Immunophenotyping Predictive of Mycoplasma Infection in Patients with CFS

 

http://www.immed.org/publications/fatigue_illness/Nijs_ImmunoMycoJCFS.html

Journal of Chronic Fatigue Syndrome 2003; 11(2):51-70.


Nijs Jo 1, MSc; Coomans Danny 2, PhD; Nicolson Garth 3, PhD; De Becker Pascale 1, PhD; Christian Demanet 4, MD, PhD; De Meirleir Kenny 1, 5, MD, PhD

    1. Department of Human Physiology – Faculty of Physical Education and Physiotherapy – Vrije Universiteit Brussel (VUB), Belgium
    2. School of Mathematical and Physical Sciences - James Cook University, Australia
    3. Institute for Molecular Medicine - Huntington Beach, California, USA
    4. Division of Hematology and Immunology, Academic Hospital Vrije Universiteit Brussel, Belgium
    5. Chronic Fatigue Clinic - Vrije Universiteit Brussel (VUB), Belgium

Address correspondence to Jo Nijs, Vakgroep MFYS / Sportgeneeskunde, AZ-VUB KRO-gebouw –1, Laarbeeklaan 101, 1090 Brussel – Belgium, Tel: +32 2 477 46 04, fax: +32 2 477 46 07, e-mail: Jo.Njs@vub.ac.be

Keywords: chronic fatigue syndrome, Mycoplasma, immunity, immunofluorescence

ABSTRACT. An impaired immune system and opportunistic infections are considered important characteristics in the pathophysiology of Chronic Fatigue Syndrome (CFS). Using immunofluorescence we examined healthy subjects (N=35) and two subsets of CFS patients: those without evidence of Mycoplasma (N=55) and those with evidence of a Mycoplasma infection in their blood (N=131). Using monoclonal antibodies and forensic polymerase chain reaction for detection of M. hominis, M. fermentans, M. pneumoniae and M. penetrans we examined leukocytes in peripheral blood samples. Both patients groups presented with significantly elevated CD25+ (activated) cells as compared to healthy volunteers. CFS patients without evidence of Mycoplasma infection(s) had increased CD5+ B-cell counts. Stepwise discriminant analysis indicated the number of activated cells, number of memory CD4+ cells and percentage of suppressor T-cells (lower in Mycoplasma+ patients as compared with Mycoplasma- patients) as the discriminant variables. A classification tree, for predicting the presence of Mycoplasma species in CFS patients, was constructed. Taken together, these data confirm earlier reports on immune activation among CFS patients, but this does not appear to be specific for Mycoplasma-infected CFS patients. ©2003 by the Hayworth Press, Inc. All rights reserved.

INTRODUCTION

Chronic Fatigue Syndrome (CFS), or alternatively Chronic Fatigue Immune Dysfunction Syndrome, is a chronic debilitating disease with an unknown cause (1). In such patients Mycoplasma species infections may serve as primary or secondary etiological factors. Mycoplasma are prokaryotes containing circular DNA and some ribosomes (2), lacking a cell wall and certain cellular organelles. The high prevalence of Mycoplasma species infections in CFS has been discussed at length in the scientific literature (3-8). The presence of opportunistic infections in CFS suggests an impaired immune system in these patients. Indeed, a deregulated 2’5’ oligoadenylate synthetase ribonuclease antiviral pathway (9-11) and a reduced natural killer cell function (13-15) are two characteristics of impaired immunity in at least subsets of CFS patients. Although some Mycoplasma species are part of the normal human flora, others are capable of causing complex systemic infections in immunocompromised hosts, as seen in HIV-AIDS (16) and CFS (3-7).

The immunological properties of Mycoplasma species have been studied extensively. M. fermentans has been shown to affect the immune system via T- or B-cell activation and macrophage stimulation [reviewed in (2)]. A membrane-associated complement C3 -activating protein [M161Ag], from M. fermentans, efficiently promotes the production of nitric oxide [NO], interleukin 1ß [IL-1ß], tumour necrosis factor [TNF-ß], IL-10 and IL-6 in human peripheral blood monocytes (2,17-19). Activation of the complement system by C3 in M. fermentans-infected cells in turn stimulates the release of C5a chemotactic factor and enhances phagocytic activity. Kikkawa et al. (17) have demonstrated that the human complement system was able to clear M. fermentans from the surface of infected human cells, but was not able to prevent persisting low-grade infections in human tumour cell lines, probably due to rapid invasion of the infected cells into tissues.

There are some data available that describe host defence action against respiratory mycoplasmosis caused by M. pneumoniae and M. pulmonis. Both innate [mediated by alveolar macrophages] and humoral immunity appear to be of prime importance in the defence action against these Mycoplasma infections (20). Moreover, T-cell responses in respiratory Mycoplasma infections may result in exacerbation of lung lesions, whereas innate immunity is crucial in defence of the lungs and humoral immunity in preventing dissemination of infection to extrapulmonary sites (20). In accordance with these observations, a recent report provides an explanation for the characteristic chronic nature of Mycoplasma spp. infections. In this report M. pulmonis infection induced a transitional shift of the TH1 [T helper cells type 1] – TH2 balance in favour of humoral immunity, which is thought to allow the microorganism to survive (21). The exact role of cell-mediated immunity in Mycoplasma infections, however, remains to be established.

To our knowledge, the possible effects of Mycoplasma species on immune cells in CFS patients have not been examined. Therefore, we examined CFS patients with or without Mycoplasma infection[s] and healthy volunteers to see if certain immune patterns were characteristic of chronic infections in these patients.

METHODS

Sample
The study was conducted in Brussels, at a university-based outpatient clinic [Vrije Universiteit Brussel]. One hundred and eighty-six consecutive patients seeking care for prolonged fatigue as major complaint, and who complied with the 1994 CDC case definition (1), were enrolled. Thirty-five age matched healthy volunteers were recruited among university students, health care professionals and hospital employees and served as control subjects.

To fulfil the CDC criteria for CFS clinically evaluated, unexplained, persistent or relapsing chronic fatigue that is of new or definite onset, should result in a substantial reduction in previous levels of occupational, educational, social, or personal activities (1). Additionally, at least four of the following symptoms must have persisted or recurred during 6 or more consecutive months and must have not predated the fatigue: impairment in short-term memory or concentration, tender cervical or axillary lymph nodes, muscle pain, multi-joint pain, headache, unrefreshing sleep and post-exertional malaise >24 hours (1). Any active medical condition that may explain the presence of chronic fatigue was excluded from the diagnosis of CFS. All subjects underwent an extensive medical evaluation, consisting of a standard physical examination, medical history, exercise capacity test and routine laboratory tests. The laboratory tests included a complete blood cell count, determination of the erythrocyte sedimentation rate, serum electrolyte panel, measures of renal, hepatic and thyroid function, as well as rheumatic and viral screens. When judged necessary, a structured psychiatric interview was performed. In a number of cases further neurological, gynaecological, endocrine, cardiac and/or gastrointestinal evaluations were performed. The medical records were also reviewed to determine if patients suffered from organic or psychiatric illnesses that could explain their signs and symptoms.

Control subjects were questioned about their health status. They had to be free of disease at least three months prior to data collection. All subjects were excluded if they were < 18 or ? 60 years of age, or if they had received antibiotics two months prior to phlebotomy. All patients and controls were Caucasian.

Immunological Data
Anticoagulated blood [EDTA] was collected between 9:00 and 11:00 A.M. and used for white blood cell enumeration, differential counts [Celldyn 4000, Abbott Laboratories, Abbott Park, IL 60064, USA] and flow cytometric studies. Lymphocyte populations were analysed with dual colour direct immunofluorescence on a EPICS® xl flow cytometer [Coulter, Miami, Florida, USA], with aid of the System I TM computer software. One hundred µ1 of whole blood was incubated with the appropriate combination of monoclonal antibodies for 25 minutes at 4°C. Red blood cells were lysed using lysis buffer [Becton Dickinson] for 7 minutes, spun down and washed once with 2 ml phosphate buffered saline [PBS]. Resuspension was immediately followed by cell analysis. Commercially available [Becton-Dickinson] phycoerythrin [PE] or fluorescein isothiocyanate [FITC] labelled monoclonal antibodies were used [Table 1]. Estimates of absolute numbers of lymphocyte subsets were determined by multiplying peripheral lymphocyte counts by the percentage of each surface marker.

Mycoplasma detection: Forensic PCR
Collection of Blood
Subjects’ blood was collected between 9.00 and 11.00 A.M. at the Chronic Fatigue Clinic of the academic hospital of the Vrije Universiteit Brussel. Blood was collected in EDTA-containing tubes and immediately brought to ice bath temperature and flash frozen as described previously (3,7). Samples were shipped with dry ice by air courier to the Institute for Molecular Medicine for analysis. All blood samples were blinded. Whole blood [50 µl] was used for preparation of DNA using Chelex [Biorad, Hercules, USA] as follows. Blood cells were lysed with nano-pure water [1.3 ml] at room temperature for 30 min. After centrifugation at 13000 x g for 2 minutes, the supernatants were discarded. Chelex solution [200 µl] was added, and the samples were incubated at 56°C and at 100°C for 15 minutes each. Aliquots from the centrifuged samples were used immediately for PCR or stored at –70°C until use. Multiple Mycoplasma tests were performed on all patients.

Amplification of Gene Sequences
Amplification of the target gene sequences (3) was performed in a total volume of 50 µl PCR buffer [10 mM Tris-HCl, 50 mM KCl, pH 9] containing 0.1% Triton X-100, 200 µm each of dATP, dTTP, dGTP, dCTP, 100 pmol of each primer, and 0.5 – 1 ?g of chromosomal DNA. Purified Mycoplasma DNA [0.1 – 1 ng of DNA] was used as a positive control for amplification. Additional primers were used as necessary to confirm the species. The amplification was carried out for 40 cycles with denaturing at 94°C and annealing at 60°C [genus-specific primers and M. penetrans] or 55°C [M. pneumoniae, M. hominis, M. fermentans]. Extension temperature was 72°C in all cases. Finally, product extension was performed at 72°C for 10 min. Negative and positive controls were present in each experimental run.

Southern Blot Confirmation
The amplified samples were run on a 1% agarose gel containing 5 ml/100 ml of ethidium bromide in TAE buffer [0.04 M Tris-Acetate, 0.001 M EDTA, pH 8.0]. After denaturation and neutralization, Southern blotting was performed as follows. The PCR product was transferred to a Nytran membrane. After transfer, UV cross-linking was performed. Membranes were prehybridized with hybridisation buffer consisting of 1x Denhardt’s solution and 1 mg/ml salmon sperm DNA as blocking reagent. Membranes were then hybridized with 32P-labeled internal probe [107 cpm per bag]. After hybrization and washing to remove unbounded probe, the membranes were exposed to autoradiography film for 1-2 days at –70°C.

Statistics
Subjects’ age characteristics were analysed using descriptive statistics and Bonferroni multiple comparisons. Sex differences were assessed using binomial nonparametric testing and Chi-Square tests. Descriptive statistics were computed for all variables among different groups [Mycoplasma-infected CFS patients, CFS patients without Mycoplasma infections and healthy volunteers]. Uni-variant group differences were assessed on untransformed data using one-way ANOVA [analysis of variances], followed by pair wise multi-comparison Bonferroni test [t-test]. For a few variables heterogeneity of variances was observed [Levene’s test]; therefore, we decided to double-check the ANOVA-results by the non-parametric Kruskal-Wallice test. The significance level of the different tests was at 0.05. The data were processed using SPSS 10.0? for Windows [Prentice Hall]. Discriminate variables were determined using stepwise linear discriminate analysis of the differences between the 3 groups. At each step, the variable that minimizes the overall Wilks’ Lambda is entered [the maximum significance of F to enter is 0.20; the minimum significance of F to remove is 0.25].

Classification trees [CT], originally invented by Brieman et al. (22) classify the groups of a single discrete variable [such as species of Mycoplasma] using explanatory variables, which may be numeric or categorical. This is done by establishing a tree structure, which divides the data into mutually exclusive groups, that have similar values for the response variable. The entire data are represented by a single node at the top of the tree, and the tree is grown by repeated binary splitting of the data. Each split defines a simple rule and forms two nodes. Rules are usually based on a single explanatory variable, and the form of the rule depends on the type of explanatory variable. For numeric explanatory variables, a split is defined by values less than and greater than a certain value. The terminal nodes [i.e. unsplit nodes] represent the groups of data formed by the tree and are also called the leaves of the tree.

Splits are [generally] chosen to maximise the homogeneity [minimise the impurity] of the resulting two nodes. Thus for each split a search is made through all possible splits of all explanatory variables and the best is selected. The impurity of a node can be defined in many ways. For classification trees the impurity is the proportion of cases from other groups present in the node together with predominant group. Total purity means that a node contains only cases from one group. The splitting procedure is continued until an over large tree is grown, which is then pruned back to the desired size. The size may depend on the objective of the analysis: be it exploration, description or prediction. For prediction we select the size that is the most accurate predictor. Often cross validation is used to select the tree size, with the chosen tree having the lowest misclassification rate. This tree can be thought of as the best predictive tree in the sense that on average it should give the most accurate predictions of all possible trees obtainable by pruning back the over large tree.


RESULTS

Sample
The demographic data are presented in Table 2. Mean age characteristics ranged between 34 and 36 years for the three groups. Statistical analysis revealed no significant differences in mean age between the three groups. No sex differences between the two patients’ groups [CFS patients with or without Mycoplasma spp. Detected] were found. Significantly more female subjects were present in the control sample as compared to both patients groups.

Immunological Data
One-way ANOVA analysis [Table 3] revealed significant differences among groups in the number of CD25+ cells [activated cells] [p=0.007] in peripheral blood. Using Bonferroni analysis it was concluded that the number of activated cells was significantly elevated in both CFS-groups [Mycoplasma-positive as well as –negative CFS patients] compared to healthy volunteers [Table 4 summarises the statistically significant differences]. Bonferroni analysis can be used to analyse this variable, because Levene’s test suggested homogeneity of variances for the number of activated cells [data not shown]. No differences between the two patients’ groups were observed concerning this variable. Kruskal-Wallice test confirmed the differences in CD25+ count and revealed significant differences in the number [p=0.009] as well as the percentage [p=0.002] of CD5+ B-cells [CD19+CD5+] [Table 5]. These observations were confirmed by Tanhame’s pair wise multi-comparisons, indicating significantly increased CD19+CD5+ cells between CFS patients without evidence of Mycoplasma infection[s] and healthy volunteers [p<0.001] [Table 4].

Stepwise linear discriminant analysis of the differences between the three groups indicated the number of activated cells was the primary discriminate variable [Wilks’?=0.953, F=5.254, p=0.006], followed by the number of memory CD4+ cells [CD4+CD45RA-] [Wilks’?=0.922, F=4.411, p=0.002] and the percentage of suppressor T-cells [Wilks’?=0.908, F=3.507, p=0.002][Table 6]. In comparison to healthy volunteers, this sample of CFS patients had elevated numbers of activated cells and lower numbers of memory CD4+ cells. Additionally, among these CFS patients, Mycoplasma spp. infected patients presented with lower numbers of suppressor cells compared to patients without Mycoplasma spp. detection.

Case Report Using the Classification Tree
The classification tree for predicting the presence of a Mycoplasma spp. infection in individual CFS patients using immunological data is presented in Figure 1; corresponding probabilities of the root endings are listed in Table 7. It is impossible to explain the entire classification tree in this manuscript; however, to demonstrate the usefulness of this approach we can discuss the classification tree using the case of a patient that did not participate in the trial. An Endocrinologist referred a 54-year old woman to the Chronic Fatigue Clinic of the Vrije Universiteit Brussel [V.U.B.]. Her current health problems had started 15 years ago with an autoimmune disorder of the thyroid gland. At the time she visited the fatigue clinic this patient complained of severe fatigue, especially following exercise, and many other symptoms [myalgia, arthralgia, non-refreshing sleep, neurocognitive impairments, among other complaints]. Blood was collected for routine laboratory tests, for immunophenotyping and for Mycoplasma spp. PCR detection. According to the classification tree, the CD5+ B-cell [#=77] count was checked first and exceeded the threshold value of 75.04. We were therefore guided into the right part of the tree, with the number of CD3+CD16CD56+ [subset cytotoxic T-cells; #=103] being the next discriminating variable. Because this parameter did not fulfil the tree equation, the cytotoxic T-cells count [CD8+CD11b-] was screened and it exceeded the tree threshold value [step 3: 515>283.5]. Next, the B-cell count fulfilled the equation [335<587] as well as the percentage virgin CD4-cells in step 5. Using this method we were quite sure that this was a CFS patient, because the next step consists of a binary split into two root endings suggesting CFS. The patient’s immune profile revealed a slightly increased number of activated T-cells [232 exceeded the threshold value in the tree and exceeded the reference range used in the laboratory - see Table 3]. According to the classification tree, this patient was judged to have a 100% probability [see Table 7 for the probability data for each root ending] for being a Mycoplasma-infected CFS patient. Indeed, this patient fulfilled both the 1988 (23) and the 1994 (1) CDC case definitions for CFS and was subsequently diagnosed as having a M. hominis infection. Interestingly, the Mycoplasma PCR was performed at a different laboratory [RED laboratories Zellik, Belgium] from the analyses used for the construction of the classification tree.


DISCUSSION

A significant increase in activated cells [p = 0.007 in the one-way ANOVA] was observed in Mycoplasma-positive and –negative CFS patients compared to healthy controls, and the stepwise linear discriminant analysis indicated this as the primary discriminant variable of the differences between the patient and control groups. These data are consistent with earlier reports on immune activation in patients with CFS (24-26). To our knowledge, however, data providing evidence of immune activation using CD25+ cells is currently unavailable. Both the CFS patients with or without evidence of a Mycoplasma spp. infection presented with elevated CD25+ cells, suggesting an immune response against another pathogen in the Mycoplasma-negative patients. Immune activation appears to be less controlled in Mycoplasma-infected CFS patients in comparison to Mycoplasma negative CFS patients, because the former group presented with a lower percentage of suppressor cells. Both increased (27) and reduced numbers (28) of CD8+CD11b+ cells have been reported in CFS compared to healthy controls, but the lack of sub-grouping in these research reports may account for this discrepancy.

The Mycoplasma spp. PCR negative CFS patients presented with significantly elevated numbers of CD5+ B-cells compared to healthy volunteers [Tanhame’s pair-wise multi-comparisons: p<0.001], which was not observed in Mycoplasma spp.-infected CFS patients. This obviously explains why this marker is used on top in the classification tree. The CD5+ surface marker was originally found only on T-cells; its function on B-cells is uncertain. It may be involved in the regulation of the B-cell activation and is mainly displayed on B-1 cells, which are biased towards auto antigens and bacteria (29). Mycoplasma have been reported to activate B-cells (2,20), and this may account for the increased but insignificant increase in CD19+CD5+ cell numbers observed in these Mycoplasma–positive CFS patients compared with healthy controls. Tirelli et al. (30) found increased numbers of CD19+ cells in CFS patients compared to asymptomatic volunteers. Interestingly, in 11 out of 30 CFS patients examined the increase in B-lymphocytes was sustained by the expansion of the CD5+ subset of B-cells (30). Klimas et al (31) reported additional evidence for a high proportion of CD5+ bearing B-cells in CFS patients, a finding that has been disputed by others (32).

The data presented in this manuscript suggest an important discriminating role for memory CD4+ cells [Wilks’ ?=0.922, F=4.411, p=0.002]. The lower numbers of memory cells observed in these CFS patients compared to healthy controls, however, may not be completely valid because of the high number of hospital employees in the control group.

In conclusion, analysing these data using ‘classical’ statistics did not reveal major differences in immunological profiles between Mycoplasma-positive and -negative CFS patients. One could conclude that immunophenotyping is not indicative of Mycoplasma infections in CFS patients, or these patients could have other chronic bacterial or viral infections that cause differences in immune parameters. We therefore chose to reanalyse these data using a recently developed statistical method [the construction of a classification tree]. The classification tree [Figure 1 and Table 7] enables one to use immunophenotyping for prediction of Mycoplasma infections in CFS patients. In the presented case report, the classification tree was able to predict the presence of a Mycoplasma infection. A Mycoplasma spp. PCR was still required because the classification tree is unable to distinguish between different Mycoplasma spp., which has important therapeutic implications. The classification tree should not be used to diagnose CFS, as suggested in the case study. It can only be utilised as a differential diagnostic aid for predicting a Mycoplasma infection in case of CFS.

These data should be interpreted with caution because no attempt was made to monitor Mycoplasma species in the control group and a high percentage of health care workers were used for the control group. The main goal of this study, however, was to examine subsets of CFS to see if differences exist in various immunological and infection parameters. Moreover, the prevalence of Mycoplasma infections among asymptomatic individuals ranges between 5 and 15% [reviewed in (6)], while in a previous study among European citizens 2 out of 36 [5.6%] healthy control subjects were infected by one species of Mycoplasma (33). Additionally, the control subjects differed in sex distribution as compared to both patient groups. This may well have biased these results. On the other hand, this classification tree should aid in predicting the presence of Mycoplasma infections in CFS patients rather than differentiating between patients and controls. Finally, this sample was not randomly selected. A sample consisting of consecutive patients visiting a clinic however is more likely to represent routine clinical practice. Although much work was done to validate the methods used for the construction of the classification tree, the use of this technique still needs to be validated in clinical situations [in accordance with the presented case report].

In this research report, minor differences in immune cells were observed between healthy volunteers and Mycoplasma-positive and –negative CFS patients. The number of CD25+ activated cells, number of memory CD4+ cells and percentage of suppressor T-cells were identified as the primary discriminate variables. Nevertheless, the construction of this classification tree should enable physicians to more accurately predict the presence of a Mycoplasma infection in CFS more accurately and consequently reduce the number and costs of laboratory tests.

REFERENCES

    1. Fukuda K, Strauss SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A, and the International Chronic Fatigue Syndrome Study Group. The Chronic Fatigue Syndrome, a comprehensive approach to its definition and study. Ann Intern Med 1994; 121:953-959.
    2. Rawadi G. Mycoplasma fermentans interaction with monocytes/macrophages:molecular basis. Microbes Infect 2000;2:955-964.
    3. Nasralla M, Haier J, Nicsolosn GL. Multiple Mycoplasmal infections detected in blood of patients with Chronic Fatigue Syndrome and / or Fibromyalgia. Eur J Clin Microbiol Infect Dis 1999;18:859-865.
    4. Vojdani A, Choppa PC, Tagle C, Andrin R, Samimi B, Lapp CW. Detection of Mycoplasma genus and Mycoplasma fermentans by PCR in patients with Chronic Fatigue Syndrome. FEMS Immunol Med Microbiol 1998;22:355-365.
    5. Nasralla MY, Haier J, Nicolson NL, Nicolson GL. Examination of Mycoplasmas in blood of 565 chronic illness patients by polymerase chain reaction. In J Med Biol Environ 2000;28:15-23.
    6. Nicolson GL, Nasralla MY, Franco AR, De Meirleir K, Nicolson NL, Ngwenya R, Haier J. Role of Mycoplasmal infections in Fatigue Illnesses: Chronic Fatigue and Fibromylagia Syndromes, Gulf War Illness and Rheumatoid Arthritis. J Chronic Fatigue Syndr 2000;6,23-39.
    7. Nicolson GL, Nasralla M, Haier J, Nicolson NL. Diagnosis and treatment of chronic Mycoplasmal infections in fibromylagia and chronic fatigue syndromes: relationship to gulf war illness. Biomed Ther 1998;16:266-271.
    8. Nicolson GL, Nicolson NL. Diagnosis and treatment of Mycoplamsmal infections in Persian Gulf War Illness - CFIDS patients. International Journal of Occupational Medicine, Immunology and Toxicology 1996;5:67-78.
    9. Suhadolnik RJ, Reichenbach NL, Hitzges P, Sobol RW, Peterson DL, Henry B, Ablashi DV, Müller WEG, Schröder HC, Carter WA, Strayer DR. Upregulation of the 2-5A synthetase/Rnase L antiviral pathway associated with chronic fatigue syndrome. Clin Infect Dis 1994;18:S96-104.
    10. De Meirleir K, Bisbal C, Campine I, De Becker P, Salehzada T, Demettre E, Lebleu B. A 37 kDa 2-5A binding protein as a potential biochemical marker for Chronic Fatigue Syndrome. Am J Med 2000;108:99-105.
    11. Suhadolnik RJ, Peterson DL, O’Brien K, Cheney PR, Herst CVT, Reichenbach NL, Kon N, Horvath SE, Iacono KT, Adelson ME, De Meirleir K, De Becker P, Charubala R, Pfleiderer W. Biochemical Evidence for a novel low molecular weight 2-5A-dependent RNase L in chronic fatigue syndrome. J Interferon Cytokine Res 1997;17:377-385.
    12. Baseman JB, Tully JG. Mycoplasmas: sophisticated, re-emerging and burdened by their notoriety. Emerg Infect Dis 1997;3:21-3.
    13. Zhang Q, Zhou X-D, Denny T, Ottenweller JE, Lange G, LaManca JJ, Lavietes MH, Pollet C, Gause WC, Natelson BH. Changes in immune parameters seen in gulf war veterans but not in civilians with Chronic Fatigue Syndrome. Clin Diagn Lab Immunol 1999; 6:6-13.
    14. Levine PH, Whiteside TL, Friberg D, Bryant J, Colclough G, Herbertman RB. Dysfunction of natural killer cell activity in a family with chronic fatigue syndrome. Clin Immunol Immunopathol 1998;88:99-104.
    15. Whiteside TL, Friberg D. Natural killer cells and natural killer cell activity in chronic fatigue syndrome. Am J Med 1998;105,27S-34S.
    16. Baum SC. Mycoplasma diseases. In: Principles and practice of infectious diseases. (Mandell, G.L., Bennett, J.E., Dolin, R., Eds.) 1995;pp.1701-1712. Churchill Livingstone, New York.
    17. Kikkawa S, Matsumoto M, Sasaki T, Nishiguchi M, Tanaka K, Toyoshima K, Seya T. Complement activation in Mycoplasma fermentans-induced Mycoplasma clearence from infected cells: probing of the organism with monoclonal antibodies against M161Ag. Infect Immun 2000;68:1672-1680.
    18. Nishiguchi M, Matsumoto M, Takao T, Hoshino M, Shimonishi Y, Tsuji S, Begum NA, Takeuchi O, Akira S, Toyoshima K, Seya T. Mycoplasma fermentans lipoprotein M161Ag-induced cell activation is mediated by toll-like receptor 2: role of N-terminal hydrophobic portion in its multiple functions. J Immunol 2001;166:2610-2616.
    19. Matsumoto M, Seya T. M161Ag is a potent cytokine inducer with complement activating function (review). Int J Mol Med 1999;3:291-5.
    20. Cartner SC, Lindsey JR, Gibbs-Erwin J, Cassell GH, Simecka JW. Roles of innate and adaptive immunity in respiratory Mycoplasmosis. Infect Immun 1998;66:3485-3491.
    21. Romero-Rojas A, Ponce-Hernandez C, Mendoza SE, Reyes-Esparza JA, Estrada-Parra S, Hadden JW. Immunomodulatory properties of Mycoplasma pulmonis. II Studies on the mechanisms of immunomodulation. Int Immunopharmacol 2001;1:1689-97.
    22. Brieman L, Friedman JH, Ohlsen R, Stone CJ. Classification and Regression Trees. Belmont California: Wadsworth International Group. 1984.
    23. Holmes GP, Kaplan JE, Gantz NM, Komaroff AL, Schonberger LB, Straus SE, Jones JF, Dubois RE, Cunningham-Rundles C, Pahwa S, Tosato G, Zegans LS, Purtilo DT, Brown N, Schooley RT, Brus I. Chronic Fatigue Syndrome, a working case definition. Ann Intern Med 1988;108:387-389.
    24. Levy JA. Viral studies of Chronic Fatigue Syndrome. Clin Infect Dis 1994;18;S117-20.
    25. Barker E, Fujimura SF, Fadem MB, Landay AL, Levy JA. Immunologic abnormalities associated with chronic fatigue syndrome. Clin Infect Dis 1994;18:S136-41.
    26. Hassan IS, Bannister BA, Akbar A, Weir W, Bofill M. A study of the immunology of the chronic fatigue syndrome: correlation of immunologic parameters to health dysfunction. Clin Immunol Immunopathol 1998;87:60-7.
    27. Peakman M, Deale A, Field R, Mahalingham M, Wessely S. Clinical improvement in chronic fatigue syndrome is not associated with lymphocyte subsets or function or activation. Clin Immunol Immunopathol 1997;82:83-91.
    28. Swanink CM, Vercoulen JH, Galama JM, Roos MT, Meyaard L, van der Ven-Jongekrijg J, de Nijs R, Bleijenberg G, Fennis JF, Miedema F, van der Meer JW. Lymphocyte subsets, apoptosis, and cytokines in patients with chronic fatigue syndrome. J Infect Dis 1996;173:460-3.
    29. Roitt I, Brostoff J, Male D. Cells involved in the immune response. In: Immunology: Fifth edition. (Roitt, I., Brostoff, J., Male, D., Eds.),1998;p19. Mosby, London.
    30. Tirelli U, Marotta G, Improta S, Pinto A. Immunological abnormailities in patients with Chronic Fatigue Syndrome. Scand J Immunol 1994;40:601-8.
    31. Klimas N, Salvato F, Morgan R, Fletcher MA. Immunologic abnormailities in chronic fatigue syndrome. J Clin Microbiol 1990;28:1403-1410.
    32. Landay AL, Jessop C, Lennette ET, Levy JA. Chronic Fatigue Syndrome: clinical condition associated with immune activation. Lancet 1991;338:707-712.
    33. Nijs J, Nicolson GL, De Becker P, Coomans D, De Meirleir K. High prevalence of Mycoplasma infections among European Chronic Fatigue Syndrome patients – Examination of four Mycoplasma species in blood of Chronic Fatigue Syndrome patients. FEMS Immunology and Medical Microbiology; 2002; 34:209-214.