Reinfection with Mycobacterium tuberculosis in an urban tuberculosis hospital

Concise Communications

ventricles and died subsequently due to the central nervous system infection. This report indicates the clinical spectrum of disease caused by this pathogen. However, we did not encounter in our series cases of pleuropulmonary, ophthalmic or amniotic fluid infections, as reported by other authors [7-lo]. Ophthalmologic infection secondary to surgical procedures has been reported as one of the commonest infections caused by this organism [9,10]. However, this diagnosis is a difficult one, because of the need for relevant deep samples for culture (e.g., intraocular lenses); in our series a few such specimens were processed and it is probable that this infection could have been missed in a retrospective investigation. Foreign bodies and previous surgery play important roles in the pathogenesis of infection due to l? acnes, as shown in this report. The high concentrations of l? acnes in the sebaceous glands of the scalp and forehead make neurologic sites a target for its pathogenic potential [2]. Perhaps for this reason, in 26 of 44 cases (59.1%) the infection had been detected in a neurologic setting. The clinical presentation of these infections due to I? acnes (brain abscess, surgical wound infection, epidural or subdural empyema) did not differ from that of the same diseases caused by other organisms. However, in some cases, the course of I? acnes infection in the central nervous system may be insidious and features may become evident later [4]. I? acnes is susceptible to p-lactam antibiotics, macrolides and lincosamides but resistant to imidazoles [ll]. Antimicrobial therapy was important in the therapy of the patients in our series. However, surgical procedures seem to have been of equally great importance in the management of some patients, mainly those who had infections that required a mixed surgical-medical approach, such as brain abscesses. It is important to recognize that I? acnes, especially when it appears in pure culture, can be more than simply a ‘contaminant’, and adequate clinical and microbiological evaluation of each isolate must be made in order to decide its true significance. Jaime Esteban, Jose-Manuel Ramos, Francisco Soriano Department of Medical Microbiology, Fundacibn JimCnez Diaz, Madrid, Spain Revised version accepted 22 July 1997

References 1. C u m i n s CS, Johnson JL. Genus Propionibacterium. In Sneath PHA, Mair NS, Sharpe ME, Holt JG, eds. Bergey’s

49

2. 3. 4.

5.

6.

7.

8.

9.

10. 11.

manual of systematic bacteriology. Baltimore: Wilhams & Wilkins, 1986: 1346-53. Ray LF, Kellum RE. Corynebacterium acnes from human s h n . Arch Dermatol 1970; 101: 36-40. Brook I, Frazier EH. Infections caused by Propionibacterium species. Rev Infect Dis 1991; 13: 819-22. Ramos JM, Esteban J, Soriano E Isolation of Propionibacterium acnes from central nervous system infections. Anaerobe 1995; 1: 17-20. Esteban J, Ramos JM, Jimenez-Castdlo F’, Soriano E Surgical wound infections due to Propionibacterium acnes: a study of 10 cases. J Hosp Infect 1995; 30: 229-32. Esteban J, Cuenca-Estrella M, Ramos JM, Soriano E Granulomatous infection due to Propionibacterium acnes mimiclung mahgnant disease. Eur J Clin Microbiol Infect Dis 1994; 13: 1084. Reed CS, Feldman RG, Cochrane RM. Amnionitis caused by Propionibacterium acnes. Br J Obstet Gynaecol 1988; 95: 626-7. Claeys G, Verschraegen G, De Potter C, Cuvelier C, Pauwels R. Bronchopneumonia caused by Propionibacteriumacne5. Eur J Clin Microbiol Infect Dis 1994; 13: 747-9. Winward KE, Pflugfelder SC, Flynn HW, Roussel TJ, Davis JL. Postoperative Propionibacterium endophthalnutis. Ophthalmology 1993; 100: 447-51. Zaidman GW. Propionibacterium acne5 keratitis. Am J Ophthalmol 1992; 113: 596-8. Denys GA, Jerris R C , Swenson JM, Thornsberry C. Susceptibdity of Propionibacterium acnes clinical isolates to 22 antimicrobial agents, Antimicrob Agents Chemother 1983; 23: 335-7.

Reinfection with Mycdacterium tuberculosis in an urban tuberculosis hospital

Clin Microbiol Infect 1998; 4: 49-51

We report on the reinfection with susceptible Mycobacterium tuberculosis strains in patients in King George V Hospital in Durban, KwaZulu-Natal. Twenty-two hospitalized patients who remained sputum positive after 5 weeks or more of conventional short-course antituberculosis chemotherapy (2 months of ethambutol, isoniazid, rifampicin and pyrazinamide followed by 4 months of three times weekly isoniazid and rifampicin) were investigated to determine whether they were still infected by the original strain of M . tuberculosis. Pretreatment (PT) and the last positive (LP) isolates were fingerprinted by means of IS6210 restriction fragment length polymorphism (RFLP) analysis according to a standard method [l]. RFLPs for eight of the 22 patients are shown in Figure 1. For 16 patients, RFLPs of paired isolates were identical (including one pair in which the LP isolate had undergone a single transpositional event). These

50

1

C l i n i c a l M i c r o b i o l o g y a n d I n f e c t i o n , V o l u m e 4 N u m b e r 1, J a n u a r y 1 9 9 8

B B B B A A A A A 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1819 20

Figure 1 Lanes 1, 10 and 19 contain the reference strain Mt14323. Lane 20 contains a molecular weight marker. Paired clinical isolates &om eight of the 22 patients are present in lanes 2 and 3, 4 and 5, 6 and 7 (patient 6), 8 and 9 (patient 4), 11 and 12 (patient l), 13 and 14, 15 and 16, and 17 and 18, with pretreatment isolates preceding the last positive. A discordant RFLP is shown in lane 12 and is highly similar to that of cluster A shown in lanes 6-9. Two isolates belonging to cluster B are shown in the figure. The isolate in lane 16 did not produce a fingerprint and was identified as a MOTT.

patients are regarded as treatment failures. O f these 16, three clusters containing eight patients were identified, indicative of recently acquired infection prior to hospitalization. The LP isolate in three of the 22 patients did not hybridize with IS61 10, and these were subsequently identified as mycobacteria other than tuberculosis (MOTTS). The RFLPs of the LP isolate of the remaining three patients were discordant with those of the P T isolates. The RFLPs of two of these patients (patients 1 and 2) were nearly identical to each other and either identical or nearly identical to those belonging to a cluster of three (patients 3, 4 and 5). These isolates were fully susceptible to antituberculous drugs. The time lapse between the paired isolates was 11 and 19 weeks. Review of the patients’ records showed an overlap of their hospitalization periods as follows: patients 1 and 2 were hospitalized between 23 March 1994 and 3June 1994, and 16 March 1994 and 11 July 1994, respectively. Patients 4, 5 and 6 were hospitalized between 9 February 1994 and 24 June 1994, 16 February 1994 and 19 May 1994, and 21 February 1994 and 27 May 1994. The second strain ofpatient 1 was isolated on 16 August 1994 (i.e. at follow-up 10 weeks after discharge) and that of patient 2 on 16 June 1994 (during the

hospitalization period). This means that patient 2 did not leave the hospital during the period in which reiniection could have occurred. Even though the discordant isolate of patient 1 was cultured after discharge, the possibility of reinfection having occurred from the clustered patients in the hospital is extremely high, especially since patient 1 and patient 4 had shared a common ward. Patients are not forcibly incarcerated in their respective wards for prolonged periods and therefore have free contact with one another, thus promoting the possibility of reinfection. It is highly unlikely that laboratory cross-contarnination was responsible for these discrepant results, as the second strains from both patients were cultured from specimens that were received and processed on different days from those of the three clustered patients from whom they may have acquired infection. Mixed infection could possibly explain the discordant RFLPs, but our findings that these two patients were hospitalized at the same time as the clustered patients support our hypothesis that they were reinfected with a second strain of M. tuberculosis while still on treatment during hospitalization. HIV testing was performed on admission and 2 months later. Patients 1 and 3 were HIV negative on both occasions, and patient 2 was HIV negative at the baseline and with no further serologic data 2 months later. Patient 4 tested HIV positive on both occasions and patient 5 was negative at the baseline but positive at month 2 (possible ‘window period’). This group of five patients is too small to link HIV status with the chance of reinfection. However, these patients had active T B which, depending on severity, has been known to result in impaired immunity which might increase susceptibility to reinfection [2,3]. The non-matching RFLP of the third patient was identical to that of a PT isolate of one of the patients whose LP isolate was a MOTT. The lack of overlap in their hospitalization periods suggests that these two patients were most probably infected by a third patient not included in the cohort studied. Patients 1, 3 and 5 were reported to be well and non-productive at the last visit at 18, 24 and 18 months respectively. Patient 2 was reported well at 3 months but defaulted at 6 months. Patient 4 was well and culture negative at month 5 but had defaulted thereafter. After the first readmission between 4 January 1995 and 8 May 1995, the patient defaulted again and was readmitted for the third time between 27 June 1996 and 18 December 1996. At this point, this patient was reported to be well. The presence of susceptible strains after prolonged periods of treatment is not an unknown phenomenon, but is usually related to advanced disease which

Concise Communications

influences the pharmacokinetics of the drugs adversely. This preliminary report of an ongoing study confirms the long-held suspicion that reinfection does occur in a TB hospital setting in a developing country. What is surprising, however, is the transmission of fully susceptible strains to patients who are still on therapy. It is, therefore, tempting to hypothesize that prolonged hospitalization aimed at improving compliance will provide the environment for the acquisition of different strains by patients on treatment, thereby prolonging the time it takes to effect cure. The recently completed TB Review by a panel of international experts indicates a cure rate of 30-60% in South Africa [4]. The WHO has suggested that when cure rates fall below 50%, there can be a negative epidemiologic impact on the prevalence of smear-positive cases, especially in the presence of improved case finding. This is a result of increased numbers of chronic transmitters emanating fiom therapeutic failures [ 5 ] . It is hoped that ongoing studies in this laboratory will serve to firmly document this phenomenon with a view to influencing admission policy in hospital settings. Acknowledgments

This work has been partially presented as part of a paper entitled ‘RFLP analysis of the Mycobacterium tuberculosis complex in KwaZuWNatal’ at the following meetings: 5th Joint Congress ofthe Sexually Transmitted Diseases and Infectious Diseases Societies of Southern Africa in Durban, South Africa, in May 1995. Glaxo Action TB Scientific Conference and Workshops in Cape Town, South Africa, in October 1995. 2nd meeting of the EC Concerted Action on Genetic markers and TB epidemiology in Paris, France, in November 1995. Manormoney Pillay I , Philip Onyebujoh2, A . Willem Sturm *Department of Medical Microbiology, Faculty of Medicine, University of Natal, South M i c a ; ’Medical Research Council, South Africa Revised version accepted 3 August 1997

References 1. Van Embden JDA, Cave MD, Crawford JT, et al. Strain identification of Mycobacterium tuberrulosis by DNA fingerprinting: recommendations for a standardized methodology. J Clin Microbiol 1993; 31: 406-9.

51

2. Singhal M, Banavalikar JN, Sharma S, Saha K. Peripheral blood T lymphocyte subpopulations in patients with TB and the effect of chemotherapy. Tubercle 1989; 70: 171-8. 3. Ainslie GM, Solomon JA, Bateman ED. Lymphocyte and lymphocyte subset numbers in blood and in bronchoalveolar lavage and pleural fluid in various forms of human pulmonary TB at presentation and during recovery. Thorax 1992; 47: 513-18. 4. Report on the review of the tuberculosis program of South Africa Department of Health, Republic of South Africa. Epidemiol Comments 1996; 23(1): 2-20. 5. Tuberculosis research and development: report of a WHO meeting, Geneva, 23-25 October 1990.

Catheter-related bacteremia due to Tsukamurellapulmonis

Clin Microbiol Infect 1998; 4: 51-53

Catheter-related bacteremia is an important problem for patients with hematologic malignancies in whom long-term venous access is required for the safe administration of high-dose chemotherapy and optimal supportive care. These infections are mainly caused by organisms colonizing the skin. Coagulase-negative staphylococci and Staphylococcus aureus are the leading causes of catheter-related infection. Other reported pathogens include Stenotrophomonas maltophilia, Burkholderia cepacia, Corynebacterium sp. and Candida spp. Here we report the first case of bacteremia caused by Tsukamurella pulmonis; a Hickman catheter-related bacteremia in a patient with malignant lymphoma. A 38-year-old woman had been well until January 1995, when she presented with a subacute superior vena cava syndrome due to the mediastinal involvement of an extended-stage high-grade mahgnant B-cell nonHodgkin’s lymphoma (Burkitt-like). After a brief period of radiotherapy and steroids, a Hickman catheter was implanted for the administration of several cycles of intensive cytoreductive therapy. Remission-induction therapy, consisting of daunorubicin, vincristine, cyclophosphamide, steroids and intrathecal prophylaxis, resulted in complete remission. Therapy was completed by two consolidation courses and radiotherapy. In October 1995 she was readmitted to the hospital for autologous peripheral blood progenitor cell transplantation. Four days before the transplantation, the patient developed spiking fever, diarrhea and cramping abdominal pain with nausea and vomiting. Because of her neutropenic status (l100/mm3), she was started on intravenous ceftazidime for a total of 7 days. Repeated cultures from blood and stool samples yielded no pathogens; chest and abdominal X-ray and ultrasound examination were unremarkable. Within the next 2