detection of M. tuberculosis, and its tion of resistant strains of M. tuberculosis was shown in Fig. 3.
turnaround time are 36 h [34]. This method was based on the fact Based on the advantages of the phage-piezoelectric sensor and
that M. tuberculosis and M. smegmatis can all be infected by the the characteristics of phage D29 only infected viable bacterium,
phage D29, and they can protect the phage D29 from being killed the phage-piezoelectric method for the drug susceptibility testing
by FAS. The phage which added to the sample was transferred into was constructed according to the criterion of resistant strains of M.
the detection cells by the M. tuberculosis as a carrier. The amount tuberculosis proposed by CLSI. The basic idea was that: when the
of phage D29 which transferred into the detection cell was propor- clinical isolates are treated with drugs at a critical concentration
tional to the concentration of M. tuberculosis in the sample. Then, prescribed by CLSI, the resistant strains were survived from the
718 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722

Fig. 2. The response curve of M. tuberculosis at different concentration by phage- Fig. 4. The phage-piezoelectric curve of resistant strains and sensitive strains: (a) M.
piezoelectric method: (a) 0 cfu/mL; (b) 102 cfu/mL; (c) 103 cfu/mL; (d) 104 cfu/mL; smegmatis control; (b) sensitive strains; (c) resistant strains; (d) phage D29 control.
(e) 105 cfu/mL;

killing of drug and the sensitive strains were inactivated. The resis- were proliferated, and M. smegmatis growth was inhibited (e); for
tant strains can be rapidly detected by phage-piezoelectric sensor. sensitive strains, M. smegmatis grown well (e ).
The whole process can be divided into 5 steps: First, the suspen- The phage-piezoelectric curves of drug susceptibility testing
sion of M. tuberculosis (105 cfu) was exposed to the environment in were shown in Fig. 4. Curve a showed phage-piezoelectric response
the presence of the anti-mycobacterial agents for 48 h. The resis- curve of 0 cfu/mL M. tuberculosis with frequency shift more than
tant strains were survived from the antibiotic killing and sensitive 350 Hz, and it’s essentially the growth curve of M. smegmatis in
strains were killed by drug (a, a ). Second, the viable M. tuberculosis detection cells. curve b showed phage-piezoelectric response curve
can be infected by the phage D29, thus protecting phage D29 from of 105 cfu of sensitive M. tuberculosis with frequency shift more than
killing by FAS (b); Being killed sensitive strains can’t infected by 140 Hz, and it’s one of the sensitive strains that its frequency shift
the phage D29, phage D29 were killed by FAS (b ). Third, sample is close to that of the criterion of clinical concentration; curve c
solution was transferred to detection medium after 10-fold dilu- demonstrated phage-piezoelectric response curve of 105 cfu resis-
tion to eliminate the role of FAS, and phage D29 in viable resistant tant M. tuberculosis with the frequency shift of 30 Hz; curve d is
M. tuberculosis replicates and ultimately lyze their host cells(c); for phage D29 control with concentration of 105 pfu/mL.
sensitive strains, no phage D29 in sample solution (c ). Forth, the
released phage infected M. smegmatis host and undergone a rapid 3.2. The effect of medium on the turnaround time and sensitivity
cycle process of infection, replication and lysis (d); For sensitive of detection method
strains, no phage was amplified and no M. smegmatis was infected
by phage D29 (d ). Finally, the phage-piezoelectric signal of resis- M. smegmatis can grow in YC medium and M7H9 medium. To
tant strains was obtained because a large number of phage D29 obtain more sensitive and earlier piezoelectric signal, the response

Fig. 3. Flowchart of the phage-piezoelectric sensor for the detection of resistant strains of M. tuberculosis.
 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722 719

according to the equations 3, 4, 5. Additionally, since the phos-

phate buffer system neutralized the hydrogen ion, the frequency
shift was not affected by hydrogen ion caused by CO2 in catabolism.
All the results imply that the curve b is a signal in piezoelectric
sensor indicated anabolism, not catabolism. The turning point in
curve b demonstrates that (1) the conductive ingredients were
entirely consumed in the only presence of M. smegmatis; (2) all
M. smegmatis were infected by phage D29 when M. smegmatis and
phage D29 coexisted in the detection medium. After turning point,
the frequency shift arrived at the maximum. Beyond the turning
point, medium conductance no longer changed, the frequency shift
seemed to level off even when the time is up to 45 h, and a steady
frequency shift value is obtained.
The turnaround time of curve b is 25 h, faster than that of the
curve a (50 h). In addition, the intensity of the curve b is 350 Hz,
and much more than that of the curve a (130 Hz). Based on the
comparison between the curves a and b, anabolism of M. smegma-
Fig. 5. Growth curve of 105 cfu/mL M. smegmtis in YC and M7H9 medium
tis in M7H9 medium has earlier and more sensitive response than
that in YC, and the as-prepared M7H9 medium is more suitable for
signals of piezoelectric sensor of M. smegmatis in different media piezoelectric detection of M. smegmatis.
are analyzed. The piezoelectric responsive curves of M. smegmatis
in YC medium and M7H9 medium are shown in Fig. 5. The curve 3.3. The criterion for resistant M. tuberculosis
a is the growth signal curve of M. smegmatis in YC medium. The
level stage of the curve a means M. smegmatis utilize the non- 3.3.1. The criterion of resistant strains for rifapicin, isoniazid,
conductance ingredients in anabolic. For the downward phase of ethambutol
curve a reveals that M. smegmatis release the conductive ingre- According to the agar proportional method for drug susceptibil-
dients in their catabolism after 30 h, such as CO2 . Therefore, the ity testing for rifapicin, isoniazid, ethambutol proposed by the CLSI,
concentration of hydrogen ions caused by the saturated carbon drug-resistance was defined as the growth on tubes containing
dioxide is 1.3 × 10−4 mol/L, as shown in Eq. (3). drug greater than 1% of that on drug-free medium for rifapicin, iso-
    niazid, ethambutol [38]. For rifapicin, isoniazid, ethambutol, when
H+ = Ka1 × C = 4.3 × 10−7 × 0.04 = 1.30 × 10−4 mol/L (3) the concentration of M. tuberculosis in drug free is 105 cfu/mL, the
critical concentration of the viable bacterium between the sensi-
Because the molar conductance of hydrogen ion in dilute solu-
tive strains and resistant strains is 103 cfu/mL after drug treatment;
tion is 350 S cm2 /mol, the conductivity of hydrogen ion in detection
the average of frequency shift and standard deviation of 11 paral-
cell is 4.55 × 10−5 (S/cm), as shown in Eq. (4):
lel critical concentration at 103 cfu/mL are 171 Hz (138, 148, 153,
C × 350 × 1.3 × 10−4 160, 167, 172, 178, 186, 189, 195 and 199) and 20 Hz by phage-
= = = 4.55 × 10−5 (S/cm) (4) piezoelectric method, respectively. The frequency shift was linear
1000 1000
to the negative logarithm of M. tuberuculosis concentration [34].
The constant of the detection cell in this experiment is 0.978 cm, The higher the concentration of M. Tuberculosis was, the lower the
and the conductance caused by hydrogen ion can be expressed as frequency shift of page-piezoelectric method was. So, the isolate
Eq. (5): can be judged as the resistant strain when its frequency shift is
A less than that of critical concentration. At 95% confidence interval
L= = 4.55 × 10−5 × 0.978 = 4.5 × 10−5 (S) (5) range (x̄ ± 2Sd , 131–213), it is defined as resistant strains when the
L
frequency shift of the sample is less than 131 Hz; while sensitive
The frequency shift of a piezoelectric sensor caused by hydrogen strain more than 131 Hz.
ion is shown by Eq. (6):

H = −2.9 × L = −2.9 × 45(S) = −130.5 (Hz) (6) 3.3.2. The criterion of resistant strains for streptomycin
According to the agar proportional method for drug suscep-
The results calculated by the mentioned equations further tibility testing for streptomycin proposed by the CLSI, resistance
prove that the frequency shift (130 Hz) of the piezoelectric sen- was defined as the growth on drug containing tubes greater than
sor is obtained due to the saturated carbon dioxide resulted from 10% of the growth on drug free medium for streptmycin [38]. For
complete decomposition of carbohydrates in water during the streptomycin, the critical concentration of the viable bacterium
catabolism. between the sensitive strains and resistant strains is 104 cfu/mL
The growth state of M. smegmatis in M7H9 medium is show in M. tuberculosis after drug treatment. The average frequency shift of
the curve b. The curve b including the rising phase, turning point (T), 11 parallel sample concentration at 104 cfu/mL M. tuberculosis was
and plateau phase. The rising phase indicates that the concentra- 122 Hz (101, 104, 105, 112, 118, 123, 126, 135, 136, 139 and 144 Hz),
tion of M. smegmatis is increased along M. smegmatis. Therefore, the and standard deviation was 15 Hz. When the frequency shift of the
value of frequency shift rose up, and the speed of frequency shift- sample is less than 92 Hz, it is defined as resistant strains at 95%
up gradually increased from the point of inoculation to 30 h. The confidence interval range (x̄ ± 2Sd , 92–152); more than 92 Hz, as
reasons are that M. smegmatis can specifically utilize the conduc- sensitive strain.
tive ingredients (e.g., citrate, calcium, and iron ion), thus resulting
in the increase frequency shift in anabolism process. As a mat- 3.4. The comparison of the results of drug susceptibility testing
ter of fact, the concentrations of citrate, calcium, and iron ion are for clinical isolates by phage-piezoelectric method and MGIT 960
3.8 × 10−4 , 1.0 × 10−3 , and 8.2 × 10−5 mol/L in the M7H9 medium,
respectively, and the citrate, calcium, and iron ions consumed by The susceptibility testing of clinical strains, collected by Hunan
M. smegmatis can give rise to 43, 334, and 16 Hz frequency shift Institute of Tuberculosis Control (Changsha China), was performed
720 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722

Table 1 by the phage-piezoelectric method and MGIT 960 method. The

Comparison of susceptibility to rifampicine of clinical strains by the phage-
experimental procedure for the collection, numbering, detection
piezoelectric method and MGIT method.
and analysis were designed according to the principle of double
Phage-piezoelectric MGIT960 Total blind. Tables 1–4 show detection results of susceptibility testing of
S R clinical isolates to rifampicine, isoniazid, ethambutol and strepto-
mycin, respectively by two methods. In four drugs susceptibility
S 73(a) 4(b) 77
R 8(c) 49(d) 57 testing, the chi-square value of correlation and difference of two
detection results are shown in Table 5. The detection results are
Total 81 53 134
associated with each other at a = 0.05 level (P < 0.005), and there
are no significant difference between the results of two methods at
Table 2 a = 0.05 level (P > 0.25).
Comparison of susceptibility to isoniazid of clinical strains by the phage-
piezoelectric method and MGIT method.

Phage-piezoelectric MGIT960 Total 3.5. The comparison of turnaround time of the

S R phage-piezoelectric method and MGIT method for the
susceptibility testing
S 88(a) 2(b) 90
R 4(c) 45(d) 49
The turnaround times of the phage-piezoelectric method and
Total 92 47 139
MGIT 960 method for the susceptibility testing of M. tuberculo-
sis are stated in Table 6. The turnaround times for four first-line
Table 3
anti-tuberculosis drugs are all less than 35 h by phage-piezoelectric
Comparison of susceptibility to Ethambutol of clinical strains by the phage- method, but they are all more than 110 h by MGIT 960 method. The
piezoelectric method and MGIT method. reasons for this are stated as follows: first, the MGIT 960 method
MGIT960
was monitored the growth state of M. tuberculosis in liqium, and its
Phage-piezoelectric Total
generation time is 18 h. While the phage-piezoelectric method was
S R detected the growth process of M. smegmatis, its generation time is
S 112(a) 2(b) 114 4 h. Second, the detection of M. tuberculosis can be converted to the
R 6(c) 18(d) 24 detection of phage D29 in M. smegmatis cells in which resistant M.
Total 118 20 138 smegmatis was used as carrier of phage D29. Third, the burst size of
phage D29 in M. smegmatis was 15 much more than that the binary
fission of M. tuberculosis in MGIT 960 method, so the chain biomag-
Table 4 nifications of phage D29 in liquid medium can shorten the detection
Comparison of susceptibility to streptomycin of clinical strains by the phage-
time and distinguish the resistant and sensitive strain quickly, but
piezoelectric method and MGIT method.
the role of that doesn’t work in MGIT 960 method. Forth, the phage-
Phage-piezoelectric MGIT960 Total piezoelectric method can sensitively detect conductance changes
S R in M7H9 medium. So, the phage-piezoelectric method can quickly
complete detection and identification of susceptibility testing of M.
S 95(a) 3(b) 98
R 5(c) 43(d) 48 tuberculosis.

Total 100 46 146

Table 5
The chi-square analysis of comparison of clinical strain susceptibility results by the phage-piezoelectric method and MGIT method.

Drug Correlation Difference Sensitivity (%) Specificity (%)

2 p Conclusion 2 P Conclusion

Rifampicine 89.7 <0.005 Y 0.75 >0.25 N 91 93

Isoniazid 113 <0.005 Y 0.16 >0.50 N 96 96
Ethambutol 85 <0.005 Y 1.125 >0.50 N 95 96
Streptomycin 111 <0.005 Y 0.125 >0.25 N 95 94

Table 6
the comparison of turnaround time of drug susceptibility testing by the phage-piezoelectric method and MGIT method.

Drug Result Number of samples Phage-piezoelectric method MGIT 960 method

Turnaround time (h) Turnaround time (h)

Mean Range Mean Range

Rifampicine R 73 25 18–30 240 115–402

S 49 30 25–35 162 112–264
Isoniazid R 45 22 18–30 210 120–420
S 88 30 25–35 150 110–240
Ethambutol R 18 25 18–30 230 112–400
S 112 30 25–35 210 120–250
Streptomycin R 43 25 18–30 250 125–420
S 95 30 25–35 180 130–250
 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722 721

4. Conclusions [15] X. Wu, J. Zhang, L. Chao, J. Liang, Y. Lu, H. Li, et al., Identification of rifampin-
resistant genotypes in Mycobacterium tuberculosis by PCR-reverse dot blot
hybridization, Mol. Biotechnol. 41 (2009) 1–7.
The phage-piezoelectric method for rapid drug susceptibility [16] J. Nair, D.A. Rouse, G.H. Bai, S.L. Morris, The rpsL gene and streptomycin
testing of M. tuberculosis was successfully constructed and applied resistance in single and multiple drug-resistant strains of Mycobacterium tuber-
in clinical strains. The turnaround time of the phage-piezoelectric culosis, Mol. Microbiol. 10 (1993) 521–527.
[17] N. Siddiqi, M. Shamim, N. Jain, A. Rattan, A. Amin, V. Katoch, et al., Molecular
method was shortened to the 35 h, lower than that of MGIT 960 genetic analysis of multi-drug resistance in Indian isolates of Mycobacterium
method (110 h). The sensitivity and specificity of the proposed tuberculosis, Mem. Inst. Oswaldo Cruz 93 (1998) 589–594.
method were 91% and 93%, respectively, and consistent with that [18] A. Davies, O. Billington, B. Bannister, W. Weir, T. McHugh, S. Gillespie, Compar-
ison of fitness of two isolates of Mycobacterium tuberculosis, one of which had
of the MGIT 960 method. So the phage-piezoelectric method can
developed multi-drug resistance during the course of treatment, J. Infect. 41
replace the MGIT 960 method in rapid drug susceptibility testing (2000) 184–187.
of M. tuberculosis with practical clinical value. The proposed method [19] V.A. Steingrube, R.W. Wilson, B.A. Brown, K. Jost, Z. Blacklock, J.L. Gibson,
et al., Rapid identification of clinically significant species and taxa of aero-
is rapid and specifical for the detection of resistant strains. The sen-
bic actinomycetes, including Actinomadura, Gordona, Nocardia, Rhodococcus,
sitivity of described method was restricted by efficiency of plating Streptomyces, and Tsukamurella isolates, by DNA amplification and restriction
of phage D29 to M. tuberculosis. The detection and drug suscepti- endonuclease analysis, J. Clin. Microbiol. 35 (1997) 817–822.
bility testing can be simultaneously completed if the efficiency of [20] M.A. Saubolle, D. Sussland, Nocardiosis review of clinical and laboratory expe-
rience, J. Clin. Microbiol. 41 (2003) 4497–4501.
plating is improved. [21] E. Tortoli, M. Benedetti, A. Fontanelli, M. Simonetti, Evaluation of automated
BACTEC MGIT 960 system for testing susceptibility of Mycobacterium tuber-
culosis to four major antituberculous drugs: comparison with the radiometric
BACTEC 460TB method and the agar plate method of proportion, J. Clin. Micro-
Acknowledgements
biol. 40 (2002) 607–610.
[22] C. Carriere, P.F. Riska, O. Zimhony, J. Kriakov, S. Bardarov, J. Burns, et al., Condi-
This research work was supported by the National Natural tionally replicating luciferase reporter phages: improved sensitivity for rapid
detection and assessment of drug susceptibility of Mycobacterium tuberculosis,
Science Foundation of China (no. 21275042), and National High
J. Clin. Microbiol. 35 (1997) 3232–3239.
Technology Research and Development Program of China (863 Pro- [23] K.L. Brown, G.J. Sarkis, C. Wadsworth, G.F. Hatfull, Transcriptional
gram) (no. 2011AA02A107). silencing by the mycobacteriophage L5 repressor, EMBO J. 16 (1997)
5914–5921.
[24] P.F. Riska, W.R. Jacobs Jr., The use of luciferase-reporter phage for antibiotic-
susceptibility testing of mycobacteria, in: Mycobacteria Protocols, Springer,
References 1998, pp. 431–455.
[25] H. David, S. Clavel, F. Clement, Adsorption and growth of the bacteriophage D29
[1] M.C. Raviglione, The TB epidemic from 1992 to 2002, Tuberculosis (Edinburgh, in selected mycobacteria, Ann.Inst. Pasteur/Virol. (1980) 167–184.
Scotland) 83 (2003) 4. [26] V. Chihota, A. Grant, K. Fielding, B. Ndibongo, A. van Zyl, D. Muirhead, et al.,
[2] E.T.-Y. Leung, K.-M. Kam, A. Chiu, P.-L. Ho, W.-H. Seto, K.-Y. Yuen, et al., Detec- Liquid vs. solid culture for tuberculosis: performance and cost in a resource-
tion of katG Ser315Thr substitution in respiratory specimens from patients constrained setting, Int. J. Tuberc. Lung Dis. 14 (2010) 1024–1031.
with isoniazid-resistant Mycobacterium tuberculosis using PCR-RFLP, J. Med. [27] J. Dhillon, D.B. Lowrie, D.A. Mitchison, Mycobacterium tuberculosis from chronic
Microbiol. 52 (2003) 999–1003. murine infections that grows in liquid but not on solid medium, BMC Infect.
[3] S. Ahmad, E. Mokaddas, A. Jaber, Rapid detection of ethambutol-resistant Dis. 4 (2004) 51–52.
Mycobacterium tuberculosis strains by PCR-RFLP targeting embB codons 306 [28] C. Ang, M. Cajucom, Y. Kim, H. Bang, H. Lee, S. Cho, et al., Evaluation of a rapid
and 497 and iniA codon 501 mutations, Mol. Cell. Probes 18 (2004) 299. assay for identifi cation of Mycobacterium tuberculosis grown in solid and liquid
[4] M. Patnaik, K. Liegmann, J.B. Peter, Rapid detection of Smear- media, Int. J. Tuberc. Lung Dis. 15 (2011) 1475–1478.
NegativeMycobacterium tuberculosis by PCR and Sequencing for Rifampin [29] M. Polreichova, U. Latif, F.L. Dickert, Functionalized polymers as receptors for
Resistance with DNA Extracted Directly from Slides, J. Clin. Microbiol. 39 detection of cells, Aust. J. Chem. 64 (2011) 1256–1260.
(2001) 51–52. [30] F. He, X. Zhang, J. Zhou, Z. Liu, A new MSPQC system for rapid detection of
[5] R. Shi, J. Zhang, C. Li, Y. Kazumi, I. Sugawara, Detection of streptomycin resis- pathogens in clinical samples, J. Microbiol. Methods 66 (2006) 56–62.
tance in Mycobacterium tuberculosis clinical isolates from China as determined [31] F. He, J. Ren, Z. Liu, The study and application of a new IDE–PQC sensor, Sens.
by denaturing HPLC analysis and DNA sequencing, Microbes Infect. 9 (2007) Actuators, B 123 (2007) 1057–1063.
1538–1544. [32] J. Ren, F. He, S. Yi, X. Cui, A new MSPQC for rapid growth and detection of
[6] C. Mani, N. Selvakumar, V. Kumar, S. Narayanan, P.R. Narayanan, Comparison of Mycobacterium tuberculosis, Biosens. Bioelectron. 24 (2008) 403–409.
DNA sequencing, PCR-SSCP and PhaB assays with indirect sensitivity testing for [33] J.L. Ren, L.N. Ma, Z.H. Li, Q.L. Lin, H.X. Huang, S.L. Yi, Simultaneous and early
detection of rifampicin resistance in Mycobacterium tuberculosis, Int. J. Tuberc. detection of Mycobacterium tuberculosis resistance to antituberculosis drugs
Lung Dis. 7 (2003) 652–659. using an indirect series piezoelectric system, Biosens. Bioelectron. 43 (2013)
[7] X. Cheng, J. Zhang, L. Yang, X. Xu, J. Liu, W. Yu, et al., A new Multi-PCR-SSCP assay 115–119.
for simultaneous detection of isoniazid and rifampin resistance in Mycobac- [34] X.W. Mi, F.J. He, M.Y. Xiang, Y. Lian, S.L. Yi, Novel phage amplified multichannel
terium tuberculosis, J. Microbiol. Methods 70 (2007) 301–305. series piezoelectric quartz crystal sensor for rapid and sensitive detection of
[8] L.M. Aragón, F. Navarro, V. Heiser, M. Garrigó, M. Español, P. Coll, Rapid Mycobacterium tuberculosis, Anal. Chem. 84 (2012) 939–946.
detection of specific gene mutations associated with isoniazid or rifampicin [35] L.B. Reller, M.P. Weinstein, G.L. Woods, Susceptibility testing for mycobacteria,
resistance in Mycobacterium tuberculosis clinical isolates using non-fluorescent Clin. Infect. Dis. 31 (2000) 1209–1215.
low-density DNA microarrays, J. Antimicrob. Chemother. 57 (2006) 825–831. [36] G.L. Woods, B. Brown-Elliott, E.P. Desmond, G.S. Hall, L. Heifets, G.E. Pfyffer,
[9] S.-Y.G. Lin, W. Probert, M. Lo, E. Desmond, Rapid detection of isoniazid and et al., Susceptibility Testing of Mycobacteria, Nocardiae, and Other Aerobic
rifampin resistance mutations in Mycobacterium tuberculosis complex from cul- Actinomycetes: Approved Standard, NCCLS, 2003.
tures or smear-positive sputa by use of molecular beacons, J. Clin. Microbiol. [37] D.-Z. Shen, W. Zhu, L.-H. Nie, S.-Z. Yao, Behaviour of a series piezoelectric sensor
42 (2004) 4204–4208. in electrolyte solution: Part I. Theory, Anal. Chim. Acta 276 (1993) 87–97.
[10] A.S. Piatek, A. Telenti, M.R. Murray, H. El-Hajj, W.R. Jacobs, F.R. Kramer, et al., [38] C.Q.F. Leite, A.L.R.Z. Beretta, I.S. Anno, M.A.d.S. Telles, Standartization of broth
Genotypic analysis of Mycobacterium tuberculosis in two distinct populations microdilution method for Mycobacterium tuberculosis, Mem. Inst. Oswaldo Cruz
using molecular beacons: implications for rapid susceptibility testing, Antimi- 95 (2000) 127–129.
crob. Agents Chemother. 44 (2000) 103–110.
[11] F. Brossier, N. Veziris, C. Truffot-Pernot, V. Jarlier, W. Sougakoff, Performance
of the genotype MTBDR line probe assay for detection of resistance to rifampin Biographies
and isoniazid in strains of Mycobacterium tuberculosis with low-and high-level
resistance, J. Clin. Microbiol. 44 (2006) 3659–3664.
[12] H. Park, E.J. Song, E.S. Song, E.Y. Lee, C.M. Kim, S.H. Jeong, et al., Comparison
Fengjiao He. She received her PhD degree in analytical chemistry at Hunan University
of a conventional antimicrobial susceptibility assay to an oligonucleotide chip
in 1994, PR China. She was a post-doctoral fellow at UCLA from 1997 to 1999. She is
system for detection of drug resistance in Mycobacterium tuberculosis isolates,
now a professor in the department of Chemistry and Chemical Engineering of Hunan
J. Clin. Microbiol. 44 (2006) 1619–1624.
University. She is interested in development and application of piezoelectric quartz
[13] D. Bang, Å.B. Andersen, V.Ø. Thomsen, Rapid genotypic detection of
crystal biosensor in microorganism.
rifampin-and isoniazid-resistant Mycobacterium tuberculosis directly in clinical
specimens, J. Clin. Microbiol. 44 (2006) 2605–2608. Xianwen Mi. He is a doctoral student of State Key Laboratory of Chemo/Biological
[14] L.M. Steinlein, J.T. Crawford, Reverse dot blot assay (insertion site typing) for Sensing and Chemometrics, Hunan University, PR China. His main research interest
precise detection of sites of IS6110 insertion in the Mycobacterium tuberculosis is in study, development and application of novel piezoelectric quartz sensor for
genome, J. Clin. Microbiol. 39 (2001) 871–878. microorganism detection.
722 X. Mi et al. / Sensors and Actuators B 193 (2014) 715–722

Meiyu Xiang. She received her MS degree in State Key Laboratory of Chemo/Biological is in study, development and application of novel piezoelectric quartz sensor for
Sensing and Chemometrics, Hunan University, PR China. She is interested in biolog- microorganism detection.
ical and chemical sensors for applications.
Songlin Yi. He is assistant director technician in Hunan institute of tuberculosis con-
Yan Lian. He is a doctoral student of State Key Laboratory of Chemo/Biological trol. He mainly engaged in detection and the improvement of detection method of
Sensing and Chemometrics, Hunan University, PR China. His main research interest Mycobacterium tuberculosis.