Orientia, rickettsia, and leptospira pathogens as causes of CNS infections in Laos: a prospective study

Summary Background Scrub typhus (caused by Orientia tsutsugamushi), murine typhus (caused by Rickettsia typhi), and leptospirosis are common causes of febrile illness in Asia; meningitis and meningoencephalitis are severe complications. However, scarce data exist for the burden of these pathogens in patients with CNS disease in endemic countries. Laos is representative of vast economically poor rural areas in Asia with little medical information to guide public health policy. We assessed whether these pathogens are important causes of CNS infections in Laos. Methods Between Jan 10, 2003, and Nov 25, 2011, we enrolled 1112 consecutive patients of all ages admitted with CNS symptoms or signs requiring a lumbar puncture at Mahosot Hospital, Vientiane, Laos. Microbiological examinations (culture, PCR, and serology) targeted so-called conventional bacterial infections (Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus influenzae, S suis) and O tsutsugamushi, Rickettsia typhi/Rickettsia spp, and Leptospira spp infections in blood or cerebrospinal fluid (CSF). We analysed and compared causes and clinical and CSF characteristics between patient groups. Findings 1051 (95%) of 1112 patients who presented had CSF available for analysis, of whom 254 (24%) had a CNS infection attributable to a bacterial or fungal pathogen. 90 (35%) of these 254 infections were caused by O tsutsugamushi, R typhi/Rickettsia spp, or Leptospira spp. These pathogens were significantly more frequent than conventional bacterial infections (90/1051 [9%] vs 42/1051 [4%]; p<0·0001) by use of conservative diagnostic definitions. CNS infections had a high mortality (236/876 [27%]), with 18% (13/71) for R typhi/Rickettsia spp, O tsutsugamushi, and Leptospira spp combined, and 33% (13/39) for conventional bacterial infections (p=0·076). Interpretation Our data suggest that R typhi/Rickettsia spp, O tsutsugamushi, and Leptospira spp infections are important causes of CNS infections in Laos. Antibiotics, such as tetracyclines, needed for the treatment of murine typhus and scrub typhus, are not routinely advised for empirical treatment of CNS infections. These severely neglected infections represent a potentially large proportion of treatable CNS disease burden across vast endemic areas and need more attention. Funding Wellcome Trust UK.


Introduction
The most common bacterial pathogens responsible for meningitis in southeast Asia are Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus infl uenzae, S suis, and Mycobacterium tuberculosis. 1 Timely empirical and specifi c pathogen-directed treatment is essential, usually, except for M tuberculosis, including a third-generation cephalosporin. However, many patients with CNS infections do not receive a causal diagnosis despite cerebrospinal fl uid (CSF) culture and DNA molecular assays, 2 partly because of low CSF pathogen density and previous antibiotic use. 2 Other neglected bacteria probably cause CNS infections in Asia, including pathogens not expected to respond to third-generation cephalosporins. During World War 2, scrub typhus (caused by Orientia tsutsugamushi) was a well-recognised cause of lethal meningitis in the Asia-Pacifi c region, but this clinical experience has largely been forgotten. 3 Findings of studies in India and Thailand showed that up to 15% of patients with scrub typhus had neurological complications. [4][5][6] O tsutsugamushi DNA was detected in the CSF of Taiwanese patients serologically confi rmed to have scrub typhus. 7 Indeed, altered CNS function is implicit in the name typhus, which means stupor. Similarly, Rickettsia typhi (the cause of murine typhus) and other Rickettsia species cause meningoencephalitis. Another neglected but common group of pathogens, the Leptospira spp have received little attention as causes of CNS infection. In a study in the Philippines, 5% of patients with aseptic meningitis had high serological titres to Leptospira spp; 8 in a Brazilian study, more than 50% of patients with aseptic meningitis were CSF PCR positive for Leptospira spp. 9 However, leptospiral meningitis would be expected to respond to thirdgeneration cephalosporins, if severe leptospirosis does respond to antibiotics. 10 Leptospira and Rickettsia species are distributed worldwide 11,12 and O tsutsugamushi is endemic across Asia, the Pacifi c islands, and northern Australia. 13 Although a vast human population is potentially exposed to treatable rickettsial and leptospiral diseases, few data exist on the incidence and clinical features of rickettsial and leptospiral CNS infections. Appropriate diagnostic methods or trialled optimum treatments are scarce. Scrub typhus, leptospirosis, and murine typhus are common diseases in Laos, both in the capital, Vientiane, and in rural areas 14,15 and in adjacent countries, including China and Thailand. Although Asia is geographically, culturally, economically, and ethnically diverse, Laos is an example of the vast areas of rural Asia that are economically poor with little medical information to guide public health policy. Therefore, we assessed whether these pathogens are important causes of CNS infections in Laos.

Study design and participants
In this prospective study, patients were enrolled between Jan 10, 2003, andNov 25, 2011, at Mahosot Hospital, Vientiane, Laos. 16 Inpatients of all ages were recruited if a diagnostic lumbar puncture was indicated on the basis of altered consciousness or neurological fi ndings by the attending physicians, and if there were no contraindications. Informed consent (verbal during 2003-06; written during 2006-11) was given by the patient, parents, or guardian (Dubot-Pérès A, et al, unpublished). Ethical approval was granted by OXTREC (University of Oxford, UK) and the Faculty of Medical Sciences Committee (University of Health Sciences, Laos).

Procedures
Acute encephalitis syndrome and meningitis were defi ned according to WHO 2003 guidelines. 17 Acute encephalitis syndrome was defi ned as the acute onset of fever and either a change in mental status (including symptoms such as confusion, disorientation, coma, or inability to talk) and new onset of seizures (excluding simple febrile seizures) in a person of any age. Meningitis was defi ned as a sudden onset of fever (>38·5°C rectal or 38·0°C axillary) with one of the following signs: neck stiff ness, altered consciousness, or other meningeal signs. If a patient fulfi lled criteria for both disorders, we used the term meningoencephalitis. We recorded demographic and clinical data on standardised forms and grouped data according to guidelines. 17,18 Occupations were classifi ed as farmer, housewife, teacher, government offi cial, driver, building worker, merchant, health worker, police, monk, mechanic, soldier, child (<5 years), schoolboy or girl (5-15 years), student (>15 years), or unemployed.
We measured the CSF opening pressure with manometers. The target CSF volumes were 8 mL for adults (>15 years), 3·5 mL for children (1-15 years), and 2·5 mL for infants (<1 year). We measured CSF lactate and glucose concentrations with Olympus AU400/ AU400e Chemistry ImmunoAnalyzers (V-Diagnostic Center, Bangkok, Thailand). Whole blood samples were taken for two blood culture bottles: non-anticoagulated blood for tests on serum and blood clots, and EDTA blood for tests on whole blood, plasma, and buff y coat samples. 14 CSF and blood cultures were processed as described previously. 14,16,17,19 The median interval between admission and convalescent serum samples was 10·5 days (range 2-90). We tested for rickettsial antibodies (IgM and IgG) with batched indirect immunofl uorescence assays for scrub typhus and murine typhus. 14 We did leptospiral microscopic agglutination tests in one batch, which were interpreted by the WHO/AO/OIE Collaborating Centre for Reference and Research on Leptospirosis, Brisbane, Australia. 14 We regarded a four-fold increase between admission and convalescent samples (by immunofl uorescence assay or microscopic agglutination test) as evidence of acute infection, and a two-fold increase or decrease (with microscopic agglutination test), a titre of 1:400 or more (with microscopic agglutination test), or a high static titre (≥1:12 800, with immunofl uorescence assay) as evidence of probable or recent infection.
Leptospira spp (from 2006), Rickettsia spp, and O tsutsugamushi (from 2008) were cultured as described previously. 14 In-vitro isolation was attempted from buff y coat for patients with admission-positive murine typhus (ImmunoDot, GenBio, USA) or scrub typhus IgM rapid test (Standard Diagnostics, Korea) 14 results on serum analysis.
PCR templates were prepared from EDTA buff y coat or CSF samples. We extracted DNA with the QIAGEN DNA Mini kit or QIAGEN EZ-1 extraction-robot. 14

Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
Between 2003 and 2011, 1112 patients were recruited and physicians collected CSF from 1051 (95%) patients (appendix p 1). Most patients were male and had a history of antibiotic administration (  (table 2). Of 446 patients with data available for both direct and serological results, we detected evidence of scrub typhus, Rickettsia spp or murine typhus, or leptospirosis in 59 (13%) of 446 patients. Leptospira interrogans was the most commonly identifi ed Leptospira species by microscopic agglutination test or sequencing (GenBank KJ150298-KJ150302; appendix p 3).We identifi ed Rickettsia spp as R typhi for 16 (84%) of 19 patients who were qPCR positive, but for three patients the Rickettsia species could not be established because no template remained. In addition to the 90 patients with O tsutsugamushi, R typhi/ Rickettsia spp and Leptospira spp monoinfections, we recorded grade 1 and grade 2 multiple infections in an additional 14 patients (appendix p 4). 24 More than half of all patients met the WHO criteria for meningitis (709/1093 [65%]) or acute encephalitis syndrome (610 [56%]), 84 (74%) met either criteria, and 521 (48%) fulfi lled both criteria (table 1 and fi gure 2). Patients with a conventional bacterial CNS infection presented with the shortest median duration of fever (p=0·005), the highest frequencies of convulsions (p=0·023) and neck stiff ness (p=0·182), the lowest median Glasgow coma scale (GCS) scores on admission (p=0·062), and the highest mortality (p=0·076), compared with patients with R typhi or Rickettsia spp and Leptospira spp infections (table 1).
Patients with a CNS infection caused by R typhi/ Rickettsia spp tended to present later in their illness than patients with conventional bacteria (table 1). They had the lowest frequency of vomiting and none reported photophobia. The mortality of patients with R typhi or Rickettsia spp (27%) was nearly double that of patients with scrub typhus (14%, p=0·307) and leptospirosis (13%, p=0·299); however, these diff erences were not signifi cant. Notably, the mortality of patients with R typhi/Rickettsia spp was similar to those infected with  conventional bacteria (about 33%). Furthermore, compared with patients with conventional bacteria, a similar proportion of patients presented with a reduced GCS score (table 1).
Patients with O tsutsugamushi also presented late in their illness and had a high frequency of rash (20%; table 1). 81% fulfi lled the WHO meningitis criteria (fi gure 2) with a signifi cantly higher median GCS score and the highest reported preadmission antibiotic use (86%) of investigated groups.
Fewer patients in the Leptospira group had convulsions (19%) than those in the conventional bacterial group (44%) and all other investigated groups, but presented with the highest frequency of peripheral neurological abnormalities (11%). One patient had GCS  Patients with conventional bacterial infections had high opening pressures, frequent CSF turbidity, and high cellularity. Cells were mainly neutrophils, with high CSF lactate, low glucose, and high protein concentrations (table 1).
Despite similar clinical severity and mortality, the CSF characteristics of patients infected with R typhi or Rickettsia spp diff ered from the conventional bacteria group. The opening pressure tended to be lower and only one patient had CSF turbidity. CSF cellularity, neutrophil to lymphocyte ratio, and lactate and protein concentrations were also signifi cantly diff erent from conventional bacterial infections (table 1).
For patients infected with O tsutsugamushi, opening pressures were similar to those with conventional bacterial illness, but CSF turbidity and cellularity were signifi cantly less common and CSF lactate concentrations were the lowest of all groups (table 1). Compared with those with R typhi or Rickettsia spp infections, patients with scrub typhus had signifi cantly higher CSF white cell counts (p=0·018) and a two-fold higher neutrophil to lymphocyte ratio (p=0·045).
For patients infected with Leptospira spp, the CSF opening pressure was similar to the other groups, but CSF turbidity and cellularity were rare, with lower protein and higher glucose concentrations than patients in the conventional bacteria group (table 1).

Discussion
These data suggest that O tsutsugamushi, R typhi or Rickettsia spp, and Leptospira spp infections are the leading causes of bacterial CNS infections in Laos. With increasing evidence that O tsutsugamushi, R typhi or Rickettsia spp, and Leptospira spp are important causes of fevers, these fi ndings raise concerns that these infections are responsible for a large proportion of neglected but treatable CNS disease burden in the many endemic countries and in travellers (panel). 12,14,25,26 8% of all CNS infections and 46% of identifi ed bacterial causes were attributable to O tsutsugamushi, R typhi/ Rickettsia spp, or Leptospira spp, with conservative and robust diagnostic defi nitions; more than double the 22% attributed to four conventional bacteria species. The overall mortality of CNS infections in Laos was 27%, with group-specifi c mortality of 18% for Orientia, Rickettsia, and Leptospira spp, and 33% for conventional bacteria, which emphasises the importance of improving diagnostic and treatment strategies. Comparisons of clinical fi ndings highlight diff erences that might serve as diagnostic clues. For example, patients with R typhi/ Rickettsia spp and Leptospira spp infections were older and patients with O tsutsugamushi presented later in their illness, consistent with reports from India, 6 and commonly had skin rashes. Elsewhere, hearing loss has often been noted in patients with O tsutsugamushi, 27 but was not signifi cantly more common in patients with O tsutsugamushi and CNS disease in Laos than other patient groups. Consistent with previous fi ndings, 6 patients with O tsutsugamushi and R typhi/Rickettsia spp infections presented with low, but abnormal CSF white cell counts; turbid CSF and raised CSF lactate concentrations were infrequent.
Data from recent reports suggest that admission interstitial pneumonitis is associated with meningitis in patients with O tsutsugamushi infection. 6,28 Although most Lao patients with O tsutsugamushi presented with tachypnoea, this was not signifi cantly more frequent than in other groups (p=0·587). Furthermore, O tsutsugamushi infection commonly presents with tachypnoea without CNS involvement. 15 R typhi/Rickettsia spp infections were more severe than O tsutsugamushi and leptospirosis in terms of GCS and mortality. However, R typhi infection is generally regarded as benign 29 and although it is distributed worldwide, is rarely included in the diff erential diagnosis of CNS disease.
Reduced consciousness and seizures were the most common neurological symptoms in Indian patients with neuroleptospirosis. 30 We did not record this fi nding in Laos, which raises questions about variation in strain virulence and host susceptibility. Three patients with leptospirosis had some neurological evidence of intracerebral lesions, consistent with the putative association between leptospirosis and Moyamoya disease, but cerebral angiography, which is not available in Laos, is needed to confi rm the diagnosis. 31 Notably, only patients with leptospirosis had abnormal peripheral neurological symptoms and signs.
Conventional bacterial infections were signifi cantly associated with turbid, cellular CSF containing high neutrophil counts. The high frequency of CSF abnormalities (including increased white cell counts, turbidity, and lactate concentrations) in all patients who had a lumbar puncture, but without a causal diagnosis, suggests that many of these patients had undetected infections and that patients with important CNS pathology are not receiving lumbar punctures.
During this 9 year investigation, we noted a distinct seasonal pattern, similar to fi ndings for non-malarial fevers in Laos. 14 Conventional bacterial infections (eg, N meningitidis and S pneumoniae), were most frequent in the dry season, peaking in January, consistent with data from India, 32 whereas Orientia, Rickettsia, and Leptospira species were detected in nearly 20% of febrile patients diagnosed with CNS disease at the end of the rainy season. This seasonality and CSF characteristics could help to guide clinicians' diff erential diagnosis.

Systematic review
We searched PubMed for relevant articles published in English up to April, 2014, using the search terms "central nervous system infections" or "meningitis" or "encephalitis" together with "scrub typhus" or "O tsutsugamushi", "murine typhus" or "Rick*", or "leptospir*". 195 relevant publications were identifi ed, describing case series and case reports of CNS disease caused by these pathogens with some including clinical and cerebrospinal fl uid characteristics or both. However, none of the published scientifi c literature described the prospective investigation of consecutive hospital patients to assess the contribution of scrub typhus, murine typhus, and leptospirosis to the CNS disease burden in endemic areas. Articles published in PubMed before 1966 alluded to the potential of Orientia tsutsugamushi, Rickettsia spp, and Leptospira spp to cause CNS disease and underlined the need for further investigation.

Interpretation
Our fi ndings show that rickettsial and leptospiral pathogens are important causes of meningitis, encephalitis, and meningoencephalitis in Laos. Our fi ndings show the importance of these neglected but important diseases and suggest that clinicians and microbiologists need to be aware of these treatable causes of severe CNS infection. In view of the ineff ectiveness of the standard empirical cephalosporin and penicillin treatments against rickettsial pathogens, these data emphasise the need to rethink treatment guidelines in endemic regions to consider including an antirickettsial antibiotic.
Our study has important limitations, including the use of suboptimum samples for leptospiral culture, 33 the known limitations of rickettsial diagnostics, 34,35 and the use of stored samples. However, these limitations probably led to the underestimation of the incidence of Leptospira spp, O tsutsugamushi, and R typhi/Rickettsia spp infections. The widespread use of over-the-counter antibiotics 2 before admittance to hospital probably reduced culture rates for conventional bacteria, and we did not do serology assays for these pathogens. The lack of clustered results and stringent sample handling protocols, physical separation of processes, and the use of uracil-DNA glycosylase in the PCR mix, makes specimen contamination very unlikely. The three Rickettsia spp infections that could not be speciated were probably R typhi because this is the main species in Laos. 14,36 Our data suggest that empirical treatment practice for CNS infections in Laos, where third-generation cephalosporin monotherapy is generally used, should be reconsidered. Although current guidelines are probably eff ective for leptospiral CNS disease, there is no evidence for their effi cacy against O tsutsugamushi and R typhi/ Rickettsia spp, which should be treated with doxycycline or chloramphenicol. Azithromycin or rifampicin might be active against O tsutsugamushi CNS disease. 37 More data are needed for CSF drug levels in patients with typhus and variability in minimum inhibitory concentrations against R typhi/Rickettsia spp and O tsutsugamushi, with clinical trials to inform optimum treatment. 28 Because of the paucity of accessible and accurate admission laboratory typhus diagnostics, 34,38 empirical treatment with doxycycline plus a thirdgeneration cephalosporin might be appropriate in areas endemic for scrub typhus and murine typhus. However, combination of bacteriostatic tetracyclines with bactericidal cephalosporins might reduce the effi cacy in typhus-endemic areas for treatment of conventional bacterial pathogens, such as S pneumoniae. 39 The optimum management of patients with either confi rmed O tsutsugamushi, R typhi/Rickettsia spp, or Leptospira spp CNS infection or of those in which these or conventional bacteria are suspected remains unclear, with little evidence to guide policy. Indeed, there is very little evidence on the pharmacokinetics of the tetracyclines in CSF, or optimum dose for CNS disease. 40 These fi ndings suggest that greater appreciation and further investigation of Orientia, Rickettsia, and Leptospira spp as neglected but treatable causes of CNS disease in other endemic areas globally is urgently needed. 3,41 Although pathogen discovery has an important role, we suggest that optimising diagnosis, treatment, and prevention of these neglected but common bacteria might have a more rapid benefi cial public health outcome.

Contributors
PNN, SD, and DHP designed the study. SR and PNN collected the data and did the data cleaning. SD and SJL designed and interpreted statistical results. SBC, SMT SDB, DABD, and AD-P contributed to data collection and data interpretation. AS, PPa, PPh, and SD did the laboratory investigations. SD wrote the fi rst draft of the paper and PNN, DHP, and SD contributed to the fi nal report. The corresponding author had full access to all the data in the study. All authors reviewed the report and agreed to submit for publication.

Declaration of interests
We declare no competing interests.

Rickettsioses as causes of CNS infection in southeast Asia
In The Lancet Global Health, Sabine Dittrich and colleagues 1 report that scrub typhus caused by Orientia tsutsugamushi, murine typhus caused by Rickettsia typhi, and leptospirosis caused by various Leptospira species account for more than a third of CNS infections diagnosed over 8 years in Vientiane Hospital in Laos. The study is one more great contribution from this team in their investigation of undocumented syndromes, as well as in the public health challenge of rickettsial diseases in southeast Asia. The same investigators have previously reported that scrub typhus was the second most common microbial cause of fever of unknown origin in rural Laos (122 [15%] of 799 diagnosed cases). 2 In 2006, rickettsial infection was detected in 115 (27%) of 427 adults admitted to Vientiane Hospital for fever with negative blood culture. 3 The most common rickettsial agent was O tsutsugamushi followed by R typhi. Fewer data are available about the prevalence of these diseases in other southeast Asian countries. In Thailand, scrub and murine typhus has been reported 4 in 16% and 2%, respectively, of fever of unknown origin, with mortality of 3-17% for scrub typhus. Even if epidemiological data for the whole region are unavailable, the substantial presence of rickettsial infections is shown by frequent reports in travellers returning from this area. 5 Because ecotourism and adventure travel are increasingly popular, the incidence of tick-borne rickettsioses among travellers is likely to continue to increase.
All patients with rickettsial infections reported by Dittrich and colleagues presented with fever at admission, and few patients presented with typical eschars of inoculation (only 3·6% of patients with murine typhus and 6·7% with scrub typhus). 1 This fi nding might be a result of poor awareness about pathognomonic signs of rickettsioses among clinicians rather than an absence of such disease.
Rickettsioses are treatable but remain underestimated. Besides murine typhus and scrub typhus, tick-borne spotted fever group rickettsioses cause much fever of unknown origin in tropical countries. 6 R felis is one such rickettsia; it has been detected worldwide in arthropod hosts (mainly fl eas), with the cat fl ea Ctenocephalides felis the only confi rmed biological vector. A growing number of reports implicate R felis in human disease, particularly in the tropics. It has been detected in 3-4% of cases of fever of unknown origin in rural Mali 7 and Kenya, 8 and 6% of cases in rural Senegal. 7 R felis has also been detected in mosquitoes 7 and it is common in countries in Africa with high prevalences of malaria. 7 Cases of R felis infection in Thailand have also been reported. 9 Collaborations of investigators in the tropics combined with powerful diagnostic methods have increased the recognition of neglected pathogens in patients with acute undiff erentiated fever. However, when studying the causes of fever of unknown origin, inclusion of local negative controls is essential because the incidence of many pathogens in these regions is totally diff erent from their incidence in Europe, changing the predictive value of diagnostic tests. The lack of local negative controls causes false positives because of the commonness of asymptomatic forms of infectious diseases; therefore, European controls cannot be used to evaluate the specifi city of diagnostic techniques in tropical countries.
Rickettsioses are severe diseases that can be fatal, yet in Dittrich and colleagues' study few patients with scrub and murine typhus received doxycycline (55% and 39%, respectively). 1 As a consequence, the use of an empirical doxycycline treatment for patients with fever of unknown origin should be discussed, especially when empirical treatment with β-lactams has failed or in cases with severe clinical presentation. Doxycycline is used for malaria prophylaxis in travellers. Only two prospective randomised studies 10 have shown the eff ectiveness of prophylactic doxycycline to prevent scrub typhus. Designing comparative studies to test the drug's eff ectiveness for preventing scrub typhus would be diffi cult. However, we propose that doxycycline be given as a priority for chemoprophylaxis against malaria in travellers to the tropics to thereby protect against rickettsioses and leptospirosis.