Coxiella burnetii and Bartonella species serology of febrile patients with an established infectious or inflammatory diagnosis in Sudan, Nepal, and Cambodia

ABSTRACT

Coxiella burnetii and Bartonella species cause febrile illness and infective endocarditis in low- and middle-income countries (LMICs). This study investigated whether seropositivity to C. burnetii or Bartonella could be detected among patients with persistent fever for which an infectious or inflammatory etiological diagnosis had been previously established in three LMICs. Our study tested sera from Cambodian, Nepalese, and Sudanese participants using indirect immunofluorescent antibody assays (IFA) for C. burnetii and Bartonella. Seropositivity rates for both pathogens were assessed across tropical and inflammatory etiologies of fever and compared to ubiquitous bacterial infections considered as a “reference group,” as they were not expected to cause serologic cross-reactivity. A total of 1,313 individuals underwent IFA, including 560/1,313 (42.7%) from Sudan, 432 (32.9%) from Nepal, and 321 (24.5%) from Cambodia. Overall, 57 (4.3%) and 60 (4.6%) participants tested positive for C. burnetii and Bartonella species, respectively. Forty-four (3.4%) individuals tested positive for both C. burnetii and Bartonella species (75.4% positive agreement). C. burnetii positivity did not differ significantly between the three countries (P = 0.44), while Bartonella seropositivity was predominantly identified in Nepal (P < 0.001). Compared to the reference group, C. burnetii and Bartonella seropositivity were more common among participants with visceral leishmaniasis, P. falciparum malaria, leptospirosis, brucellosis, scrub typhus, and systemic lupus erythematosus (SLE), though only statistically significant for the latter two diagnoses. Further studies are necessary to investigate C. burnetii and Bartonella seropositivity in LMICs and to disentangle cross-reactivity, previous infection, or co-infection.

IMPORTANCE

Coxiella burnetii and Bartonella spp. are important but under-recognized causes of febrile illness and infective endocarditis in low- and middle-income countries (LMICs). This study evaluated the seroprevalence of these pathogens among patients with confirmed causes of persistent fever in Sudan, Nepal, and Cambodia. Despite the diagnostic utility of serologic testing for these infections, its performance in LMICs—where co-infections and background seropositivity are common—remains poorly characterized. The findings suggest notable seropositivity for C. burnetii and Bartonella among patients with a set of tropical and inflammatory diagnoses, including visceral leishmaniasis, Plasmodium falciparum malaria, leptospirosis, brucellosis, scrub typhus, and systemic lupus erythematosus. These results highlight the potential for cross-reactivity and underscore the need for context-specific validation. Enhanced understanding of serologic test characteristics is essential for accurate diagnosis in resource-limited settings.

INTRODUCTION

Infections due to zoonotic and vector-borne diseases, including those caused by Coxiella burnetii and Bartonella species, are common in low- and middle-income countries (LMICs) (1–4). Both bacteria cause febrile illness and infective endocarditis across low-resource areas globally (1–3, 5, 6). Transmission of C. burnetii, the etiologic agent of Q fever, predominantly occurs via inhaling aerosols contaminated by the excreta and birth products of infected animals, though transmission via contaminated animal product ingestion may also occur (2). Over 40 different Bartonella species infect a wide range of animal reservoirs (7). Bartonella species are primarily transmitted via hematophagous arthropod vectors, including lice, fleas, ticks, and sand flies (7, 8). The most common Bartonella species to infect humans are the louse-borne B. quintana (historically, the agent of trench fever), the flea-borne B. henselae (cat scratch disease), and the sandfly-borne B. bacilliformis (Carrion’s disease/Oroya fever), though human infections due to other Bartonella species have been described (7, 9).

C. burnetii and Bartonella species are fastidious bacilli that cannot be identified by routine culture on agar with 5-day incubation (2, 7). These pathogens are thus considered culture-negative bacteria, necessitating specialized microbiologic techniques, including serologic and molecular methods for identification (2, 9). Symptoms due to both infections are frequently nonspecific, which further complicates diagnosis (2, 7). In light of these considerations, the revised modified Duke criteria include seropositivity to C. burnetii and Bartonella species with immunoglobulin G (IgG) titers ≥ 1:800 as major criteria for diagnosing infective endocarditis (10, 11).

Serologic testing for C. burnetii and Bartonella is mainly performed via indirect immunofluorescence antibody assays (IFA), which are limited by low specificity. Cross-reactivity occurs between different species within the Bartonella genus and between Bartonella species and C. burnetii (12–14). Furthermore, C. burnetii IFA cross-reacts with Legionella, Chlamydia, Ehrlichia, Anaplasma, and Rickettsia serologies (15–19).

Despite the central role of serology in diagnosing C. burnetii and Bartonella infections and the prevalence of these diseases in LMICs, few studies evaluate the performance of C. burnetii and Bartonella serologic testing in LMICs where other infections are common. A thorough understanding of Coxiella burnetii and Bartonella serologic test characteristics is crucial for accurately diagnosing infection and interpreting seroprevalence studies in low-resource settings.

The NIDIAG consortium (“Better DIAGnosis of Neglected Infectious Diseases,” NIDIAG) was established in 2010 to address knowledge gaps in infectious syndromes in LMICs, including persistent fever in Sudan, Nepal, Cambodia, and the Democratic Republic of the Congo (DRC) (20). This prospective multicentric study focused on severe infections known to cause persistent fever (e.g., malaria, tuberculosis, visceral leishmaniasis, enteric fever, brucellosis, melioidosis, leptospirosis) but also included other infectious syndromes (e.g., community-acquired pneumonia, skin and soft tissue infections, bacterial meningitis, cholangitis) (20, 21). NIDIAG samples were analyzed in national and international (France, Belgium) reference laboratories, and the remaining samples were biobanked in French and Belgian laboratories (21). Testing for Coxiella burnetii and Bartonella species was not included in the original NIDIAG protocol (21). Our study aimed to assess Coxiella burnetii and Bartonella seropositivity among biobanked serum samples from the NIDIAG fever cohort with existing diagnoses as determined by the original NIDIAG fever studies (20, 21).

MATERIALS AND METHODS

Study design and populations

The original NIDIAG fever samples were collected from January 2013 to October 2014 from the following study sites in Sudan, Nepal, Cambodia, and the DRC: a rural hospital in Gedaref State, Sudan; a rural district hospital and an outpatient health center in Mosango, Kwilu province, DRC; a rural district hospital in Dhankuta and a university hospital in Dharan, Nepal; and a referral-level Sihanouk hospital center of Hope in Phnom Penh, Cambodia (20). The NIDIAG fever study compiled clinical and diagnostic data from consenting patients over 5 years of age presenting with over 7 days of fever (21). Patients requiring initial intensive care were excluded (21). Insufficient volume of serum originating from the DRC precluded analysis; the included samples originated from Nepal, Cambodia, and Sudan. All samples of patients in whom an infectious or inflammatory condition had been diagnosed during the NIDIAG studies were included, provided that sufficient serum was available for testing. Original NIDIAG diagnoses included the following: ubiquitous culture-positive bacterial infections (further considered as the reference group as these infections were not expected to interfere with C. burnetii or Bartonella spp. serology; see more detail below); tropical diseases such as malaria, visceral leishmaniasis, leptospirosis, etc.; and one autoimmune condition, systemic lupus erythematosus (SLE). The latter two groups were hypothesized to demonstrate serologic cross-reactivity to C. burnetii and/or Bartonella species. No samples were previously tested for C. burnetii and Bartonella species in the initial studies.

Due to the persistence of immunoglobulin G (IgG) to C. burnetii and Bartonella antigens after infection, the reference group was created as a proxy for background seroprevalence. This group comprised all participants diagnosed with ubiquitous extracellular bacterial infections likely accounting for the presenting fever, but not typically associated with the clinical features of C. burnetii infection or bartonellosis, and not known to cause serologic cross-reactivity with these pathogens. All participants meeting the following criteria were included in the reference group: diagnosed with a common infectious syndrome that is not typically associated with C. burnetii or Bartonella infection; associated with a cultivated bacterial pathogen with no known serologic cross-reactivity to C. burnetii or Bartonella species; and had remaining serum available for testing. The reference group included the following syndromes: bacterial meningitis, cholangitis/cholecystitis, tonsillitis/pharyngitis, skin and soft tissue infection, bacterial gastroenteritis/dysentery, bacterial liver abscess, appendicitis, spontaneous bacterial peritonitis (SBP), pyelonephritis, and documented bacteremia with non-fastidious organisms (other than Salmonella and Brucella species; cases of enteric fever and brucellosis were analyzed separately due to their facultatively intracellular localization, similar to Bartonella spp.). Serologic positivity to C. burnetii and Bartonella spp. in the reference group was considered to represent positivity from previous exposure rather than an association with the participant’s presenting fever.

Indirect immunofluorescent antibody assays

An in-house IFA was performed on stored serum samples to assess antibody response to C. burnetii antigens (phase I and II lipopolysaccharides) and Bartonella antigens (B. quintana and B. henselae antigens), as previously described (2, 9, 12, 22). Sera were diluted in phosphate-buffered saline with 3% powdered milk to mitigate nonspecific antibody fixation (12). Initial screening was conducted using titers of 1:50 and 1:100. Titers ≥ 1:100 were considered positive and underwent further doubling dilutions (9, 12, 23). This cut-off is the established positivity threshold for the in-house IFA assay and is considered equivalent to the 1:128 cut-off used in the FOCUS-DiaSorin IFA assay (9, 12, 23). Samples exhibiting a titer of 1:50 but testing negative at 1:100 were classified as indeterminate and were not subjected to further serial dilutions (23). For C. burnetii, IFA screening was performed with a combination of anti-phase I and II antigens, and positive sera subsequently underwent serial dilution and testing for anti-phase I and II IgG. For Bartonella, screening was performed with B. henselae and B. quintana antigens. Samples with IgG titers ≥ 1:100 to either B. quintana or B. henselae antigens were counted as a positive Bartonella result.

Analysis and interpretation

Participants with C. burnetii anti-phase I IgG titers ≥ 1:800 or anti-Bartonella IgG titers ≥ 1:800 were described separately and linked to the original clinical descriptions. These thresholds were chosen for focused analysis as they are major diagnostic criteria for infective endocarditis (10). Due to known serologic cross-reactivity between C. burnetii and Bartonella species, the antibody response with the highest titer was deemed to indicate exposure to the causative genus. Separate analyses compared IFA results from the reference group with results from participants diagnosed with visceral leishmaniasis (VL), tuberculosis, malaria (Plasmodium falciparum and Plasmodium vivax), leptospirosis, melioidosis, brucellosis, scrub typhus, enteric fever, and SLE as diagnosed using the original NIDIAG fever protocol (21).

Visceral leishmaniasis (VL) was diagnosed using a combination of either promastigote culture (Novy–McNeal–Nicolle medium), microscopic visualization of Giemsa-stained amastigotes, or direct agglutination test (DAT) and response to treatment (21). TB was diagnosed using either microscopic visualization of acid-fast bacilli (Ziehl-Neelsen) and/or the GeneXpert Mtb/Rif assay. Malaria (P. falciparum and P. vivax) was diagnosed using microscopy of Giemsa-stained thick and thin films and/or rapid diagnostics tests targeting both the pan-pLDH and HRP2/3 antigens (SD Malaria Ag P.f/Pan and CareStart malaria) (21, 24, 25). Leptospirosis was diagnosed using either polymerase chain reaction (PCR) and/or microscopic agglutination tests (MAT). Melioidosis was diagnosed by a positive bacterial culture for Burkholderia pseudomallei. Brucellosis was diagnosed using either bacteriologic culture and/or serology. Scrub typhus was diagnosed using serology detecting immunoglobulin M (IgM) to Orientia tsutsugamushi antigens (21). Enteric fever was diagnosed by blood culture growth of Salmonella enterica serotypes Typhi and Paratyphi. SLE was diagnosed by positive antinuclear antibody test (ANA) in combination with clinical symptoms compatible with SLE, as determined by the treating clinician.

Statistical methods

Statistics were performed using R version 4.2.2 software (2022-10-31). The chi-squared and Fisher’s exact tests were used to compare categorical variables, with the latter used for smaller sample sizes. The chi-squared test was performed to compare seropositivity between the three countries to assess whether seropositivity was equal or different between the countries. Fisher’s exact test was used to compare seropositivity rates between the disease and reference groups. Unless otherwise specified, comparisons were performed using positive serologies (titers ≥ 1:100). Kappa coefficient and positive and negative agreements were calculated to assess potential cross-reactivity between C. burnetii and Bartonella IFAs. Relative risk (RR) ratios were calculated to determine the probability of seropositivity in the diseased group compared to the reference group. Attributable risk percent (AR%) was calculated to estimate the percentage of serologic positivity attributable to exposure to the alternate disease. The threshold for statistical significance was set at a P-value of <0.05.

RESULTS

Overall seropositivity and country-specific seropositivity rates

A total of 1,313 individuals, including the reference group, underwent serologic testing for C. burnetii and Bartonella species. This involved 560 (42.7%) individuals from Sudan, 432 (32.9%) from Nepal, and 321 (24.5%) from Cambodia (Table 1). Overall, 57 (4.3%) individuals tested positive for C. burnetii, 182 (13.9%) individuals had indeterminate testing, and 1,074 (81.8%) tested negative. Seropositivity rates were 5.2% (29/560), 3.7% (16/432), and 3.7% (12/321) for Sudan, Nepal, and Cambodia, respectively (Table 1). C. burnetii positivity did not differ significantly between the three countries (X² [df = 2, N = 1,313] =1.65, P = 0.44). Regarding testing for Bartonella, 60 (4.6%) individuals were positive, 73 (5.6%) had indeterminate IFA, and 1,181 (89.9%) tested negative. Bartonella seropositivity rates were 1.6% (9/560), 11.3% (49/432), and 0.6% (2/321) for Sudan, Nepal, and Cambodia, respectively. Bartonella seropositivity differed significantly between the three countries (X² [2, N = 1,313 ]=68.18, P < 0.001).

TABLE 1.

C. burnetii and Bartonella species seropositivity among febrile individuals in Sudan, Nepal, and Cambodia according to country^a

	Country				P value
	Sudan	Nepal	Cambodia	Total
N positive C. burnetii IFA (seroprev)	29 (5.2%)	16 (3.7%)	12 (3.7%)	57 (4.3%)	0.44
N positive Bartonella IFA (seroprev)	9 (1.6%)	49 (11.3%)	2 (0.6%)	60 (4.6%)	<0.001
Total	560	432	321	1,313

^a N positive C. burnetii indirect immunofluorescent antibody assay (IFA, seroprev): number of participants with positive C. burnetii IFA and associated seroprevalence rate from the affiliated country. N positive Bartonella IFA (seroprev): number of participants with positive Bartonella spp. IFA with the associated seroprevalence rate from the affiliated country. P-value: P-value of chi-squared statistic comparing seropositivity rates between the three countries.

Agreement between C. burnetii and Bartonella IFA

There were 1,024 (78.0%) agreements between C. burnetii and Bartonella IFA results (kappa coefficient = 0.14, 95% CI: 0.08–0.19), indicating slight agreement. Percent positive and negative agreements were 75.4% (95% CI: 93.0%–77.7%) and 91.7% (95% CI: 90.1%–93.1%), respectively. Forty-four individuals tested positive for both C. burnetii and Bartonella species. Agreement increased when indeterminate results were considered positive, with a resulting kappa coefficient of 0.41 (95% CI: 0.32–0.50).

Reference group

A total of 67 cases met criteria to be included in the reference group, including 14 (20.9%) cases of skin and soft tissue infection, 13 (19.4%) tonsillitis/pharyngitis, 10 (14.9%) bacteremia, 10 (14.9%) bacterial gastroenteritis/dysentery, 6 (9.0%) bacterial meningitis, 5 (7.5%) spontaneous bacterial peritonitis, 4 (6.0%) cholangitis/cholecystitis, 2 (3.0%) cases of appendicitis, 2 (3.0%) bacterial liver abscesses, and 1 (1.5%) pyelonephritis. Most cases (37/67 = 55.2%) were from Cambodia. 27 (40.3%) were from Sudan and 3 (4.5%) from Nepal. Two (3.0%) individuals in the reference group screened positive for C. burnetii, and 9 (13.4%) had indeterminate serology. One individual (1.5%) screened positive for Bartonella species, and one individual (1.5%) was Bartonella indeterminate.

Visceral leishmaniasis

One hundred nine participants were diagnosed with visceral leishmaniasis. Of these 109 participants, 6 (5.5%) tested positive for C. burnetii, 15 (13.8%) were indeterminate, and 88 (80.7%) tested negative (Table 2). C. burnetii seropositivity was more common among individuals with VL than in the reference group (RR = 1.844, 95% CI: 0.38–88.87), though this difference was not statistically significant (P = 0.47). Ten (9.2%) and eight (7.3%) participants with VL tested positive and indeterminate for Bartonella IFA, respectively (Table 3). Bartonella seropositivity was more common among individuals with VL than in the control group (RR = 6.15, 95% CI: 0.81–46.94). This difference tends toward statistical significance (P = 0.051). Two participants diagnosed with VL had C. burnetii IFA titers > 1:800 (Table 4).

TABLE 2.

C. burnetii seropositivity by disease and compared to the reference group^a

	C. burnetii seropositive/total tested (%)	C. burnetii indeterminate/total tested (%)	Relative risk ratio (95% CI)	Attributable risk percent (95% CI)	P value
Reference group	2/67 (3.0%)	9/67 (13.4%)	N/Ab	N/A	N/A
VL	6/109 (5.5%)	15/109 (13.8%)	1.84 (0.4–88.9)	45.8% (−69.4 to 160.9)	0.47
TB	2/93 (2.2%)	7/93 (7.5%)	0.78 (0.1–5.4)	−28.36 (−247.5 to 190.8%)	1
P.f malaria	2/46 (4.3%)	17/46 (37.0%)	2.09 (0.3–14.2)	53.2% (80.4–184.9)	0.59
Leptospirosis	5/52 (9.6%)	5/52 (9.6%)	3.56 (0.7–17.6)	71.94% (12.2%–156.1%)	0.12
Melioidosis	0/13 (0.0%)	1/13 (7.7%)	0	N/A	1
Brucellosis	2/18 (11.1%)	5/18 (27.8%)	3.722 (0.6–24.7)	73.13 (–26.0 to 172.3%)	0.12
Scrub typhus	0/11 (0.0%)	0/11 (0.0%)	0	N/A	1
Enteric fever	2/22 (9.1%)	0/22 (0.0%)	3.04 (0.5–20.4)	AR%: 67.2 (−42.6% to 176.9%)	0.25
SLE	3/7 (42.9%)	0/7 (0.0%)	14.36 (2.9–71.9)	93.03 (47.4%–138.6%)	0.005c

^a C. burnetii seropositive/total tested (%): number of participants who tested positive for C. burnetii (Cb) (titers ≥ 1:100) over total, with percent positivity in parentheses. C. burnetii indeterminant/total tested (%): number of participants who tested indeterminant for C. burnetii (titers = 1:50) over total with percent positivity. Relative risk ratio (95% CI): relative risk ratio of positive cases from associated disease and control with 95% CI (CI). Attributable risk percent (95% CI): attributable risk percent with 95% CI. P value: Fisher’s exact test P value. VL, visceral leishmaniasis; P. f malaria, Plasmodium falciparum malaria; SLE, systemic lupus erythematosus.

^b N/A, not applicable.

^c Bold value indicates statistical significance.

TABLE 3.

Bartonella seropositivity by disease compared to the reference group^a,c

	Bartonella seropositive/total tested (%)	Bartonella indeterminant/total tested (%)	Relative risk ratio (95% CI)	Attributable risk percent (95% CI)	P value
Reference group	1/67 (1.5%)	1/67 (1.5%)	N/Ab	N/A	N/A
VL	10/109 (9.2%)	9/109 (8.3%)	6.15 (0.805–46.94)	83.73% (3.45%–164.01%)	0.05
TB	0/93 (0.0%)	4/93 (4.3%)	0	N/A	0.45
P.f malaria	1/46 (2.2%)	3/46 (6.5%)	1.42 (0.09–22.2)	29.85% (−200.2% to 259.95%)	1
Leptospirosis	2/52 (3.8%)	1/52 (1.9%)	2.62 (0.244–28.18)	61.94 (−84.26 to 208.1365)	0.58
Melioidosis	0/13 (0.0%)	1/13 (7.7%)	N/A	N/A	1
Brucellosis	1/18 (5.5%)	1/18 (5.5%)	3.722 (0.24–56.648)	73.13% (−68.83% to 215.03%)	0.28
Scrub typhus	7/11 (63.6%)	2/11 (18.5%)	42.63 (5.794–313.76	97.65 (67.26%–128.05%)	0.001
Enteric fever	0/22 (0.0%)	1/22 (4.5%)	0	N/A	1
SLE	2/7 (28.7%)	1/7 (14.3%)	19.14 (1.97–185.42)	94.78 (41.03%–148.5%)	0.02

^a Bartonella Seropositive/total tested (%): number of participants who tested positive for Bartonella species (titers ≥ 1:100) over total testing, with percent positivity. Bartonella indeterminant/total tested (%): number of participants who tested indeterminant for Bartonella (titers = 1:50) over total with percent positivity. Relative risk ratio (95% CI): relative risk ratio of positive cases from associated disease and control with 95% CI. Attributable risk percent (95% CI): Attributable risk percent with 95% CI. P value: Fisher’s exact test P value. VL, visceral leishmaniasis; P. f malaria, Plasmodium falciparum malaria; SLE, systemic lupus erythematosus.

^b N/A, not applicable.

^c Bold values indicate statistical significance.

TABLE 4.

Participants with C. burnetii anti-phase I IgG titers ≥ 1:800 or Bartonella IgG titers ≥ 1:800^a

Case number	Age (sex, occupation)	Original diagnosis (diagnostic modality)	Country	C. burnetii or Bartonella IFA result	Cardiopulmonary auscultation	Death
1	28 (M, farmer)	VL (Culture, DAT, and microscopy)	Nepal	C. burnetii anti-phase I IgG 1:800 and anti-phase II IgG titers 1:400	Normal examination (no murmur)	No
2	47 (F, unspecified worker)	VL (Culture and microscopy)	Sudan	C. burnetii anti-phase I IgG titers of 1:1,600 and anti-phase II IgG 1:800	Normal examination (no murmur)	No
3	78 (M, dependent)	TB (sputum GeneXpert)	Nepal	C. burnetii anti-phase I IgG titers 1:800 and anti-phase II IgG titers 1:200	Normal examination (no murmur)	No
4	44 (M, farmer)	Leptospirosis (MAT)	Sudan	C. burnetii anti-phase I IgG titers of 1:3,200 and anti-phase II IgG titers of 1:1,600	Normal examination (no murmur)	No
5	35 (F, housewife)	Brucellosis (positive Brucella species blood cultures)	Sudan	B. quintana 1:800 and B. henselae 1:200	Normal examination (no murmur)	No

^a M, male; F, female; VL, visceral leishmaniasis; TB, tuberculosis; DAT, direct agglutination test; MAT, microscopic agglutination test; Death, determined at 1-month follow-up.

Tuberculosis

Ninety-three participants were diagnosed with TB. Two (2.2%) tested positive for C. burnetii, and 7 (7.5%) had indeterminate serologies (Table 2). C. burnetii seropositivity rate was similar among individuals diagnosed with TB and those in the reference group (P = 1). No participants with TB tested positive for Bartonella IFA, though four (4.3%) had indeterminate Bartonella serology (Table 3). One Nepalese participant diagnosed with TB had C. burnetii anti-phase I IgG titers of 1:800 and anti-phase II IgG titers of 1:200 (Table 4).

Malaria

Fifty participants were diagnosed with malaria. There were 46 (92.0%) cases of P. falciparum and 4 (8.0%) cases of P. vivax. All cases of P. vivax were negative for both C. burnetii and Bartonella serology, except for one participant with indeterminate C. burnetii serology (Tables 2 and 3). Of the 46 cases of P. falciparum malaria, 2 (4.3%) tested positive for C. burnetii, and 17 (37.0%) were indeterminate. C. burnetii seropositivity was more common among individuals with malaria than in the control group (RR = 2.09, 95% CI: 0.309–14.20, AR% = 53.24%, 95% CI: 80.44–184.93), though this difference was not statistically significant (P = 0.59). The difference became significant when comparing indeterminate results (P = 0.008). Furthermore, when indeterminate results were treated as positive, statistical significance remained (P = 0.004). One (2.2%) participant with P. falciparum malaria tested positive for Bartonella, and three (6.5%) had indeterminate Bartonella IFA. Bartonella seropositivity was equally common among individuals with malaria and the control group (P = 1).

Leptospirosis

Fifty-two participants were diagnosed with leptospirosis. Of these 52 participants, 5 (9.6%) tested positive for C. burnetii and 5 (9.6%) were indeterminate (Table 2). C. burnetii seropositivity was more common among individuals with leptospirosis than in the control group (RR = 3.56383, 95% CI: 0.72–17.59), though not statistically significant (P = 0.12). Two (3.8%) participants with leptospirosis tested positive for Bartonella, and one (1.9%) had indeterminate Bartonella IFA (Table 3). One participant from Sudan had C. burnetii anti-phase I IgG titers of 1:3,200 and anti-phase II IgG titers of 1:1,600 (Table 4).

Melioidosis

Thirteen participants were diagnosed with melioidosis; none tested positive for C. burnetii nor Bartonella (Tables 2 and 3). One individual (7.7%) had an indeterminate IFA for C. burnetii, and another had an indeterminate IFA for Bartonella species.

Brucellosis

Eighteen participants were diagnosed with brucellosis. Two (11.1%) tested positive for C. burnetii, and 5 (27.8%) were C. burnetii indeterminate (Table 2). C. burnetii seropositivity was more common among individuals with brucellosis than in the control group (RR = 3.722, 95% CI: 0.56–24.698), though this difference was not statistically significant (P = 0.12). One (5.5%) participant with brucellosis tested positive for Bartonella, and one (5.5%) had an indeterminate Bartonella IFA (Table 3). One participant from Sudan with positive Brucella species blood cultures had anti-Bartonella quintana IgG titers of 1:800 (Table 4).

Scrub typhus

Eleven participants were diagnosed with scrub typhus. No participants diagnosed with scrub typhus had positive or indeterminate C. burnetii serologies. However, the majority of participants diagnosed with scrub typhus (7/11 = 63.6%) had positive Bartonella IFAs, all of which had low positive titers of 1:100. Bartonella seropositivity was significantly more elevated among individuals diagnosed with scrub typhus than the reference group (P < 0.001, RR = 42.63, 95% CI: 5.794–313.76).

Enteric fever

Twenty-two participants diagnosed with enteric (typhoid and paratyphoid) fever had C. burnetii and Bartonella IFA testing. Two (9.1%) participants had positive C. burnetii serology (Table 2). C. burnetii seropositivity was similar among individuals diagnosed with enteric compared to the control group (P = 0.25). One (4.5%) individual diagnosed with enteric fever had indeterminate Bartonella IFA results (Table 3). None had positive Bartonella serology.

Systemic lupus erythematosus

Seven participants were diagnosed with SLE. Three (42.9%) participants had positive C. burnetii serology (Table 2). C. burnetii seropositivity was more common among individuals diagnosed with SLE compared to the control group (P = 0.005, RR = 14.36, 95% CI: 2.86–71.89). Two (28.7%) individuals diagnosed with SLE had positive Bartonella IFA results (Table 3). Bartonella seropositivity was more common among individuals diagnosed with SLE compared to the control group (P = 0.017, RR = 19.14, 95% CI: 1.97–185.42).

DISCUSSION

Our findings suggest that seropositivity to C. burnetii and Bartonella species may be commonly identified among patients with persistent fever in Sudan, Nepal, and Cambodia. Seropositivity for either genus may be more frequently detected among cases with certain tropical or autoimmune etiologies, compared to those diagnosed with ubiquitous bacterial syndromes, though further studies with larger sample sizes are required to confirm this observation.

Seropositivity rates in our study exceed those described in most high-income countries (2, 12, 26). There was no statistically significant difference in C. burnetii seropositivity between Sudan, Nepal, and Cambodia. Most participants with C. burnetii seropositivity originated from Sudan, a country where C. burnetii infection is prevalent in livestock and human cases have been described (1, 27). Bartonella seropositivity was significantly more common among Nepalese participants. This finding may be linked to the endemicity of B. quintana in Nepal (28). Lice in Nepal commonly harbor B. quintana, and previous cases of B. quintana endocarditis have been described among Nepalese individuals (3, 28, 29). The association between altitude elevation and increased rates of B. quintana infection has been documented elsewhere and remains ecologically plausible due to body lice’s niche in layers of clothing and the sartorial requirements of high-altitude settings (30).

Among 60 Bartonella-seropositive individuals, 44 were also seropositive for C. burnetii, suggesting cross-reactivity. While cross-reactivity between C. burnetii and Bartonella IFAs is well established, our study indicates possible cross-reactivity of these assays to other infections in LMICs (12). C. burnetii IFA may be falsely positive among individuals with brucellosis, leptospirosis, and SLE and demonstrate falsely indeterminate results among individuals with P. falciparum malaria. Bartonella IFA may be falsely positive among individuals with visceral leishmaniasis, scrub typhus, brucellosis, and SLE. Neither C. burnetii nor Bartonella seropositivity was associated with TB, melioidosis, or enteric fever. However, without cross-adsorption, protein immunoblotting, and molecular methods, our study could not distinguish serologic cross-reactivity from possible co-infection. Identifying the true cause of serological positivity is further complicated by the low sensitivity of molecular testing for C. burnetii and Bartonella species on peripheral blood specimens (2, 9, 31). Even if our study included PCR testing of paired blood samples, negative PCR results would be incapable of ruling out either disease (2, 9, 31).

The issue of serologic cross-reactivity may lead to misdiagnoses. In Tanzania, cases of C. burnetii and B. quintana disease confirmed by 16S rRNA metagenomic sequencing were initially misdiagnosed as “probable acute leptospirosis” due to positive MAT (32). Without confirmatory testing, it remains uncertain whether the NIDIAG cases of fever diagnosed as leptospirosis and scrub typhus by serology alone were infections due to these pathogens or whether C. burnetii, Bartonella species, or another undetermined etiology caused the fever. This highlights the need for improved diagnostic testing for culture-negative bacterial pathogens, especially in LMICs. A similar phenomenon may be at play for the NIDIAG participants diagnosed with probable SLE due to non-specific symptoms and a positive ANA. Both Bartonella and C. burnetii are known to cause auto-antibody production, which calls into question the veracity of these auto-immune diagnoses (33–35).

While seropositivity was present at low titers in most cases, a few cases demonstrated C. burnetii and Bartonella IFA positivity with elevated titers that would meet diagnostic criteria for infective endocarditis (10). Two cases of VL, one case of leptospirosis, and one case of TB had C. burnetii anti-phase I IgG ≥1:800. One case of culture-proven brucellosis had Bartonella titers of 1:800. While culture-negative bacterial pathogens disproportionately cause infective endocarditis in LMICs, co-infection of two unusual infections in a single individual warrants further interrogation. None of the cases with elevated serologies had abnormal cardiac examinations suggestive of underlying endocarditis, though this alone does not exclude true infection. Further prospective infective endocarditis studies are needed to validate the revised, modified Duke criteria in LMICs.

This study is subject to several limitations. Without cross-adsorption, protein immunoblot or specialized culture-based or molecular techniques, IFA positivity could indicate previous exposure or cross-reactivity to a different pathogen or co-infection. Inadequate remaining sera impeded our ability to compare paired acute and convalescent sera, as follow-up was challenging in many low-resource settings (2, 36). Certain disease comparisons had low sample sizes, and the reference group included few cases from Nepal. The limited number of individuals meeting criteria for the reference group precluded our ability to subdivide the reference group into three country-specific reference groups. This limitation impeded our ability to match disease etiology with country of origin, preventing us from addressing inter-country differences in disease-specific seropositivity.

Febrile individuals in Sudan, Nepal, and Cambodia may be commonly exposed to C. burnetii and Bartonella species. Serologic tests targeting these pathogens may cross-react with other tropical and autoimmune diseases. Further studies are necessary to improve diagnostics for culture-negative bacteria and elucidate the true burden of these infections in LMICs.

ETHICS APPROVAL

The NIDIAG study was initially registered at clinicaltrials.gov under the identifier NCT01766830 and was conducted in line with a specifically designed ethics Charter (www.NIDIAG.eu). The study was initially approved by the Institutional Review Board of the ITM under the number 12125818 (June 5, 2012), with a granted amendment for additional testing number 793/11 (January 30, 2024). The study was also approved by the Ethics Committee of the Antwerp University Hospital/ University of Antwerp (UZA/ UA) under the Belgium registration number B300201214571 (June 25, 2012) with granted approval for additional testing “Project ID 6871 - Edge n/a - BUN: B3002024000144” (August 12, 2024). The study was approved by ethical committees of partner institutions in the countries where data were collected: The Ethics Committee of the University of Khartoum (Khartoum, Sudan, FEM/DO/EC102; Submission: June 3, 2012; Approval: June 13, 2012; Update: January 18, 2024), the National Research Ethics Review Committee, Sudan (Khartoum, Sudan; Submission: September 5, 2012; Approval: November 18, 2012; Update: January 21, 2024), the Ethics Committee of B.P. Koirala Institute of Health Sciences (BPKIHS, Dharan, Nepal, 57/2012; Submission: May 29, 2012; Approval: July 3, 2012; Update: January 18, 2024), National Ethics Committee for Health Research of Cambodia (Phnom Penh, Cambodia, 134/2012; Submission: May 27, 2012; Approval: August 24, 2012; Update: January 20, 2024) and the University of Kinshasa School of Public Health (Kinshasa, Democratic Republic of the Congo, ESP/CE/016; Submission: December 4, 2012; Approval: January 14, 2013; Update: January 19, 2024). All study participants or their guardians gave written informed consent, and children <18 years also gave their assent. The original informed consent included a statement about sample storage for future research on infectious diseases. Samples from participants who opted for post-study destruction at the conclusion of the original NIDIAG protocol were excluded from biobanking. For further details, please see: https://nidiag.eu/.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.01675-25.

OPEN PEER REVIEW. reviewer-comments.pdf.

An accounting of the reviewer comments and feedback.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.