Detection of Volatile Organic Compounds Associated ... - Springer Link

2 downloads 0 Views 136KB Size Report
Abstract—Macrosmatic animals (dogs and mice) have been proved to be able to distinguish between the .... Dogs can detect cancer by odor at early stages;.
ISSN 10623590, Biology Bulletin, 2015, Vol. 42, No. 3, pp. 239–245. © Pleiades Publishing, Inc., 2015. Original Russian Text © E.I. Rodionova, M.Yu. Kochevalina, E.V. Kotenkova, O.V. Morozova, G.A. Kogun’, E.L. Bataeva, A.V. Ambaryan, 2015, published in Izvestiya Akademii Nauk, Seriya Biologicheskaya, 2015, No. 3, pp. 293–301.

ANIMAL AND HUMAN PHYSIOLOGY

Detection of Volatile Organic Compounds Associated with Hepatocellular Carcinoma by Macrosmatic Animals: Approaches to the Search for New Tumor Markers E. I. Rodionovaa, M. Yu. Kochevalinaa, E. V. Kotenkovab, O. V. Morozovac, G. A. Kogun’d, E. L. Bataevad, and A. V. Ambaryanb a

Kharkevich Institute for Information Transmission Problems, Bol’shoi Karetnyi per. 19, str. 1, Moscow, 127051 Russia b Severtsov Institute of Ecology and Evolution, Leninskii pr. 33, Moscow, 119071 Russia c Blokhin Cancer Research Institute, Institute of Carcinogenesis, Kashirskoe sh. 24, Moscow, 115478 Russia d Cynological Division of Aviation Security Service, Aeroflot, Russian Airlines, ul. Arbat 10, Moscow, 119002 Russia email: [email protected] Received May 21, 2014

Abstract—Macrosmatic animals (dogs and mice) have been proved to be able to distinguish between the urine or feces of mice with transplanted hepatocellular carcinoma and those of healthy mice by odor. The chemical composition of animal excreta was found to change with tumor growth; however, it is not clear yet if this results from tumor growth itself, inflammation, or immune response. We suggested that the use of the ability of macrosmatic animals to compare odor mixtures combined with mouse cancer models is a promising trend in the search for new tumor markers. DOI: 10.1134/S1062359015030103

INTRODUCTION The number of people with malignant diseases increases annually (Jemal et al., 2011). It is obvious that treatment success of cancer depends directly on the early detection of disease. However, no efficient methods for early diagnostics, and in particular effi cient screening tests, are available at present. The use of known biomarkers is not yet effective enough because of their low specificity (Horváth et al., 2009). One of the relatively new and rapidly growing trends in clinical diagnostics is the analysis of volatile organic compounds (VOCs) found in human biologi cal specimens, which are typical of different diseases, such as lung, breast, colorectal, and prostate cancers (Horváth et al., 2009; Hakim et al., 2010; Horvath et al., 2010). It is known for some time that the growth of malignant tumors of various nature changes the composition of VOCs in patient specimens (O’Neill et al., 1988). The VOCs associated with malignant tumor growth have been studied for almost 30 years (Persaud and Dodd, 1982; Tisch and Haick, 2010). The technology for detecting VOCs and techniques for their discrimination are continually being improved. At present, the sensitivity of such equipment, as well as the accuracy of analysis, is rather high (Kuano et al., 2011; Kwak et al., 2013). A number of VOCs of various functional groups, such as alkanes, methylated alkanes, benzene derivatives, and others, were found to be potentially useful as markers of certain tumors (Kuano et al., 2011). The concentration and propor

tion of volatile compounds were proved to differ in air exhaled by patients and healthy subjects (Horváth et al., 2009; Hakim et al., 2010), as well as in urine and other specimens of patients and healthy individuals (Hakim et al., 2010; Hanai et al., 2012), and in the headspace of the of normal and cancer cells cultures (Amal et al., 2012; Kwak et al., 2013). However, despite certain progress in investigating VOCs associated with cancer, we still have a lot to study before we can identify reliable tumor biomarkers and develop an effective method for noninvasive diag nostics of tumors. There is limited data on VOCs that can be used as markers; it is not clear yet if VOCs occur in all cancer types, if there are common VOCs for all or several cancers, or if each tumor type has a unique set of them. We do not yet know if VOCs that are typi cal of cancer are present in all kinds of excreta or only in those that are related most closely to the affected organ. The above is only a small part of all pending issues. Until now, there has been no successful study of volatile biomarkers leading to an attempt to introduce them into clinical practice; most likely, it means that no results were verified. There may be many causes of that. On the one hand, studies of volatile markers are performed in different laboratories where samples of various kinds of excreta are used (exhaled breath, urine, or blood), so it is almost impossible to compare the data obtained. The measurement of VOCs concen tration by instrumental methods, such as gas chroma tography–mass spectrometry (GCMS), are not yet

239

240

RODIONOVA et al.

standardized, and the tuning of equipment, sampling, collection of VOCs from the samples, VOC analysis, and other procedures have a significant effect on the results (Koczulla et al., 2011). On the other hand, a patient specimens VOCs anal ysis involves great difficulties. The VOCs composition is affected not only by the type of tumor and diseased organ, but also by individual’s sex, age, emotional state, some habits (smoking, diet, etc.), medicine taken, and many other factors (Penn and Potts, 1998; Ackerl et al., 2001; Qin et al., 2010). A significant part of VOCs is genetically predetermined, and the human population is very heterogenous (Havlicek and Rob erts, 2009). Each type of excretion is known to contain large number of VOCs; more than 3000 of them can be detected in exhaled breath, for example (Horváth et al., 2009). Therefore, the composition of VOCs dif fers significantly among individuals. Additional problems arise in the study of VOCs biological spacimens of cancer patients. Those speci men are usually taken from those individuals who have their diagnostic set; i.e., they stay in the hospital and are under stress conditions, even if the treatment has not yet started. This may influence the composition of VOCs. Healthy individuals in control groups are not exposed to the factors described, and composition of their VOCs may differ not only in the content of can cer markers. Thus, the VOCs that are thought to indi cate the presence of cancer may actually not be associ ated with any disease at all. Such an effect has been observed and discussed in some works (Hanai et al., 2012; Walczak et al., 2012). The lack of ideas about what substances may be rel evant for the diagnosis makes it difficult to search for substances associated with the development of con crete tumors. Human studies do not allow us to answer this question since the genetic differences between individuals are great, and it is too difficult to detect the substances associated with a cancer among a great number of VOCs. For the detection of VOCs marking cancer it is nec essary not only to analyze the composition of secre tions, but also have a tool to compare odor mixtures from patients and healthy individuals, revealing the presence of components associated with the disease. No instrumental methods for this purpose are avail able at present. Obviously, some new approaches to the search for volatile markers are needed to solve the main problems in these studies; these approaches include reducing the diversity of the analyzed VOCs, tumor development monitoring, and a rapid method for comparing VOC spectra and detecting the presence of cancer markers. We suggest solving the problem of diversity of VOCs analyzed by using laboratory mice instead of human patients at early stages of studies; these mice can serve as “unified patients” since genetic and other differ ences among them are much less than among humans. The process of tumor growth in mice differs from that

in humans (Anisimov et al., 2005); however, our key aim is to develop some principles of searching and not to detect VOCs themselves. We suggest studying the strains of mice bred for investigating the processes of tumor growth with the use of transplanted tumors and transplanted tumors. This procedure allows us to transplant approximately the same amount of tumor tissue to all experimental animals at the same time, i.e., to observe synchronous tumor growth. This model makes it possible to monitor the tumor development and analyze VOCs spectrum changes associated with the stages of tumor development and characteristics. In spite of the high sensitivity of modern analytical equipment, comparison of samples and discrimina tion of components associated with cancer remains a complicated task. Macrosmatic animals deal success fully with this problem. Therefore, we suggest using dogs and mice for comparing odor mixtures of healthy and sick animals, and as a proof that odors are signifi cantly different and that these differences are caused by the tumor growth. The olfactory system of these animals is very sensitive; it contains more than 1000 olfactory receptors. Because of that, and due to a number of behavior tasks that these animals complete in nature by means of olfaction, they can memorize many substances (Williams and Johnston, 2002), ana lyze and compare mixtures, and discriminate signifi cant components (Sokolov et al., 1990; Schoon, 1996; McCulloch et al., 2006). The odor of a disease is a strong stimulus for animals, and they can respond to it without prior training (Arakawa et al., 2011). More over, a specialized gene family of mammalian olfac tory receptors—formyl peptide receptors (FPR)— have proved recently that it is very likely for mice to detect infected individuals by their discharge (Riviere et al., 2009). Animals cannot give us any information about the composition of the odor mixture, but unlike machines they can discriminate certain patterns of a VOC mix ture that are associated with some properties of an organism (sex, individual smell, or disease) (Sokolov et al., 1990; Matsumura et al., 2010). We can train the animal, or design the experiment in such a way that the animal will indicate the mixtures that contain similar patterns of VOCs. Animals’ abilities enable us to use dogs and mice as biosensors. Dogs can detect cancer by odor at early stages; moreover, the reliability of dogs is higher than that of electronic sensors for detecting volatile tumor markers (Bransbury et al., 2004; Gordon et al., 2008; Horvath et al., 2008; Cornu et al., 2011). Dogs discriminate healthy individuals from patients with skin, prostate, lung, bladder, breast, and ovarian cancer (Bransbury et al., 2004; McCulloch et al., 2006; Gordon et al., 2008; Horvath et al., 2008, 2010; Matsumura et al., 2010; Cornu et al., 2011; Sonoda et al., 2011). In all those studies, animals were first trained for a long period (5–12 months) to distinguish the odor of patients with cancer, and after that for them were pre BIOLOGY BULLETIN

Vol. 42

No. 3

2015

DETECTION OF VOLATILE ORGANIC COMPOUNDS ASSOCIATED

sented 5–7 excreta samples, among which one belonged to a patient with a certain cancer. The dogs discriminated the patient sample with high probability. The results of such studies have been criticized by many authors, partly because of the different designs of these experiments (Boedeker et al., 2012), and partly because it is not clear enough what kind of olfactory “clues” a dog uses to discriminate between a healthy subject and a sick one and whether these “clues” are associated with the disease. It is also not obvious how to use the results obtained. Obviously, animals are unlikely to be widely used in clinical prac tice since the training of dogs and trainers requires considerable costs, selection of dogs with best olfac tory acuity, etc. Dog performance is not always high since some physiological changes can affect it. The learned smell of one disease also limits the opportuni ties of using dogs as biosensors in laboratories. We suggest using “line up” procedure for detecting cancer volatile markers by dogs; this procedure has already been used successfully in a number of coun tries, including Russia. According to this method, dogs are used as a “forensic tool,” and the evidence obtained is accepted in court (Schoon and Haak, 2002), including Russian courts (Korenevskii, 2000). Animals are trained a ‘matchtosampie’ like proce dure. Like procedures used for detecting cancer, this method is based on comparison of odor mixtures and detectionof the common component. A dog smells the initial odor for a minute. After that, the dog is asked to match it to those in the lineup (Schoon, 1996; Korenevskii, 2000). The use of the preliminary dem onstration of the initial odor allows for training the animal with different odors and for changing initial odor instead of memorizing just one and restricting the use of an animal (Krutova et al., 1997). This method has also been used in a number of studies (Sokolov et al., 1990; Krutova et al., 1997; Rodionova et al., 2009). Mice are also capable of comparing smells and therefore methods based on learning (Matsumura et al., 2010) and not learning one certain odor (Heth and Todrank, 2000) were developed. Mice, just like dogs, can discriminate between a mouse with a cancer and a healthy one (Matsumura et al., 2010). This pro vides the opportunity to use these animals as biosen sors. In our studies on the scent of model transplanted hepatocellular carcinoma (HCC) in mice, we used dogs and mice as biosensors. Hepatocellular carci noma is a very aggressive disease with a dismal progno sis (Hussain et al., 2001). VOCs associated with this disease were found by GCMS in patients’ exhaled breath (O’Neill et al., 1988), in blood (Xue et al., 2008), and in the headspace of cancer cells cultures (Amal et al., 2012). Firstly, we decided to explore if there were any VOCs in the urine and feces that could allow discrim inate between healthy and sick mice. We had no infor mation about any earlier works that would show that BIOLOGY BULLETIN

Vol. 42

No. 3

2015

241

there are VOCs associated with HCC in that discharge. We supposed that the mouse model of HCC would reduce the diversity of the VOCs analyzed and allow disease monitoring and that the use of macrosmatic animals as biosensors will provide rapid comparison of odors of healthy and sick animals as well as different stages of tumor growth. MATERIALS AND METHODS Experimental models. As “unified patients” we used BDF1 hybrid males (C57B1/6 × DBA2)—experi mental biomodels for investigating tumor growth and evaluating the efficiency of antitumor drugs, based on transplanted tumors. We employed a transplantable hepatic tumor strain H33 obtained in Blokhin Cancer Research Institute, Institute of Carcinogenesis, from HCC induced by a intraperitoneal injection of dieth ylnitrosamine (90 mg/kg body weight) to the same BDF1 hybrid mice (Lazarevich et al., 2004). Tumor tissue (100 mg) suspension in 0.5 mL of saline (0.9%) was transplanted subcutaneously to 2 to 3monthold hybrid males. Healthy hybrid BDF1 mice of the same age and sex, kept under the same conditions on the same diet were used as the control group. We studied two control groups of mice: the first one consisted of intact animals, the second one, of control mice injected subcutaneously with 0.5 mL of saline (0.9%) at the same time as tumor was transplanted, with a needle of the same diameter. During the period of studies, the mice were kept in one vivarium room in daylight conditions. Sawdust was used as litter. Ani mals were given ad libitum access to feed (oats and compound feed) and water. The samples of urine and feces were collected 7– 10 days after the tumor transplantation (latestage cancer, visible subcutaneous neoplasm with a diame ter of 1–1.5 cm). were collected by gently pressing the mouse belly and placed in plastic tubes. Every tube with samples taken from a healthy or a sick animal was marked so that one could determine to which specific animal and which stage of tumor growth the sample belongs. The samples of wastes from the experimental and control mice collected during one period of time (2–3 h) were shown to biosensor animals. Urine sam ples were frozen at –20°C and thawed shortly before the experiment. Altogether, 24 sick and 57 healthy mice were used in the experiments with dogs and mice. Biosensor animals. The first set of experiments was carried out with the use of specially trained dogs (jackaldog hybrids) from the kennel of the Cynologi cal Division of Aviation Security Service, Aeroflot, Russian Airlines. We used a matchtosamplelike protocol of scent identification lineups by dogs; the method used in forensic science and research (Korenevskii, 2000). A dog was exposed to the initial scent for a minute, for memorizing it; than the dog is supposed to detect the most similar odor in the range of 12 samples (1 experimental sample and 11 control

242

RODIONOVA et al.

ones). Five dogs (two males and three females) at the age from 5 to 10 years were used as sensors. The dogs worked with professional trainers. In the second set of experiments, 11 healthy 6– 7 month old CBA mice were used as biosensors. Ten days before the experiment, every mouse was kept in a separate standard box (25 × 12 × 10 cm) with a net cover; mice were also kept in these boxes during the experiments. Dog experiments. In the experimental room (4 × 4 m), a temperature of 18–20°C and humidity of at least 35% were maintained. On the floor, a circle consisting of 12 points placed at regular intervals was marked. Urine (30–40 µL) or feces were applied onto a small strip of fabric (0.5 × 3 cm) which was then placed in a preliminarily washed and kept at 70°C 0.5 L glass jar, so that the dog could not touch the sample while smelling. The total number of jars prepared for each experiment was 13; the jars contained one type of dis charge (urine or feces) collected from different mice. Two jars contained samples of sick mice discharge and 11 jars contained samples of healthy mice discharge. The jars were covered with glass lids which were taken off shortly before the experiment and placed back right after it. Twelve jars (11 of which contained samples from healthy mice and one sample from a sick one) were placed in the points of the circle; the last jar with a sample from a sick mouse was used for preliminary demonstration to the dog and odor memorizing. Before every experiment, the dog was taken to the experimental room, placed at some distance from the circle and presented with the jar with sick mouse odor for 1 minute. After that, the dog smelled at the range of 12 jars containing tested samples. The trainer lead ing the dog did not know which of the jars contained the sample from a sick mouse. When the test sample was found, the dog sat in front of the jar and barked. When a dog found the jar with sick mouse odor, we recorded that as a right choice. If a dog chose a control sample (false signal), it was recorded as a wrong choice. If the dog smelled all the samples and did not choose any, or gave one to three false signals, we recorded that the dog did not find the test sample. A dog that gave more than three false signals during one passage was considered as nonfunctional, withdrawn from the study on that very day; the data obtained in this experiment were not taken into account. The place of the test sample was changed in random order before every passage; every time a dog started the pas sage from a new point. Each dog made one or two pas sages per day. We ran 66 experiments with urine samples and 30 experiments with feces samples. Each experiment was recorded on paper and on video tape. We used a con ventional method of calculation (Horvath et al., 2008; Cornu et al., 2011). Since every test consisted of sev eral passages, and in each of them one targeted sample and 11 controls were used, in the case of accidentally

right choice, the probability of success was 1/12. The sensitivity of the test was determined as the percentage of samples taken from sick mice, detected correctly by the dogs; specificity was determined as the percentage of control samples not chosen. Mice experiments. We used a modification of the habituationgeneralization method which allows us to compare the similarity and difference between odors (Heth and Todrank et al., 2000). The principle of this method is the demonstration of one certain odor at the start, i.e., at the habituation stage, followed by simul taneous demonstration of two other odors at the rec ognition stage. The greater the similarity of these odors and the first one, the less the time of sniffing at the recognition stage. If the mouse sniffs one sample of the recognition stage significantly longer than the other one, it means that the former odor differs from the habituation stage odor to a bigger extent than the latter. One of the modifications of the habituation generalization method is widely used for studying how the animals recognize each other by individual odor. Since the aim of our study was to investigate the scent of disease, we eliminated the possibility of recognition by individual odor and focused attention of the ani mals on the scent of disease (all animals, whos odor was presented to sensor mice, were different). At the habituation stage, a urine sample taken from an unfa miliar healthy mature male (the starting odor) was demonstrated to the biosensor mice; at the recogni tion stage, a sample taken from another unfamiliar healthy mature male and a sample taken from one more unfamiliar mature male with a tumor trans planted 7–10 days before were used. The experiments were run in transparent chambers with walls of polymethyl methacrylate (25 × 23 × 20 cm). Before each experiment, the mice were allowed to explore the chamber for 3 minutes. Then the chamber was divided in two equal parts by a septum; in the first part, the starting sample was presented to the male. The habituation stage lasted 3 minutes. The starting sample (a drop of urine, 10–15 µL) was applied on a slide so that it was easy to insert and to remove the sample quickly. After that, we took away the septum and the starting sample, and the mice moved to another part of chamber where two drops of urine (10–15 µL) was placed on the floor at a distance of 10 cm. The recognition stage lasted 5 minutes. The time of sniffing was measured by a stopwatch with an accuracy of 0.1 s. It was a blind experiment; i.e., at the recognition stage the conductor did not know which urine drop is the targeted one. The experiments started not earlier than at 6:00 p.m., i.e., in the maximum activity period. The experiments were run 2–3 times a week; the interval between experiments for each mouse was at least 48 h. After 2–3 sets a break was taken. Each urine samples was never demonstrated twice to one mouse. At least three urine donors of each possible variant were used in each set. After each pas BIOLOGY BULLETIN

Vol. 42

No. 3

2015

DETECTION OF VOLATILE ORGANIC COMPOUNDS ASSOCIATED

243

Recognition of urine odors of hybrid BDF1 male mice with HCC and controls by male CBA mice Initial odor Healthy male 1

Healthy male 1

Odors presented Time of sniffing at the recognition Significance of differences at the recognition stage stage: total (minimal–maximal) (Wilcoxon test for paired samples)* Healthy male 1 2

36.6 (1–7) 76.9 (3–15.5)

n = 11, T = 0 p = 0.005*

Healthy male 2 H33 sick male

29.7 (1–5) 82.2 (3–14)

n = 10, T = 0 p = 0.005*

Figures 1 and 2 designate different animals used for one experiment (at least three urine donors for each set of experiments and for each variant of odor). * The level of significance: p < 0.01.

sage of a mouse, the chamber was thoroughly washed and ventilated for at least 24 h. The data on the demonstrated odors, experiment sets, and statistical significance of differences in olfac tory signals are presented in the table. The results were processed by the paired nonparametric Wilcoxon test for dependent variables with the use of the STATISTICA program package. RESULTS AND DISCUSSION In 27 out of 30 experiments with feces samples, the odors of sick animals were detected correctly by the dogs (sensitivity is 90%); only 3 control samples out of 330 were falsely chosen as test sample (specificity is 99.1%). In 45 out of 66 experiments with urine sam ples, the odors of sick animals were detected correctly (sensitivity is 76.8%), and 12 control samples out of 622 were falsely chosen (specificity is 98.1%). These data show that the dogs do not always find the sample from a sick animal, but extremely rarely chose the sam ples from healthy animals instead of the targeted ones. The efficiency of biosensor mice in the habitua tiongeneralization method was estimated as follows: in the control set of experiments at the habituation stage, we demonstrated a sample from a healthy mouse; at the recognition stage, we had two samples from healthy mice, one of which was the male urine sample demonstrated at the habituation stage and another from an unfamiliar male (see table). In this set, the sensor mice discriminate well between the new odor from the familiar one. Then we presented a urine sample from a healthy mouse at the habituation stage and two samples (from another healthy and one sick mouse) to the same biosensor mice at the recognition stage. In this case, the sensor mice sniffed the sample from a sick animal significantly longer than that of a healthy one (table). This confirms that the odors of healthy animals and animals with a cancer are signifi cantly different. The results show that macrosmatic animals—both dogs and mice—can significantly distinguish dis charges of healthy animals from those of animals with BIOLOGY BULLETIN

Vol. 42

No. 3

2015

subcutaneously transplanted HCC by odor without preliminary training; i.e., the discharge of sick animals (both urine and feces) contain VOCs typical of HCC. The presence of these compounds in these kinds of discharge has not been observed before. So, the nature of this odor is still unknown; it is not yet clear if it results from tumor growth, inflammation, or immune response. In spite of the fact that some of the mice from the control group used in the dog experiments and all control mice used in the mice experiments were also traumatized like the sick ones were trauma tized during tumor transplantation (by injection of the same dose of normal saline with a needle of the same diameter), biosensor animals easily discriminate between the mice with a transplanted tumor and the control ones. This suggests that biosensor animals focused on VOCs associated with tumor growth but not with the bodily response to trauma. We used transplantable hepatic tumor strain H33 which was obtained from HCC induced in males of the BDF1 line; therefore, the immune response against tumor tissue could have been not as strong as cells of mice of other lines. This suggests that the effect of the immune response to the composition of VOCs of sick animals can be less significant than when using mice of another strain. However, this suggestion needs verifi cation. The experiments showed that the method allows for rapid changes in the targeted odors presented to sensor animals. For example, the mice switched atten tion easily from individual odor to the scent of disease. In the dog experiments, we moved easily from urine samples to feces samples using the same group of dogs. After that, the dogs discriminate sick mice samples with a high probability. Certainly, it is likely that the VOC composition associated with tumor growth is identical or very similar to urine and feces of sick ani mals; but it should also be verified. The sensitivity of our method in dog experiments was slightly lower than in other studies where the dogs detected samples of patients (Bransbury et al., 2004; McCulloch et al., 2006; Horváth et al., 2009). Such findings may result from several causes. Firstly, our

244

RODIONOVA et al.

dogs completed a more difficult task than that in other studies. The dog usually selects the test odor out of 5– 6 samples (Bransbury et al., 2004; McCulloch et al., 2006; Gordon et al., 2008; Horváth et al., 2009; Cornu et al., 2011; Sonoda et al., 2011); our dogs select 1 out of 12 samples (1 targeted sample and 11 controls). Moreover, the urine and feces samples used in this study could differed in the amount and composition of VOCs. For example, the composition of samples can be affected by sampling time. We collected our sam ples at different times of day. Furthermore, one possi ble cause of the high sensitivity of the method shown in studies on humans is that dogs can focus not only on the odor of disease but also on other complexes of VOCs during sumple discrimination. In one such study, it was suggested and proved that in cases when control samples are taken from people who work or stay for a long time at the same hospital as the patients the sensitivity of the method in dog experiments decreases approximately to the level obtained in our studies (Walczak et al., 2012). The data obtained by some authors show that every type ofcancer tumor has a specific odor (Matsumura et al., 2010). However, VOCs specific either to cancer in general or to certain types of tumors have not yet been isolated with a sufficient level of significance. This is probably because searching for VOCs present in discharges among the great number of volatile com pounds of the organism is like looking for a needle in a haystack; it is not evident which part of the VOC spec trum should be sought. Furthermore, the composition of VOCs associated with tumor growth in discharges seems to be rather complex. In some studies, the com position of VOCs in the headspace of cancer cells was proved to differ from that the normal cell (Amal et al., 2012; Kwak et al., 2013). It means that the changes in metabolic processes in tumor cells apparently lead to the excretion of VOCs typical of this type of tumor. It was shown recently than several histologically different types of lung cancer can be discriminated by detection of VOCs in headspace of the cell cultures by GCMS with the use of array sensors (Kwak et al., 2013). The dogs were proved to discriminate between the smell of cancer and healthy ovarian tissue (Hor vath et al., 2008). All these data is the evidence of the presence of volatile cancer markers in cell cultures and tissues. The changes in metabolic processes in a group of cells affect the composition of VOCs excreted by the whole organism; some of the compounds excreted by tumor cells in culture are also present in discharges of animals with transplanted tumors (Hanai et al., 2012b). We do not know yet if all cancer types affect the odor of the organism. However, since a difference between individuals in just one gene affects the individual odor of the animal recognized by conspecific individuals (Yamazaki and Beauchamp, 2007), most likely, the growth of a tumor of any type leads to some changes in the odor as cancer cause huge rearrangements in many genes of many types of cells. If the odor of a disease

reflects the changes in genes expression and the vola tile compounds appear as a result of decomposition of proteins synthesized in a “wrong way” because of these changes, these compounds must be present in the internal environment (lymph, blood, and intersti tial fluid), and consequently in all kinds of discharges, in one form or another. Indeed, the compounds asso ciated with tumor growth are excreted by the cells into interstitial fluid (Amal et al., 2012; Kwak et al., 2013) and are present in blood (Xue et al., 2008; Horváth et al., 2010;) and their metabolites are excreted by the organism. Obviously, a growing tumor causes a response of the immune system. The functioning of these cells also changes, and changed metabolites are present in wastes. The volatile markers associated with certain diseases have not yet been identified probably because in different studies VOCs from different parts of the spectrum are investigated, while the proportions of these parts vary for different patients. Our results suggest that the use of macrosmatic ani mals together with instrumental methods of analysis of VOCs associated with cancer and the controlled mouse model of tumor growth is a promising tech nique and can be useful for many tasks. Although this method does not allow for detection of the chemical structure or composition of compounds, it provides information about the similarities and differences of odors presented to a biosensor animal. Using the ani mals, we can determine the presence of VOCs typical of cancer. We suppose that the use of macrosmatic ani mals will allow discriminate between these VOCs and those associated with inflammation process, immune response, etc. This will make the instrumental detec tion of VOCs more taskoriented. ACKNOWLEDGMENTS The authors are grateful to K.T. Sulimov (Likhachev Russian Research Institute for Cultural and Natural Heritage) for their help in conducting experiments with dogs. This work was supported by the Program of the Presidium of the Russian Academy of Sciences no. 10 “Chemical Analysis and Study of the Structure of Materials: Fundamentals and New Methods.” REFERENCES Ackerl, K., Atzmüller, A., and Grammer, K., The scent of fear, Neuroendocrinol. Lett., 2001, vol. 23, pp. 79–84. Amal, H., Ding, L., Liu, B.B., et al., The scent fingerprint of hepatocarcinoma: invitro metastasis prediction with volatile organic compounds (VOCs), Int. J. Nanomed., 2012, vol. 7, pp. 4135–4146. Anisimov, V.N., Ukraintseva, S.V., and Yashin, A.I., Cancer in rodents: does it tell us about jarcer in humans?, Nat. Rev. Cancer, 2005, vol. 5, pp. 807–819. Arakawa, H., Cruz, S., and Deak, T., From models to mechanisms: odorant communication as a key determinant of social behavior in rodents during illnessassociated BIOLOGY BULLETIN

Vol. 42

No. 3

2015

DETECTION OF VOLATILE ORGANIC COMPOUNDS ASSOCIATED states, Neurosci. Biobehav. Rev., 2011, vol. 35, pp. 1916– 1928. Boedeker, E., Friedel, G., and Walles, T., Sniffer dogs as part of a bimodal bionic research approach to develop a lung cancer screening, Interact. Cardiovasc. Thorac. Surg., 2012, vol. 14, pp. 511–515. Bransbury, A.J., Church, M.R.T., and Church, J.C.T., Olfactory detection of human bladder cancer by dogs: proof of principle study, BMJ, 2004, vol. 329, pp. 712–717. Cornu, J.N., CancalTassin, G., Ondet, V., et al., Olfactory detection of prostate cancer by dogs sniffing urine: a step forward to early diagnosis, Eur. Urol., 2011, vol. 59, pp. 197–201. Gordon, R.T., Schatz, C.B., Myers, L.J., et al., The use of canines in the detection of human cancers, J. Altern. Compl. Med., 2008, vol. 14, pp. 61–67. Hakim, G.M., Broza, Y.Y., Billan, S., et al., Detection of lung, breast, colorectal, and prostate cancers from exhaled breath using a single array of nanosensors, Br. J. Cancer, 2010, vol. 103, pp. 542–551. Hanai, Y., Shimono, K., Matsumura, K., et al., Urinary volatile compounds as biomarkers for lung cancer, Biosci. Biotechnol. Biochem., 2012a, vol. 76, pp. 679–684. Hanai, Y., Shimono, K., Oka, H., et al., Analysis of volatile organic compounds released from human lung cancer cells and from the urine of tumorbearing mice, Cancer Cell Int., 2012b, vol. 12, pp. 7–19. Havlicek, J.1. and Roberts, S.C., MHCcorrelated mate choice in humans: a review, Psychoneuroendocrinology, 2009, vol. 34, no. 4, pp. 497–512. Heth, G. and Todrank, J., Individual odour similarities across species parallel phylogenetic relationships in the S. ehrenbergi superspecies of molerats, Anim. Behav., 2000, vol. 60, pp. 789–795. Horvath, G., Järverud, G., Järverud, S., and Horváth, I., Human ovarian carcinomas detected by specific odor, Inte grative Cancer Therapies, 2008, vol. 7, pp. 76–80. Horváth, I., Lázár, Z., Gyulai, N., et al., Exhaled biomark ers in lung cancer, Eur. Respir. J., 2009, vol. 34, pp. 261– 275. Horvath, G., Andersson, H., and Paulsson, G., Character istic odour in the blood reveals ovarian carcinoma, BMC Cancer, 2010, vol. 10, p. 643. Hussain, S.A., Ferry, D.R., ElGazzaz, G., et al., Hepato cellular carcinoma, Ann. Oncol., 2001, vol. 12, pp. 161–172. Jemal, A., Bray, F., Center, M.M., et al., Global cancer sta tistics, CA Cancer J. Clin., 2011, vol. 61, pp. 69–90. Koczulla, R., Hattesohl, A., Biller, H., et al., Comparison of four identical electronic noses and three measurement setups, Pneumologie, 2011, vol. 65, pp. 465–470. Korenevskii, Yu., Examination of olfactory traces in inves tigative and judicial practice, Ros. Yustitsiya, 2000, no. 8, p. 29. Krutova, V.I., Sulimov, K.T., and Zinkevich, E.P., The time of appearance of individual odor in the ontogeny of the gray rat (Rattus norvegicus) according to dog training analysis, Sensorn. Sistemy, 1997, vol. 11, no. 3, pp. 340–345. Kuano, M., Mendez, E., and Furton, K.G., Development of headspace SPME method or analysis of volatile organic compounds present in human biological specimens, Anal. Bioanal. Chem., 2011, vol. 400, pp. 1817–1826. Kwak, J., Gallagher, M., Ozdener, M.H., et al., Volatile biomarkers from human melanoma cells, J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci., 2013, vol. 931, pp. 90–96. BIOLOGY BULLETIN

Vol. 42

No. 3

2015

245

Lazarevich, N.L., Cheremnova, O.A., Varga, E.V., et al., Progression of HCC in mice is associated with a downregu lation in the expression of hepatocyte nuclear factors, Hepatology, 2004, vol. 39, pp. 1038–1047. Matsumura, K., Opiekun, M., Oka, H., et al., Urinary vol atile compounds as biomarkers for lung cancer: a proof of principle study using odor signatures in mouse models of lung cancer, PLoS One, 2010, vol. 5, p. e8819. McCulloch, M., Jezierski, T., Broffman, M., et al., Diag nostic accuracy of canine scent detection in early and late stage lung and breast cancers, Interact. Cancer. Ther., 2006, vol. 5, pp. 30–39. O’Neill, H.J., Gordon, S.M., O’Neill, M.H., et al., A com puterized classification technique for screening for the pres ence of breath biomarkers in lung cancer, Clin. Chem., 1988, vol. 34, no. 8, pp. 1613–1618. Penn, D. and Potts, W.K., Chemical signals and parasite mediated sexual selection, Trends Ecol. Evol., 1998, vol. 13, pp. 391–396. Persaud, K. and Dodd, G., Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose, Nature, 1982, vol. 299, pp. 352–355. Qin, T., Liu, H., Song, Q., et al., The screening of volatile markers for hepatocellular carcinoma, Cancer Epidemiol. Biomark. Prev., 2010, vol. 19, pp. 2247–2253. Rivière, S., Challet, L., Fluegge, D., et al., Formyl peptide receptorlike proteins are a novel family of vomeronasal chemosensors, Nature, 2009, vol. 459, pp. 574–577. Rodionova, E.I., Minor, A.V., Sulimov, K.T., and Kogun, G.A., Dogs are able to recognize insect individuals by odour, Chem. Senses, 2009, vol. 34, no. 3, p. E64. Schoon, G.A.A., Scent identification lineups by dogs (Canis familiaris): experimental design and forensic appli cation, Appl. Anim. Behav. Sci., 1996, vol. 49, pp. 257–267. Schoon, A. and Haak, R., K9 Suspect Discrimination: Train ing and Practicing Scent Identification LineUps, Calgary, Alberta, Canada: Detselig Enterprises, 2002. Sokolov, V.E., Sulimov, K.T., and Krutova, V.I., Dog identi fication of individual odors in live activity traces of four spe cies of vertebrates, Izv. Akad. Nauk SSSR, Ser. Biol., 1990, no. 4, pp. 56–64. Sonoda, H., Kohnoe, S., Yamazato, T., et al., Colorectal cancer screening with odour material by canine scent detec tion, Gut, 2011, vol. 60, pp. 814–819. Tisch, U. and Haick, H., Nanomaterials for crossreactive sensor arrays, MRS Bull., 2010, vol. 35, p. 797. Walczak, M., Jezierskie, T., GóreckaBruzda, A., et al., Impact of individual training parameters and manner of taking breath odor samples on the reliability of canines as cancer screeners, J. Vet. Behav., 2012, vol. 7, pp. 283–294. Williams, M. and Johnston, J.M., Training and maintaining the performance of dogs (Canis familiaris) on an increasing number of odor discriminations in a controlled setting, Appl. Anim. Behav. Sci., 2002, vol. 78, pp. 55–65. Xue, R., Dong, L., Zhang, S., et al., Investigation of volatile biomarkers in liver cancer blood using solidphase microex traction and gas chromatography/mass spectrometry, Rapid Commun. Mass Spectrom., 2008, vol. 22, pp. 1181–1186. Yamazaki, K. and Beauchamp, G.K., Genetic basis for MHCdependent mate choice, Adv. Genet., 2007, vol. 59, pp. 129–145.

Translated by Ya. Atmanskikh