Microbiology and resistance in urogenital tuberculosis
N. M. Francisco 2
G. Müller 1
O. Braissant 1
1 alta uro AG, Merian Iselin Klinik, Center of Biomechanics & Calorimetry (COB), University Basel, Basel, Switzerland
2 Department of Pathology, Institute of Infectious Disease and Molecular Medicine, University of Cape Town, South Africa
Mycobacteria are less common than other infections but still represent a significant threat to patients. Their slow growing nature and the late appearance of symptoms of urogenital mycobacterial infections often result in a delayed diagnostic and more severe consequences. UGTB is mostly linked to an active or inactive lung tuberculosis, as a consequence a similar immune response is expected. Unfortunately, compared to TB, very little is known on specific response to urinary tract infection caused by MOTT, still knowledge from BCG instillation for bladder cancer treatment suggest a similar process as well. With respect to treatment and drug resistance, considering the late onset of symptoms, the potentially large populations of mycobacteria in some lesion and their mutation rate, monotherapy should be avoided. Such monotherapy would mostly lead to the emergence of resistant subpopulations. In addition, some studies focusing on persisters emphasize that at least PZA and RIF should be included in the treatment regimen as those drugs have shown some activity on persisters.
Summary of recommendations
Although very little data exist with respect to the diagnostic and treatment of M. tuberculosis and other mycobacterial urogenital infections a few recommendations can be considered useful:
- Because of the late appearance of some symptoms, patient showing repeated sterile pyuria or other symptoms (if any) should be screened for slow growing mycobacteria often missed in regular urine culture. Upon positive mycobacterial culture DST should be performed.
- Considering the potentially high population in some lesion and the mutation rate for drug resistance, monotherapy should be avoided for M. tuberculosis and other mycobacterial urogenital infections.
- As persisters are of major concern during the usually long treatment course of TB and MOTT infections, PZA and RIF should be included (if possible) in the regimen as those 2 drugs have shown some efficacy against such persisters.
1 A brief introduction to mycobacteria
Mycobacteria are slow growing microorganisms with generation time of a few hours to a few days on appropriate microbiological medium , . They are conventionally divided in tuberculous mycobacteria and mycobacteria other than tuberculosis (MOTT). The first ones are mycobacteria able to cause tuberculosis (or leprosy) as the latter are mycobacteria unable to cause tuberculosis (even if they can cause other related lung disease). The diagnosis of mycobacteria is difficult because of their slow growing nature. Indeed mycobacteria are also divided based on their growth rate. The slow growing mycobacteria are expected to form a colony on Lowenstein-Jensen medium in more than 7 days (typically 10–28 ) as fast growing mycobacteria are expected to produce a colony in less than 7 days. Still a culture for diagnosis might take up to 61 days to become positive (positive MGIT sample, or colony appearing on Lowenstein Jensen medium) , , , . Most mycobacteria except for those appearing to be obligate pathogens (M. tuberculosis, M. leprae) are commonly found in soils, dust and water bodies. In addition some MOTT such as M. smegmatis are common inhabitants of the genitourinary tract , , , . Mycobacteria are responsible for a significant number of urinary tract and urogenital infections. Among mycobacterial infection, M. tuberculosis is the main pathogen encountered. However, other species of mycobacteria can represent up to 35% of isolates (Table 1). These mycobacteria are often not linked to an active or inactive lung TB.
|M. abscessus||M. avium intracellulare|
|M. gordonae||M. bovis|
|M. fortuitum||M. smegmatis*|
|M. kansasii||M. phlei†|
|M. marinum||M. scrofulaceum|
|M. xenopi||M. terrae|
|M. chelonae||M. simiae|
|M. celatum||M. haemophilum|
|M. gastrii||M. xenopi|
|*Considered not pathogenic
†Rarely causes diseases
Urinary and urogenital mycobacterial infections are often diagnosed quite late as symptoms are usually not specific (see following sections/chapters) , . Especially in urinary tuberculosis white blood cell count in urine is quite low compared to infection by MOTT . This is also consistent with the physiology of mycobacteria in urine. Indeed mycobacteria have been shown to survive in human urine . But, urine is a potent growth media for many microbes and mycobacteria can proliferate and thrive in it. Indeed mycobacteria have been shown to be able to growth in urine or in urine added with serum (growth conditions that mycobacteria could easily encounter in urinary tract, with or without lesions). Such growth took place at a slow rate  with doubling time as long as 99 hours. Indeed, a slow growth is consistent with the late appearance of symptoms as well. If several studies have focused on urinary tuberculosis, UTI caused by other mycobacteria have been much neglected in the last years, therefore very few data are available.
2 The case of urogenital tuberculosis (UGTB)
In terms of number, 10.4 million new cases of tuberculosis and 1.4 million fatalities were attributed to tuberculosis in 2015 . In addition to the 1.4 million people killed by tuberculosis, 400,000 died due to co-infection with HIV , . Data for Germany suggest that extrapulmonary tuberculosis (EPTB) represent ca 20.7% of cases of tuberculosis and among those cases 12% are UGTB (i.e., 2.5% of the total cases of TB). However, urinary tuberculosis is often linked to an active or inactive lung tuberculosis ,  thus emphasizing that the primary entry route of infection are the lungs. On the contrary, but not surprisingly, urinary tract infections caused by other mycobacteria is not linked to active or inactive TB (diagnosed by chest X-ray ).
3 Immune response to urogenital tuberculosis (UGTB)
As UGTB is mostly linked to an active or inactive lung tuberculosis, one can expect that immune response is mostly similar. In tuberculosis immune response acts as follows: Following interaction between the pattern recognition receptors (PRRs) and M. tuberculosis ligands, the microorganisms are phagocytized by antigen presenting cells (APCs). Upon phagocytosis, APCs migrate to the draining lymph nodes, initiating T-cell mediated immunity by priming naïve T lymphocytes. T-cell mediated immunity develops after two to three weeks of infection , , . The phagosome inside the cells interacts with secreted proteins and developed to an endosome. The endosome processes the mycobacteria and presents fragments to the major histocompatibility complex (MHC) class II, which in turn will present the peptides on the APC surface to the T cell receptor (TCR) of CD4+ T helper 1 cells. M. tuberculosis may also be killed by CD8+ T cells, whereby mycobacteria in the phagosome are processed by protease enzyme and transported to the endoplasmic reticulum (ER) by transporter associated with antigen processing (TAP). The peptide is thereafter loaded and transported to the Golgi apparatus, whereby it is presented by the MHC I to the TRC on the CD8+ T cells, which may exhibit cytotoxic effect against M. tuberculosis-infected cells , . The CD8+ T lymphocyte cells can directly kill M. tuberculosis using granulysin , . During APC- and T-cell interaction, various pro-inflammatory cytokines such as IL-6, IL-12, TNF and interferon-gamma (IFN-γ) are produced. These cytokines play a central role by inducing macrophage activation and inducible isoform of nitric oxide synthetase (iNOS) expression . Also, they enhance surface expression of MHC class II molecules and increase secretion of inflammatory mediators. Besides APCs, CD4+ Th1 cells and CD8+ T cells, other immune cells such as B cells, NK cells, neutrophils and regulatory T cells accumulate at the infectious focus , , , . Due to a high production of IL-12 by APC, the immune response is largely polarized towards a Th1 type. At the infection side, the effector Th1 cells undergo functional maturation  and increase their production of effector cytokines and chemokines. These two markers attract new immune cells, amplifying local inflammatory and promote the formation of granuloma. It is presumed that the formation of granuloma represents a host strategy to contain the M. tuberculosis infection and limit dissemination of pathogen. However, individuals with latent TB infection and active TB patients both develop granulomatous response.
Although a lot is known on the immune response to lung TB, very little (not to say anything) is known on specific response to urinary tract infection caused by M. tuberculosis or MOTT. Studies on the use of BCG instillation for bladder cancer suggest a similar process , . Still some variations have been noted in the proinflammatory cytokine profile in active kidney tuberculosis patients .
4 The problem of antimycobacterial resistance
Due to their extremely thick and multi-layered hydrophobic cell wall, mycobacteria have an intrinsic resistance to some antibiotics that cannot penetrate such permeation barrier. In addition mycobacteria ,  and in particular M. tuberculosis developed many ways to counteract the effect of drugs including efflux pumps, modified targets, as well as overexpression of the drug target. Also mycobacteria produce efficient beta lactamases and other drug inactivating enzymes , . Moreover, rate of emergence of resistance to antimycobacterial can be quite high. For M. tuberculosis, mutation rate leading to resistance against rifampin, isoniazid, streptomycin and ethambutol are 3.32x10-9, 2.56x10-8, 2.29x10-8, and 1.0x10-7 (expressed in mutation/bacterium/cell division), respectively. With an approximate number of 108 mycobacteria per lesion, it appears that monotherapy will surely result in the appearance of resistance , , . This situation can be observed in lung cavities but also in kidney cavities thus advocating for multiple drug therapy. Also less data are available, similar finding were made with MOTT with mutation rate ranging from 10-5 to 10-9. Among urinary tract pathogens M. fortuitum and M. avium were show to have the highest mutation rates for isoniazid and streptomycin, with 10-6 and 10-5 respectively . Thus showing that such pathogens should also be considered carefully when initiating a treatment. This also emphasize that monotherapy should be avoided with MOTT as well in the context of urogenital infections with cavities containing large populations of mycobacteria.
From a more clinical point of view, delayed drug susceptibility results, inappropriate use of antimycobacterial compounds, interrupted treatment (due to drug shortage a state level, non compliance, or poor access from patient to drug caused by poverty), and presence of counterfeit drug lead to the appearance of these acquired resistance . Further spread of the resistant strains combined with those bad practices lead to the appearance multiresistant TB strains. In 2015 MDR strains accounted for 4.6% of total TB cases (i.e., 480,000 case of MDR TB). Of these, 480,000 were primary infection/transmission of people having no previous TB history (i.e., likely transmission of an MDR strain). Only 21% (roughly 100,000) of these cases were found in individuals who received a successful TB treatments (i.e., likely acquired resistance link to previous drug treatment) , . Finally it must be noted that within different countries the proportion of XDR TB within the MDR TB ranges from <1 to 11% (laboratory confirmed cases) and poses a major threat, especially considering that laboratory confirmed cases are probably highly underestimated . As a result WHO considers that early initiation of appropriate drug treatment as well as avoiding transmission or acquisition of drug resistance is of critical importance .
5 Persistence in mycobacteria
Persisters are usually defined as a genetically identical subpopulation that is non-replicating or slow-growing and can survive bactericidal antibotics. They have a uninheritable phenotypic resistance (or tolerance) to antibiotics, but they daughter cells remains fully susceptible . Persistence in M. tuberculosis and other mycobacteria, seems can be induced by many different factors such as pH, oxygen concentration, starvation for example. Indeed, in the urinary tract the nutrient, pH and oxygen concentrations are factors which can vary a lot and that could induce persistence. Still recent research show that the traditional view of persisters being non-replicating or slow growing might be wrong. Furthermore, persistence when mycobacteria are exposed to INH might be driven by pulsing KatG expression that activates INH (that is a prodrug) in still dividing microorganisms (in this study M. smegmatis). Therefore low-frequency pulsing of KatG might be beneficial and promote persistent lineages of mycobacteria . Still mechanisms for other drug might be very different  . Overall this emphasize that persisters should be considered a threat and must be taken into account when choosing a drug regimen. Indeed, this suggest again that monotherapy should be avoided and that regimen should include at least PZA and RIF that are the two drugs which have shown some efficacy against persisters. To our knowledge no studies have been conducted to evaluated the role of persisters in urogenital infection by M. tuberculosis or other mycobacteria.
Many aspects of the biology and pathology of mycobacteria (tuberculosis but also MOTT) and their role in UGTB have not been studied in detail yet. As such gap in knowledge impairs optimal treatment and patient management further study in the field of UGTB is indispensable. Also considering their rising importance MOTT should also be better studied. Finally, mechanisms of persistence are also crucial and need to be understood as well.
O.B. work is currently supported by a grant from the Merian Iselin Stiftung (Basel, Switzerland).
References Cook GM, Berney M, Gebhard S, Heinemann M, Cox RA, Danilchanka O, Niederweis M. Physiology of mycobacteria. Adv Microb Physiol. 2009;55:81-182, 318-9. DOI: 10.1016/S0065-2911(09)05502-7
 Holt JG, Krieg NR, Sneath PHA, Staley JT, Williams ST. Bergey's manual® of determinative bacteriology. 9th ed. Baltimore: Williams & Wilkins; 1994.
 Hartmans S, de Bont JAM, Stackebrandt E. The Genus Mycobacterium--Nonmedical. In: The Prokaryotes. New York: Springer; 2006. p. 889–918. DOI: 10.1007/0-387-30743-5_33
 Chan DS, Choy MY, Wang S, Sng LH. An evaluation of the recovery of mycobacteria from urine specimens using the automated Mycobacteria Growth Indicator Tube system (BACTEC MGIT 960). J Med Microbiol. 2008 Oct;57(Pt 10):1220-2. DOI: 10.1099/jmm.0.2008/001685-0
 Cruciani M, Scarparo C, Malena M, Bosco O, Serpelloni G, Mengoli C. Meta-analysis of BACTEC MGIT 960 and BACTEC 460 TB, with or without solid media, for detection of mycobacteria. J Clin Microbiol. 2004 May;42(5):2321-5. DOI: 10.1128/JCM.42.5.2321-2325.2004
 Becton Dickinson and Company. Bactec MGIT (Myco- bacteria growth indicator tube) supplemented with OADC enrichment and PANTA antibiotic mixture user guide. 2005. Available from: http://legacy.bd.com/ds/technicalCenter/inserts/L000180JAA(04).pdf
 Falkinham JO 3rd. Surrounded by mycobacteria: nontuberculous mycobacteria in the human environment. J Appl Microbiol. 2009 Aug;107(2):356-67. DOI: 10.1111/j.1365-2672.2009.04161.x
 Falkinham JO 3rd. Growth in catheter biofilms and antibiotic resistance of Mycobacterium avium. J Med Microbiol. 2007 Feb;56(Pt 2):250-4. DOI: 10.1099/jmm.0.46935-0
 Petrini B. Non-tuberculous mycobacterial infections. Scand J Infect Dis. 2006;38(4):246-55. DOI: 10.1080/00365540500444652
 Chan TH, Ng KC, Ho A, Scheel O, Lai CKW, Leung R. Urinary tract infection caused by Mycobacterium terrae complex. Tubercle and Lung Disease. 1996 Dec;77(6):555–7. DOI: 10.1016/S0962-8479(96)90055-6
 Clark R, Cardona L, Valainis G, Hanna B. Genitourinary infections caused by mycobacteria other than mycobacterium tuberculosis. Tubercle. 1989 Dec;70(4):297–300. DOI: 10.1016/0041-3879(89)90025-1
 Dahl DM, Klein D, Morgentaler A. Penile mass caused by the newly described organism Mycobacterium Celatum. Urology. 1996 Feb;47(2):266–8. DOI: 10.1016/S0090-4295(99)80433-6
 Fox R, Mears J. Female genital tract Mycobacterium chelonae infection. J Obstet Gynaecol. 2004 Apr;24(3):326. DOI: 10.1080/01443610410001661084
 Huang CT, Chen CY, Chen HY, Chou CH, Ruan SY, Lai CC, Hsueh PR. Genitourinary infections caused by nontuberculous mycobacteria at a university hospital in Taiwan, 1996-2008. Clin Microbiol Infect. 2010 Oct;16(10):1585-90. DOI: 10.1111/j.1469-0691.2010.03180.x
 Indudhara R, Das K, Sharma M, Vaidyanathan S. Seminal vesiculitis due to Mycobacterium gastri leading to male infertility. Urol Int. 1991;46(1):99-100.
 Keller M, Mak A, Thibert L, Rene P, Klein MB. Mycobacterium haemophilum epididymal abscess in a renal transplant patient. J Clin Microbiol. 2008 Jul;46(7):2459-60. DOI: 10.1128/JCM.00735-08
 Macek P, Bodnarova M, Zavada J, Jezek P, Pavlik I, Slany M, Havelkova M, Stork J, Duskova J, Hanus T, Kocvara R. Mycobacterium marinum epididymoorchitis: case report and literature review. Urol Int. 2011;87(1):120-4. DOI: 10.1159/000328220
 Oatway WH Jr. Nontuberculous mycobacteria and associated diseases. Am Rev Respir Dis. 1979 Dec;120(6):1389-90. DOI: 10.1164/arrd.19184.108.40.2069
 Orumwense PO, Torvinen E, Heinonen-Tanski H. The survival of mycobacteria in pure human urine. Water Sci Technol. 2013;67(8):1773-7. DOI: 10.2166/wst.2013.052
 Streit M, Böhlen LM, Hunziker T, Zimmerli S, Tscharner GG, Nievergelt H, Bodmer T, Braathen LR. Disseminated Mycobacterium marinum infection with extensive cutaneous eruption and bacteremia in an immunocompromised patient. Eur J Dermatol. 2006 Jan-Feb;16(1):79-83.
 Woods GL, Washington JA. Mycobacteria Other than Mycobacterium tuberculosis: Review of Microbiologic and Clinical Aspects. Clinical Infectious Diseases. 1987 Mar 1;9(2):275–94. DOI: 10.1093/clinids/9.2.275
 Wise GJ, Shteynshlyuger A. An update on lower urinary tract tuberculosis. Current Urology Reports. 2008 Jul;9(4):305–13. DOI: 10.1007/s11934-008-0053-9
 Zarrabi AD, Heyns CF. Clinical features of confirmed versus suspected urogenital tuberculosis in region with extremely high prevalence of pulmonary tuberculosis. Urology. 2009 Jul;74(1):41-5. DOI: 10.1016/j.urology.2008.12.083
 Bonkat G, Bachmann A, Solokhina A, Widmer AF, Frei R, Gasser TC, et al. Growth of Mycobacteria in Urine Determined by Isothermal Microcalorimetry: Implications for Urogenital Tuberculosis and Other Mycobacterial Infections. Urology. 2012 Nov;80(5):1163.e9-1163.e12. DOI: 10.1016/j.urology.2012.04.050
 World Health Organization. Global tuberculosis report 2016. 2016. Available from: apps.who.int/medicinedocs/en/d/Js23098en/
 Naber KG, Schaeffer AJ, Heyns CF, Matsumoto T, Shoskes DA, Bjerklund Johansen TE. Urogenital infections. Arnhem: European Association of Urology; 2010.
 Tonkin AK, Witten DM. Genitourinary tuberculosis. Seminars in Roentgenology. 1979 Oct;14(4):305–18. DOI: 10.1016/0037-198X(79)90027-0
 Kleinnijenhuis J, Oosting M, Joosten LA, Netea MG, Van Crevel R. Innate immune recognition of Mycobacterium tuberculosis. Clin Dev Immunol. 2011;2011:405310. DOI: 10.1155/2011/405310
 Tsolaki AG. Innate Immune Recognition in Tuberculosis Infection. In: Advances in Experimental Medicine and Biology. Springer New York; 2009. p. 185–97. DOI: 10.1007/978-1-4419-0901-5_13
 van Crevel R, Ottenhoff TH, van der Meer JW. Innate immunity to Mycobacterium tuberculosis. Clin Microbiol Rev. 2002 Apr;15(2):294-309.
 Cooper AM. Cell-mediated immune responses in tuberculosis. Annu Rev Immunol. 2009;27:393-422. DOI: 10.1146/annurev.immunol.021908.132703
 Lalvani A, Brookes R, Wilkinson RJ, Malin AS, Pathan AA, Andersen P, et al. Human cytolytic and interferon gamma-secreting CD8+ T lymphocytes specific for Mycobacterium tuberculosis. Proceedings of the National Academy of Sciences. 1998 Jan 6;95(1):270–5. DOI: 10.1073/pnas.95.1.270
 Cooper AM. T cells in mycobacterial infection and disease. Curr Opin Immunol. 2009 Aug;21(4):378-84. DOI: 10.1016/j.coi.2009.06.004
 Grotzke JE, Lewinsohn DM. Role of CD8+ T lymphocytes in control of Mycobacterium tuberculosis infection. Microbes Infect. 2005 Apr;7(4):776-88. DOI: 10.1016/j.micinf.2005.03.001
 Cooper AM, Khader SA. The role of cytokines in the initiation, expansion, and control of cellular immunity to tuberculosis. Immunol Rev. 2008 Dec;226:191-204. DOI: 10.1111/j.1600-065X.2008.00702.x
 Allie N, Grivennikov SI, Keeton R, Hsu NJ, Bourigault ML, Court N, Fremond C, Yeremeev V, Shebzukhov Y, Ryffel B, Nedospasov SA, Quesniaux VF, Jacobs M. Prominent role for T cell-derived tumour necrosis factor for sustained control of Mycobacterium tuberculosis infection. Sci Rep. 2013;3:1809. DOI: 10.1038/srep01809
 Francisco NM, Hsu NJ, Keeton R, Randall P, Sebesho B, Allie N, Govender D, Quesniaux V, Ryffel B, Kellaway L, Jacobs M. TNF-dependent regulation and activation of innate immune cells are essential for host protection against cerebral tuberculosis. J Neuroinflammation. 2015 Jun 26;12:125. DOI: 10.1186/s12974-015-0345-1
 Ulrichs T, Kosmiadi GA, Trusov V, Jörg S, Pradl L, Titukhina M, Mishenko V, Gushina N, Kaufmann SH. Human tuberculous granulomas induce peripheral lymphoid follicle-like structures to orchestrate local host defence in the lung. J Pathol. 2004 Oct;204(2):217-28. DOI: 10.1002/path.1628
 Kapina MA, Shepelkova GS, Mischenko VV, Sayles P, Bogacheva P, Winslow G, Apt AS, Lyadova IV. CD27low CD4 T lymphocytes that accumulate in the mouse lungs during mycobacterial infection differentiate from CD27high precursors in situ, produce IFN-gamma, and protect the host against tuberculosis infection. J Immunol. 2007 Jan;178(2):976-85.
 Brandau S, Suttmann H. Thirty years of BCG immunotherapy for non-muscle invasive bladder cancer: a success story with room for improvement. Biomed Pharmacother. 2007 Jul;61(6):299-305. DOI: 10.1016/j.biopha.2007.05.004
 Patard JJ, Saint F, Velotti F, Abbou CC, Chopin DK. Immune response following intravesical bacillus Calmette-Guerin instillations in superficial bladder cancer: a review. Urol Res. 1998;26(3):155-9.
 Koziol-Montewka M, Kolodziejek A, Oles J, Janicka L. Proinflammatory cytokine profile in active kidney tuberculosis patients. Immunol Invest. 2004 Aug;33(3):277-85.
 Nguyen L, Pieters J. Mycobacterial subversion of chemotherapeutic reagents and host defense tactics: challenges in tuberculosis drug development. Annu Rev Pharmacol Toxicol. 2009;49:427-53. DOI: 10.1146/annurev-pharmtox-061008-103123
 Müller B, Borrell S, Rose G, Gagneux S. The heterogeneous evolution of multidrug-resistant Mycobacterium tuberculosis. Trends Genet. 2013 Mar;29(3):160-9. DOI: 10.1016/j.tig.2012.11.005
 David HL. Probability distribution of drug-resistant mutants in unselected populations of Mycobacterium tuberculosis. Appl Microbiol. 1970 Nov;20(5):810-4.
 Gillespie SH. Evolution of drug resistance in Mycobacterium tuberculosis: clinical and molecular perspective. Antimicrob Agents Chemother. 2002 Feb;46(2):267-74.
 Gandhi NR, Nunn P, Dheda K, Schaaf HS, Zignol M, van Soolingen D, et al. Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of tuberculosis. The Lancet. 2010 May;375(9728):1830–43. DOI: 10.1016/S0140-6736(10)60410-2
 Kulchavenya E. Best practice in the diagnosis and management of urogenital tuberculosis. Ther Adv Urol. 2013 Jun;5(3):143-51. DOI: 10.1177/1756287213476128
 Zhang Y, Yew WW, Barer MR. Targeting persisters for tuberculosis control. Antimicrob Agents Chemother. 2012 May;56(5):2223-30. DOI: 10.1128/AAC.06288-11
 Wakamoto Y, Dhar N, Chait R, Schneider K, Signorino-Gelo F, Leibler S, McKinney JD. Dynamic persistence of antibiotic-stressed mycobacteria. Science. 2013 Jan;339(6115):91-5. DOI: 10.1126/science.1229858
 Dhar N, McKinney JD. Microbial phenotypic heterogeneity and antibiotic tolerance. Curr Opin Microbiol. 2007 Feb;10(1):30-8. DOI: 10.1016/j.mib.2006.12.007
 Gomez JE, McKinney JD. M. tuberculosis persistence, latency, and drug tolerance. Tuberculosis. 2004 Jan;84(1–2):29–44. DOI: 10.1016/j.tube.2003.08.003