Abstract
As rickettsioses may be severe diseases and Rickettsia prowazekii is a potential agent of bioterrorism, highly efficient diagnostic techniques are required to detect rickettsiae in patients. We developed a nested PCR assay using single-use primers targeting single-use gene fragments present in the genomes of both Rickettsia conorii and R. prowazekii. We used this “suicide” PCR with DNA from 103 skin biopsy specimens from patients who definitely had a rickettiosis, 109 skin biopsy specimens from patients who possibly had a rickettsiosis, and 50 skin biopsy specimens from patients with nonrickettsial diseases. The suicide PCR detected “R. conorii conorii” in 38 biopsy specimens, R. africae in 28 biopsy specimens, R. slovaca in 12 biopsy specimens, “R. sibirica mongolotimonae” in 5 biopsy specimens, R. aeschlimannii in 2 biopsy specimens, and “R. conorii caspia” and “R. sibirica sibirica” in 1 biopsy specimen each. The technique had a specificity of 100% and a sensitivity of 68%. It was 2.2 times more sensitive than culture (P < 10−2) and 1.5 times more sensitive than regular PCR (P < 10−2). The efficacy of the suicide PCR was reduced by antibiotic therapy prior to biopsy (P < 10−2) and was increased when it was performed with eschar biopsy specimens (P = 0.03). We propose the use of the suicide PCR as a sensitive, specific, and versatile technique for improving the diagnosis of rickettsioses, especially when it is used on eschar biopsy specimens taken prior to antibiotic therapy.
The genus Rickettsia includes 19 species responsible for arthropod-borne infections causing mild to potentially lethal diseases (17). Among these, Rickettsia prowazekii, the agent of epidemic typhus, is considered a potential warfare agent and is classified on the B list of bioterrorism agents by the Centers for Diseases Control and Prevention. In addition to the pathogenic Rickettsia species that have been characterized, numerous rickettsiae have been found in ticks but have yet to be recognized as human pathogens. For example, R. parkeri has just been identified in a patient, 60 years after it was first isolated from ticks (13). PCR is a highly useful tool for the direct diagnosis of rickettsioses. However, the sensitivity of the usual PCR amplification with clinical specimens may be reduced by several factors, including the presence of DNA polymerase inhibitors, DNA degradation during sample transport, and antibiotic therapy prior to sampling. Also, although nested PCR techniques enhance the detection threshold of PCR through the use of a reamplification step, they have a major drawback, i.e., an elevated risk of contamination both by amplicons from previous assays with the same primers and by DNA carryover between the two amplification steps.
In an effort to improve the sensitivity of rickettsial DNA detection and to avoid false-positive amplifications, we designed a nested PCR technique with single-use primers targeting single-use DNA fragments selected after comparison of R. conorii and R. prowazekii genome sequences. We named this technique “suicide” PCR. It has previously been successful for the detection of Yersinia pestis in the remains of people who died from plague (18) and in a few patients with rickettsioses (21-23). In this report we compared the suicide PCR to culture and the regular PCR for the detection of rickettsiae in skin biopsy specimens submitted to our laboratory.
MATERIALS AND METHODS
Study design.
Single-use PCR primers were selected from fragments of rickettsial genes which had not been amplified previously. In 1998, we conducted a pilot study on 15 skin biopsy specimens from patients definitely considered to have rickettsial infections by use of the three diagnostic scores described below. Subsequently, we performed this technique annually on all skin biopsy specimens referred to our laboratory, with each assay incorporating a new primer set that was used only once. Patients from whom the biopsy specimens had been taken were classified as definitely having a rickettsiosis when they fulfilled one of the three diagnostic scores and/or when direct evidence of a Rickettsia species was found, as possibly having a rickettsiosis when they did not fulfill any of the diagnostic scores and had no evidence of a nonrickettsial disease, and as excluded from having a rickettsiosis when they had a proven nonrickettsial disease. The result for each positive specimen was confirmed by a second nested PCR with ompA-amplifying primers. The detection threshold of each of the suicide PCR primer sets was estimated a posteriori in vitro, after the patients' specimens had been tested. Finally, we estimated the sensitivity and specificity of the suicide PCR and compared them with those of culture and regular PCR.
Patients.
Following our 1998 pilot study, we performed suicide PCRs on all skin biopsy specimens referred to our laboratory between 1 January 1999 and 31 August 2003. Informed consent was obtained from all patients included in the present study. Patients were classified into three groups, i.e., patients with definite, possible, or no rickettsiosis, by using the case definitions described below. For each patient, epidemiological and clinical features were provided by the attending physician and a standardized questionnaire was completed. The information obtained included whether a tick bite was present, the geographical area from which the tick probably originated, any contact with animals, any underlying disease, any evidence of a nonrickettsial disease, the clinical symptoms [fever, headache, an inoculation eschar(s) and its location, the presence of a cutaneous rash and its type, presence of regional adenopathies], platelet count, transaminase levels, whether doxycycline and/or fluoroquinolones had been administered prior to the skin biopsy, and the outcome of the infection. We recorded the time of shipment during which the biopsy specimen was subject to thawing as ≥24 or <24 h. In addition, an early-phase serum sample was collected, and when possible, a second serum sample was collected 4 weeks later. Upon receipt, each biopsy specimen was trimmed aseptically into 1-mm3 fragments, which were used immediately for culture and DNA extraction.
Case definition.
The result of the suicide PCR was not considered an inclusion criterion, as it was the variable studied. All patients were screened by the use of three diagnostic scoring systems. Patients were classified as definitely having Mediterranean spotted fever (MSF), caused by “Rickettsia conorii conorii,” when they obtained a score of >25 by a previously reported diagnostic scoring system (24). Patients returning from sub-Saharan Africa were classified as definitely having African tick-bite fever (ATBF), caused by R. africae, when they fulfilled the criteria obtained by a previously described diagnostic scoring system for the disease (7, 21). Patients who became ill in Europe and who had an inoculation eschar on the scalp and cervical lymph node enlargement were classified as definitely having tick-borne lymphadenitis (TIBOLA), caused by R. slovaca, when they met one of the following criteria (23): (i) direct evidence of R. slovaca infection by culture and/or regular PCR; (ii) serology specific for a recent R. slovaca infection (seroconversion or presence of an immunoglobulin M [IgM] titer ≥1:32), with titers to R. slovaca greater than those to “R. conorii conorii” by at least 2 dilutions; and (iii) antibodies specific for R. slovaca by Western blotting or cross-adsorption assays, as described previously (23).
In addition, patients were classified as definitely having a rickettsiosis if they had direct evidence of infection with a Rickettsia sp. by culture or regular PCR.
Patients with a proven nonrickettsial disease were classified as not having a rickettsiosis and served as a negative control group.
Patients with a rash and/or an inoculation eschar who did not fulfill any of the inclusion criteria described above and who did not have a nonrickettsial disease were classified as possibly having a rickettsiosis.
Laboratory diagnosis. (i) Serological tests and culture.
Serum specimens were tested for the presence of IgG and IgM antibodies by microimmunofluorescence assay, as reported previously (9). For patients from the Mediterranean area and Europe, we tested for the presence of antibodies to “R. conorii conorii,” R. slovaca, R. helvetica, “R. sibirica mongolotimonae,” R. massiliae, R. aeschlimannii, R. felis, R. typhi, Coxiella burnetii, Francisella tularensis, and Borrelia burgdorferi. For patients returning from sub-Saharan Africa, we tested for the presence of antibodies to R. africae, “R. conorii conorii,” R. aeschlimannii, “R. sibirica mongolotimonae,” R. akari, R. felis, R. typhi, R. prowazekii, C. burnetii, and Borrelia recurrentis. Western blotting and cross-adsorption procedures were performed as described previously (10, 27) with the antigens cited above. Western blotting was performed with early-phase sera (27), while cross-adsorption was performed with test sera with IgG titers ≥1:128 (10).
Skin biopsy specimens were ground with sterile pestles and mortars, inoculated onto human embryonic lung fibroblasts by the shell vial technique (12), and cultured as described elsewhere (9). Isolates were identified by partial ompA amplification (5) and sequencing, as described below.
(ii) Molecular diagnosis.
DNA was extracted from the ground skin biopsy specimens, and the rickettsiae were isolated from the biopsy specimens with a QIAmp DNA Tissue kit (Qiagen GmbH, Hilden, Germany), according to the recommendations of the manufacturer. DNA samples extracted from skin biopsy specimens from patients from the negative control group were used as negative controls for PCR amplification. Ground biopsy samples were also stored at −80°C until DNA was extracted for suicide PCR. No sample was stored for longer than a year.
Regular PCR was performed within 7 days of receipt of a biopsy specimen by previously described methods (5, 25) with the gltA-derived primers CS877F and CS1258R and primers 190-70 and 190-701, which amplify the ompA gene. Amplifications were carried out in a PTC200 DNA thermal cycler (MJ Research, Waltham, Mass.).
The sequences of the suicide PCR primers were selected from conserved regions flanking 300- to 600-bp variable fragments of genes present in both R. conorii (GenBank accession number NC_003103) and R. prowazekii (GenBank accession number NC_000963) genomes. All primers and their characteristics are detailed in Table 1.
TABLE 1.
Characteristics of primers used for suicide PCR in our study
| Yr and primer name | Gene name (gene product) | Primer sequence (5′-3′) | Amplicon size (bp)a | Hybridization temp (°C) |
|---|---|---|---|---|
| 1998 | ||||
| lpxD1Fb | lpxD [UDP-3-O-(3-hydroxymyristoyl) glucosamine N-acyltransferase] | ATATAGGAATAGTTAAAATCGGC | 315 | 56 |
| lpxD1Rc | TAGATTGTCTATGCCAACCAT | |||
| lpxD2Fbd | AACACTACTATTGATAGAGG | 106 | 50 | |
| lpxD2Rcd | GCTACTTCCTGCTATACC | |||
| 1999 | ||||
| fabZ1Fb | fabZ [3R-hydroxymyristoyl-(acyl carrier protein)] | TTAGTAGATAGAGTACTTAAAATC | 264 | 56 |
| fabZ1Rc | CTTGAAAAYTTCCATACGTTAGC | |||
| fabZ2Fbd | AAATCGATCCTAACAAATCAATAA | 225 | 50 | |
| fabZ2Rcd | TCTTTGCTGATCAATAACAG | |||
| 1999 | ||||
| yidC1Fb | yidC (60-kDa inner membrane protein) | TAGAATCTGAATCACTTACCGG | 401 | 57 |
| yidC1Rc | CCGATAGGTCCTTGATGTAATA | |||
| yidC2Fbd | CATTAAAAGGACTTAGATTTG | 153 | 50 | |
| yid2Rcd | TTTCACTATCACTATTCCATA | |||
| 2000 | ||||
| pcnB1Fb | pcnB [poly(A) polymerase] | CTAATTCTAATTTAAAATYATTTGC | 401 | 56 |
| pcnB1Rc | AAAGATATTGAATGTAACGGYAG | |||
| pcnB2Fbd | CATAATTTTGKATAGAGAATAT | 342 | 50 | |
| pcnB2Rcd | TTTTGCCGAAGATGCCG | |||
| 2001 | ||||
| recF1Fb | recF (RecF protein) | GAACTAAAAACAGATAATACACC | 316 | 56 |
| recF1Rc | TCTATCAAGAAATTTTCTTCTATC | |||
| recF2Fbd | GGTAGCGGTAAAACTAATAT | 245 | 50 | |
| recF2Rcd | GCTTGTAAAGATTCCTTCCA | |||
| 2002 | ||||
| gltX1Fb | gltX (Glutamyl-tRNA synthetase) | AGTTATAACACGMTTTGCTCC | 533 | 56 |
| gltX1Rc | ATCAATATCATCAATAACTGAGC | |||
| gltX2Fbd | ACATTTAATCAATTAAGTCGT | 285 | 50 | |
| gltX2Rcd | TCTGCTCTTATCACTATTGG | |||
| 2003 | ||||
| mutL1Fb | mutL (DNA mismatch repair protein MutL) | GTAAAGGAATTAGTTGAAAATGCT | 420 | 55 |
| mutL1Rc | GGATGTGCTAGAGCGATTTT | |||
| mutL2Fbd | AACGTCATACAACTTCTAAG | 166 | 50 | |
| mutL2Rcd | AGTACCTTCATTATGAACAG |
The sizes of the amplicons are given with reference to the “R. conorii conorii” strain Malish (Seven) genome sequence (GenBank accession number NC_003103).
Forward primer.
Reverse primer.
Primer used for reamplification.
The suicide PCRs were conducted in a PTC200 DNA thermal cycler (MJ Research) with Elongase DNA polymerase (Gibco-BRL, Cergy-Pontoise, France). The 25-μl reaction mixtures consisted of the following (final concentrations): each of the four primers (1 pmol μl−1 each), deoxynucleoside triphosphates (dATP, dCTP, dGTP, and dTTP; 0.4 mM each), 2 μl of Elongase buffer A, 8 μl of Elongase buffer B, 1.2 μl of Elongase enzyme mixture, 2.5 μl of the DNA preparation, and sterile water. Amplifications were performed at the hybridization temperatures reported in Table 1 and with 35 cycles for the primary PCR and 35 cycles for the secondary PCR. The amplification was completed by maintaining the sample at 68°C for 10 min to allow complete extension of the PCR products. To avoid possibilities for contamination, we did not use positive controls in our suicide PCR assays. As negative controls, in addition to the DNA of skin biopsy specimens from patients classified as not having rickettsioses, for every eighth sample we used DNA, extracted as described above, from the heart valves of patients with degenerative valvulopathy. Testing was done blindly, and the PCR products were resolved on 1% agarose gels.
Positive PCR products were sequenced in both directions by using internal primers and the d-Rhodamine Terminator Cycle sequencing ready reaction kit, as described by the manufacturer. Sequencing products were resolved with an ABI 310 automated sequencer (Applied Biosystems). Multisequence alignment was done with CLUSTAL W software, version 1.81 (28).
It was only after we had tested all the patients' samples by suicide PCR that we determined the thresholds of the various suicide PCR assays that we used in the study. We used DNA from R. montanensis strain M/5-6. The number of organisms present in 10 μl of a Gimenez-stained rickettsial suspension was counted under a microscope at a magnification of ×1,000. DNA was extracted from an aliquot of 106 R. montanensis organisms, and serial 10-fold dilutions in water were used to determine the detection thresholds in a LightCycler thermal cycler (Roche Diagnostics, Mannheim Germany), according to the instructions of the manufacturer. To identify positive suicide PCR products, amplification and sequencing with each primer set were performed with DNA from “R. conorii conorii,” R. felis, R. slovaca, “R. sibirica mongolotimonae,” R. helvetica, R. massiliae, R. africae, and R. aeschlimannii. All skin biopsy specimens found to be positive by suicide PCR were tested by an additional nested PCR assay with the ompA-amplifying primer pairs AF1F-AF1R and AF2F-AF2R, as described previously (21). Negative controls were included after every seven test specimens, but we used no positive controls. Positive amplicons were sequenced as described above with the internal primers, and their sequences were compared to the ompA sequences available in GenBank.
We estimated the sensitivity of the suicide PCR in patients from our study with a definite diagnosis of rickettsiosis and its specificity among patients classified as not having a rickettsiosis.
Statistical tests.
Fisher's exact test was used to compare the sensitivities of the suicide PCR and those of culture or regular PCR. The Student t test was used to compare means. Observed differences were considered significant when P was <0.05 for two-tailed tests. We also studied the influence of each variable using univariate logistic regression analysis. To identify the variables independently associated with the PCR results, we included in a multivariate logistic regression analysis the sex ratio and variables for which the P value obtained in the univariate analysis was <0.4. STATA software (version 7.0; Stata Corporation, College Station, Tex.) was used for analysis.
RESULTS
Patients.
Overall, 262 skin biopsy specimens from 246 patients were included in our study: 15 biopsy specimens from 15 patients obtained in 1998, 55 biopsy specimens from 47 patients obtained in 1999, 59 biopsy specimens from 54 patients obtained in 2000, 70 biopsy specimens from 68 patients obtained in 2001, 39 biopsy specimens from 38 patients obtained in 2002, and 24 biopsy specimens from 24 patients obtained in 2003. The biopsy specimens were sent to the laboratory from France, Italy, Norway, Hungary, Austria, The Netherlands, Malta, and Tunisia. Ninety-three patients (103 biopsy specimens) were classified as definitely having a rickettsiosis (Table 2): 48 patients (49 biopsy specimens) had MSF, 25 patients (34 biopsy specimens) had ATBF; 12 patients (12 biopsy specimens) had TIBOLA, 5 patients were infected with “R. sibirica mongolotimonae” on the basis of a positive culture and/or regular PCR result, 1 patient was infected with “R. conorii caspia” on the basis of positive culture and PCR results (6), 1 patient had an “R. sibirica sibirica” infection on the basis of the regular PCR result, and 1 patient was infected with R. aeschlimannii on the basis of the result of PCR with serum and serology (20). Of the biopsy specimens from patients with definite rickettsioses, 32 of 103 (31.0%) were positive by culture and 47 of 103 (45.6%) were positive by regular PCR (Tables 2 and 3).
TABLE 2.
Distribution of patients included in the present study by diagnostic scores and the results of tests performed on skin biopsy specimens
| Patient score | No. of patients tested | No. of patients with:
|
|||
|---|---|---|---|---|---|
| Positive culture result | Positive regular PCR result | Positive suicide PCR result | Rickettsia species identifiedb | ||
| Definite cases | |||||
| MSF | 48 (49)a | 12 (12) | 21 (21) | 30 (30) | Rc |
| ATBF | 25 (34) | 13 (14) | 14 (16) | 22 (25) | Raf |
| TIBOLA | 12 (12) | 1 (1) | 3 (3) | 7 (7) | Rslo |
| “R. sibirica mongolotimonae” | 5 (5) | 4 (4) | 5 (5) | 5 (5) | Rm |
| “R. conorii caspia” | 1 (1) | 1 (1) | 1 (1) | 1 (1) | AFR |
| “R. sibirica sibirica” | 1 (1) | 0 | 1 (1) | 1 (1) | R sib |
| R. aeschlimannii | 1 (1) | 0 | 0 | 1 (1) | Raes |
| Possible cases | 104 (109) | 0 | 0 | 17 (17) | Rc (n = 8), Raf (n = 3), Rslo (n = 5), Raes (n = 1) |
| Other diagnoses | 49 (50) | 0 | 0 | 0 | |
For each variable, the number of biopsy specimens is indicated in parentheses.
Rc, “R. conorii conorii”; Raf, R. africae; Rslo, R. slovaca; Rm, “R. sibirica mongolotimonae”; Raes, R. aeschlimannii; AFR, “R. conorii caspia”; Rsib, “R. sibirica sibirica.”
TABLE 3.
Clinical and laboratory data for patients definitely with a rickettsiosis and patients possibly with a possible rickettsiosisa
| Variable | Patients definitely with a rickettsiosis (n = 93)
|
Patients possibly with a rickettiosis (n = 104) | ||||||
|---|---|---|---|---|---|---|---|---|
| MSF (n = 48) | ATBF (n = 25) | TIBOLA (n = 12) | Infection with Rm (n = 5) | Infection with AFR (n = 1)b | Infection with Raes (n = 1)b | Infection with Rsib (n = 1)b | ||
| Sex ratio (no. of M/no. of F) | 1.7 | 3.2 | 0.5 | 4 | 0 | 1 | 0 | 1.7 |
| Mean ± SD age (yr) | 51.1 ± 17.3 | 51.8 ± 14.3 | 34.2 ± 20.4 | 59.0 ± 8.6 | 36 | 36 | 52 | 42.2 ± 16.1 |
| Percentage of patients with: | ||||||||
| Tick bite | 60.4 | 64 | 100 | 0 | − | + | + | 48 |
| Inoculation eschar | 100 | 100 | 100 | 100 | + | + | + | 5 |
| Fever | 100 | 60 | 16.7 | 100 | + | + | + | 37.5 |
| Eruptive skin lesions | 100 | 64 | 8.3 | 60 | + | + | + | 72.1 |
| Enlarged lymph nodes | 0 | 68 | 100 | 60 | − | − | − | 19.4 |
| Liver cytolysis | 81.2 | 56 | 25 | 0 | − | + | + | 17.3 |
| Thrombocytopenia | 75 | 52 | 0 | 20 | − | − | − | 21.1 |
| Early antibiotic therapy | 18.7 | 16 | 25 | 0 | − | − | − | 20.2 |
| Eschar biopsy | 91.7 | 100 | 100 | 100 | + | + | + | 45.4 |
| Shipment delay >24 h | 39.6 | 36 | 16.7 | 0 | − | + | − | 42.3 |
| Antibodies to a Rickettsia sp. | 62.5 | 60 | 41.7 | 0 | + | − | − | 24 |
Abbreviations: M, male; F, female; Rc, “R. conorii conorii”; Raf, R. africae; Rslo, R. slovaca; Rm, “R. sibirica mongolotimonae”; Raes, R. aeschlimannii; AFR, “R. conorii caspia”; Rsib, “R. sibirica sibirica.”
Data for one patient are indicated in each of these three columns and thus are not given as means but as the value for the age and as the presence or absence for the other variables.
Forty-nine patients (50 biopsy specimens) with a proven nonrickettsial disease served as a negative control group. These included 26 patients from northern Italy with Lyme disease, determined on the basis of both positive culture and serology results (12) or positive serology results only (14), and no evidence of any another tick-borne infection. The other nonrickettsial diseases that we found among the control patients were Streptococcus pyogenes erysipelas in four patients; infected eczema in three patients; psoriaris in three patients; histologically confirmed bacillary angiomatosis in three patients, two of whom were PCR positive for Bartonella quintana; lymphoma in two patients; erythema multiforme in two patients; a spider bite in one patient; co-trimoxazole allergy in one patient; cellulitis in one patient; acute Q fever in one patient; chronic Q fever in one patient; and renal adenocarcinoma in one patient.
One hundred four patients (109 biopsy specimens) with a rash and/or an inoculation eschar were classified as possibly having a rickettsiosis (Table 2).
Suicide PCR.
All negative control samples in all assays except the 1999 assay targeting fabZ remained negative. Seven negative controls were positive by the assay targeting fabZ. Therefore, the results from this assay were not considered and a second assay targeting the yidC gene was conducted with the same specimens (Table 1). By use of such a strict interpretation of the results, the specificity of the suicide PCR was 100%.
The suicide PCR was positive for 70 of 103 skin biopsy specimens from patients classified as definitely having a rickettsiosis (68.0%) (Table 2), 17 of 109 biopsy specimens from patients with a possible rickettsiosis, and none of the biopsy specimens from excluded patients. Sequencing of positive suicide PCR products confirmed that the patients were infected with a Rickettsia species, but only pcnB sequences could identify “R. conorii conorii” (GenBank accession number AY509571), R. africae (GenBank accession number AY509572), R. slovaca (GenBank accession number AY509573), “R. sibirica mongolotimonae” (GenBank accession number AY509574), and R. aeschlimannii (GenBank accession number AY509575). Gene fragments amplified with the other primer sets were too conserved to allow identification of the infecting species. The results for all specimens positive by the suicide PCR were confirmed, and the species were identified by the ompA-based nested PCR.
Among patients with a definite rickettsiosis, the suicide PCR detected “R. conorii conorii” in 30 of 49 biopsy specimens (61.2%) from patients with MSF (Table 1), R. africae in 25 of 34 biopsy specimens from patients with ATBF (73.5%), R. slovaca in 7 of 12 biopsy specimens from patients with TIBOLA (58.3%), “R. sibirica mongolotimonae” in 5 of 5 biopsy specimens from patients previously proven to be infected with this rickettsia, R. aeschlimannii in the biopsy specimen from the patient with proven R. aeschlimannii infection, “R. conorii caspia” in the patient from whom this rickettsia was previously isolated, and “R. sibirica sibirica” in the patient with proven infection with this rickettsia. Among the 104 patients with a possible rickettsiosis, suicide PCR detected “R. conorii conorii” in 8 patients with MSF-like symptoms in the Mediterranean area, R. africae in 3 patients with ATBF-consistent symptoms in sub-Saharan Africa, “R. slovaca” in 5 patients with TIBOLA-like symptoms in Europe, and R. aeschlimannii in a patient from central France who exhibited a febrile maculopapular rash and an inoculation eschar to the leg following a tick bite (Table 1). A posteriori, we evaluated the detection thresholds of all primer sets except those amplifying fabZ. The primer sets specific for lpxD, yidC, pcnB, recF, gltX, and mutL detected 0.4 × 101, 1.9 × 101, 1.3 × 101, 0.8 × 101, 2.4 × 101, and 1.1 × 101 R. montanensis bacteria, respectively. In comparison, regular ompA-based PCR detected 1.6 × 102 bacteria. The overall sensitivity of the suicide PCR among biopsy specimens from patients with a definite rickettsiosis was 68% (70 of 103 biopsy specimens). The suicide PCR was 2.2 times more sensitive than culture (70 of 103 versus 32 of 103 biopsy specimens [P < 10−2]) and 1.5 times more sensitive than regular PCR (70 of 103 versus 47 of 103 biopsy specimens [P < 10−2]). By univariate analysis, the sex, age, type of disease, clinical presentation, and the presence of antibodies to a Rickettsia sp. had no influence on the suicide PCR results (Table 4). In contrast, significantly fewer positive results were obtained for biopsy specimens received thawed for ≥24 h (P = 0.045). However, by multivariate logistic regression analysis, this variable was not independently associated with the suicide PCR results (P = 0.7) (Table 4). In contrast, early antibiotic treatment, prior to the biopsy, was significantly associated with a reduced efficacy of the suicide PCR by both univariate and multivariate analyses (P < 10−2 for both). Likewise, eschar biopsy specimens were significantly more likely than other skin biopsy specimens to be positive by both univariate analysis (P = 0.03) and multivariate analysis (P = 0.045) (Table 4). Subsequently, we estimated the sensitivity of the technique among untreated patients to be 75.5% (65 of 86 patients), regardless of the type of skin biopsy specimen, and 78% (64 of 82 patients) among those from whom eschar biopsy specimens were tested.
TABLE 4.
Influences of sex, age, clinical presentation, type of disease, presence of antibodies, antibiotic therapy, and type of biopsy specimen on the suicide PCR results for patients definitely with a rickettsiosisa
| Variable | Positive suicide PCR result (n = 70) | Negative suicide PCR result (n = 33) | Univariate analysis
|
P by multivariate analysis | |
|---|---|---|---|---|---|
| P | Odds ratio | ||||
| Sex ratio (no. of M/no. of F) | 1.2 | 1.3 | 0.8 | 0.9 (0.4-2.0)b | 0.4 |
| Mean ± SD age (yr) | 49.1 ± 17.8 | 50.3 ± 15.3 | 0.7 | 0.9 (0.7-1.2) | ND |
| Percentage of patients with: | |||||
| MSF | 42.8 | 57.6 | 0.2 | 0.5 (0.2-1.4) | 0.08 |
| ATBF | 35.7 | 27.3 | 0.4 | 1.5 (0.5-4.1) | 0.9 |
| TIBOLA | 10.0 | 15.1 | 0.4 | 0.6 (0.2-2.5) | ND |
| Infection with Rm | 7.1 | 0 | 0.1 | Infinite | NDc |
| Infection with Raes | 1.4 | 0 | 0.7 | Infinite | ND |
| Infection with AFR | 1.4 | 0 | 0.7 | Infinite | ND |
| Infection with Rsib | 1.4 | 0 | 0.7 | Infinite | ND |
| Fever | 85.7 | 75.7 | 0.2 | 1.9 (0.6-6.1) | 0.1 |
| Lymph node | 40.0 | 39.4 | 0.9 | 1.0 (0.4-2.6) | ND |
| Rash | 75.7 | 75.7 | 1.0 | 1.0 (0.3-2.9) | ND |
| Type of skin biopsy | 98.6 | 87.9 | 0.03d | 9.5 (1.1-233.7) | 0.045 |
| Delay of shipment >24 h | 25.7 | 45.4 | 0.045 | 2.4 (0.9-6.3) | 0.7 |
| Early antibiotic therapy | 7.1 | 33.3 | <10−2 | 0.1 (0.04-0.5) | <10−2 |
| Elevated transaminase levels | 61.4 | 72.7 | 0.3 | 0.6 (0.2-1.6) | 0.1 |
| Thrombocytopenia | 55.7 | 51.5 | 0.7 | 1.2 (0.5-2.9) | ND |
| Presence of antibodies | 61.2 | 54.5 | 0.9 | 1.0 (0.4-2.7) | ND |
Abbreviations: M, male; F, female; Rm, “R. sibirica mongolotimonae”; Raes, R. aeschlimannii; AFR, “R. conorii caspia”; Rsib, “R. sibirica sibirica”; ND, not determined.
The values in parentheses are 95% confidence intervals.
Due to colinearity, this variable was not included in the multivariate analysis.
Boldface data indicate P values that were <0.05.
DISCUSSION
In our study, we showed the efficacy and reliability of suicide PCR for the molecular detection of rickettsiae in skin biopsy specimens. The method was sensitive and versatile, enabling us to identify seven different Rickettsia spp. It was also specific, as all positive results were confirmed by a second ompA-based nested PCR. Except for one assay, which was not included in our analysis of the data, we had no contamination problems.
We developed the suicide PCR as a versatile but specific test for the rapid, direct, and sensitive detection of rickettsiae in patients with severe rickettsiosis or cases of bioterrorism. In a previous study, Leitner et al. (11) described a nested PCR assay that amplified a fragment of the 17-kDa protein-encoding gene. Although this assay was sensitive, it had several drawbacks, including the need for the use of primers exhibiting mismatches with the sequences from R. typhi and R. prowazekii and a single target sequence which was identical for R. rickettsii, R. conorii, R. sibirica, R. montanensis, R. rhipicephali, and R. parkeri. Also, in-house nested PCRs are readily subject to laboratory contamination, especially when they are performed more than once with the same target. We have experienced such contamination problems in the past (unpublished data). To enable us to amplify both typhus and spotted fever group rickettsiae, we compared the complete genome sequences of R. prowazekii and R. conorii and selected primers specific for regions of genes that were conserved in both organisms. In addition, to limit the risk of vertical contamination by amplicons from previous assays, in each year of the study we targeted a new gene that had never been amplified before in our laboratory. This was facilitated by the wide choice of potential target genes, with 834 genes being conserved in both R. prowazekii and R. conorii. To limit lateral contamination due to carryover, all the reagents for the two successive PCRs were incorporated into the reaction tube prior to the first amplification step and no positive controls were used. Furthermore, we routinely included in our assays large numbers of negative controls, processed identically to the test samples. We evaluated our suicide PCR in blinded tests using skin biopsy specimens, which are the specimens of choice for the detection of rickettsioses by isolation or genomic procedures. In a pilot study performed on 15 biopsy specimens from patients who definitely had a rickettsiosis, 13 were positive by suicide PCR. There was no contamination of negative controls, and in all cases the Rickettsia sp. identified by the suicide PCR was the same as that established in previous tests. We then applied the suicide PCR to all skin biopsy specimens referred to our laboratory. The 212 skin biopsy specimens from patients with a suspected rickettsiosis evaluated in this study are the largest series tested to date. We performed only one suicide PCR assay each year, so skin biopsy specimens were frozen at −80°C for a maximum of 1 year prior to DNA extraction and PCR amplification. In patients who definitely had a rickettsiosis, the sensitivity of the suicide PCR was 68%, significantly higher than those of culture (P < 10−2) and regular PCR (P < 10−2). This suggests that the suicide PCR should be reserved for use on skin biopsy specimens from patients suspected of having a rickettsiosis but with negative regular PCR results. The specificity of the suicide PCR was 100%, but this was obtained only by strict interpretation of the results for the negative controls. We considered a suicide PCR result valid only if none of the negative controls produced an amplification product. We selected primer sequences from conserved regions of the genomes of the R. prowazekii and R. conorii flanking variable sequences, but, unfortunately, we could not predict the variability of the target gene fragments within the spotted fever and typhus groups. The infecting rickettsiae that we identified by the suicide PCRs were, in all cases, identical to those identified by previous amplification and sequencing by an ompA-based PCR assay. We identified two factors that independently influenced the suicide PCR results: (i) the administration of antibiotics highly effective against rickettsiae prior to biopsy sampling significantly reduced the rate of detection of rickettsiae (P < 10−2) (Table 4), as has previously been observed for detection by culture (8); and (ii) significantly more positive suicide PCR results were obtained on biopsy specimens from eschars than other skin biopsy specimens (P = 0.03, Table 4), as has previously been demonstrated for diagnosis by culture and regular PCR (21). The latter finding is probably linked to the initial multiplication of rickettsiae at the site of inoculation. The sensitivity of the suicide PCR increased to 78% when only results for eschar biopsy specimens taken prior to antibiotic therapy were considered. We found that delays during shipment to our laboratory and the type of rickettsiosis of the patient had no independent influence on the suicide PCR results.
For 104 patients classified as possibly having a rickettsiosis, suicide PCR provided direct evidence of infection with “R. conorii conorii” in 8 patients, R. africae in 3 patients, R. slovaca in 5 patients, and R. aeschlimannii in 1 patient. In all patients except the last patient, the etiological agent found matched the clinical and laboratory data. The final patient presented with clinical symptoms consistent with MSF in central France, where Rhipicephalus ticks, vectors of MSF, are not endemic. The suicide PCR identified R. aeschlimannii as the etiological agent of the infection. This rickettsia has been described in Hyalomma sp. ticks collected in Morocco (1), Zimbabwe (1), Mali and Niger (14), Spain (3), Croatia (16), Kazakhstan (26), and Corsica (K. Matsumoto, personal communication). The first two documented infections with R. aeschlimannii were described in a patient who had just returned from Morocco (20) and whose skin biopsy specimen was included in our study and in a hunter in South Africa (15). Prior to 1991, only “R. conorii conorii” was identified as an agent of tick-borne rickettsiosis in France, and therefore, the diagnostic scoring system used in our study included only microimmunofluorescence assay and culture as laboratory tests. Since 1991, however, another three pathogenic rickettsiae have been identified in France: R. slovaca (2), “R. sibirica mongolotimonae” (19), and R. helvetica (4). We now add R. aeschlimannii, which has recently been found in French ticks (K. Matsumoto, personal communication), to the list of Rickettsia spp. that may infect people in France. Our data suggest that the diagnosis of suspected rickettsioses, especially atypical cases, will be improved with the use of the suicide PCR. Our results also indicate that diagnostic scoring systems should be modified to include this new technique as a diagnostic criterion.
In conclusion, we have demonstrated that suicide PCR is a valuable tool for the detection of rickettsiae in skin biopsy specimens. It is specific and is 1.5 times more sensitive than regular PCR for the diagnosis of rickettsioses. Comparison of the complete sequences of R. conorii and R. prowazekii to determine appropriate primers for suicide PCRs provided us with a large choice of potential targets and improved the versatility of the suicide PCR for the identification of new pathogenic rickettsiae. Our findings show that eschar biopsy specimens for suicide PCR should be taken before antibiotic therapy is given. Sera are the most easily obtained clinical samples, and the value of suicide PCR on such specimens warrants further study (21-23). Finally, with the increasing number of complete genomes that are being sequenced, we suggest that suicide PCR will become more widely used for the detection of other bacteria.
Acknowledgments
We thank Gilbert Greub for assistance with statistical analysis and Patrick J. Kelly for reviewing the manuscript.
The authors do not have any commercial or other association that might pose a conflict of interest.
The research described in this manuscript has not benefited from any external financial support.
REFERENCES
- 1.Beati, L., M. Meskini, B. Thiers, and D. Raoult. 1997. Rickettsia aeschlimannii sp. nov., a new spotted fever group rickettsia associated with Hyalomma marginatum ticks. Int. J. Syst. Bacteriol. 47:548-554. [DOI] [PubMed] [Google Scholar]
- 2.Cazorla, C., M. Enea, F. Lucht, and D. Raoult. 2003. First isolation of Rickettsia slovaca from a patient, France. Emerg. Infect. Dis. 9:135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fernandez-Soto, P., A. Encinas-Grandes, and R. Perez-Sanchez. 2003. Rickettsia aeschlimannii in Spain: molecular evidence in Hyalomma marginatum and five other tick species that feed on humans. Emerg. Infect. Dis. 9:889-890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fournier, P. E., F. Gunnenberger, B. Jaulhac, G. Gastinger, and D. Raoult. 2000. Evidence of Rickettsia helvetica infection in humans, eastern France. Emerg. Infect. Dis. 6:389-392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Fournier, P. E., V. Roux, and D. Raoult. 1998. Phylogenetic analysis of spotted fever group rickettsiae by study of the outer surface protein rOmpA. Int. J. Syst. Bacteriol. 48:839-849. [DOI] [PubMed] [Google Scholar]
- 6.Fournier, P. E., B. Xeridat, and D. Raoult. 2003. Isolation of a rickettsia from a patient in Chad which is related to Astrakhan fever rickettsia. Ann. N. Y. Acad. Sci. 990:152-157. [DOI] [PubMed] [Google Scholar]
- 7.Jensenius, M., P. E. Fournier, S. Vene, T. Hoel, G. Hasle, A. Z. Henriksen, K. B. Hellum, D. Raoult, and B. Myrvang. 2003. African tick-bite fever in travelers to rural sub-equatorial Africa. Clin. Infect. Dis. 36:1411-1417. [DOI] [PubMed] [Google Scholar]
- 8.La Scola, B., and D. Raoult. 1996. Diagnosis of Mediterranean spotted fever by cultivation of Rickettsia conorii from blood and skin samples using the centrifugation-shell vial technique and by detection of R. conorii in circulating endothelial cells: a 6 year follow-up. J. Clin. Microbiol. 34:2722-2727. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.La Scola, B., and D. Raoult. 1997. Laboratory diagnosis of rickettsioses: current approaches to the diagnosis of old and new rickettsial diseases. J. Clin. Microbiol. 35:2715-2727. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.La Scola, B., L. Rydkina, J. B. Ndihokubwayo, S. Vene, and D. Raoult. 2000. Serological differentiation of murine typhus and epidemic typhus using cross-adsorption and Western blotting. Clin. Diagn. Lab. Immunol. 7:612-616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Leitner, M., S. Yitzhaki, S. Rzotkiewicz, and A. Keysary. 2002. Polymerase chain reaction-based diagnosis of Mediterranean spotted fever in serum and tissue samples. Am. J. Trop. Med. Hyg. 67:166-169. [DOI] [PubMed] [Google Scholar]
- 12.Marrero, M., and D. Raoult. 1989. Centrifugation-shell vial technique for rapid detection of Mediterranean spotted fever rickettsia in blood culture. Am. J. Trop. Med. Hyg. 40:197-199. [DOI] [PubMed] [Google Scholar]
- 13.Paddock, C. D., J. W. Sumner, J. A. Comer, S. R. Zaki, C. S. Goldsmith, J. Goddard, S. L. F. McLellan, C. L. Taminga, and C. A. Ohl. 2004. Rickettsia parkeri: a newly recognized cause of spotted fever rickettsiosis in the United States. Clin. Infect. Dis. 38:805-811. [DOI] [PubMed] [Google Scholar]
- 14.Parola, P., H. Inokuma, J. L. Camicas, P. Brouqui, and D. Raoult. 2001. Detection and identification of spotted fever group rickettsiae and ehrlichiae in African ticks. Emerg. Infect. Dis. 7:1014-1017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pretorius, A. M., and R. J. Birtles. 2002. Rickettsia aeschlimannii: a new pathogenetic spotted fever group Rickettsia, South Africa. Emerg. Infect. Dis. 8:874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Punda-Polic, V., M. Petrovec, T. Trilar, D. Duh, N. Bradaric, Z. Klismanic, and T. Avsic-Zupanc. 2002. Detection and identification of spotted fever group rickettsiae in ticks collected in southern Croatia. Exp. Appl. Acarol. 28:169-176. [DOI] [PubMed] [Google Scholar]
- 17.Raoult, D. 2004. A new rickettsial disease in the United States. Clin. Infect. Dis. 38:812-813. [DOI] [PubMed] [Google Scholar]
- 18.Raoult, D., G. Aboudharam, E. Crubezy, G. Larrouy, B. Ludes, and M. Drancourt. 2000. Molecular identification by “suicide PCR” of Yersinia pestis as the agent of medieval black death. Proc. Natl. Acad. Sci. USA 97:12800-12803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Raoult, D., P. Brouqui, and V. Roux. 1996. A new spotted-fever-group rickettsiosis. Lancet 348:412. [DOI] [PubMed] [Google Scholar]
- 20.Raoult, D., P. E. Fournier, P. Abboud, and F. Caron. 2002. First documented human Rickettsia aeschlimannii infection. Emerg. Infect. Dis. 8:748-749. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Raoult, D., P. E. Fournier, F. Fenollar, M. Jensenius, T. Prioe, J. J. de Pina, G. Caruso, N. Jones, H. Laferl, J. E. Rosenblatt, and T. J. Marrie. 2001. Rickettsia africae, a tick-borne pathogen in travelers to sub-Saharan Africa. N. Engl. J. Med. 344:1504-1510. [DOI] [PubMed] [Google Scholar]
- 22.Raoult, D., B. La Scola, M. Enea, P. E. Fournier, V. Roux, F. Fenollar, M. A. M. Galvao, and X. De Lamballerie. 2001. A flea-associated Rickettsia pathogenic for humans. Emerg. Infect. Dis. 7:73-81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Raoult, D., A. Lakos, F. Fenollar, J. Beytout, P. Brouqui, and P. E. Fournier. 2002. Spotless rickettsiosis caused by Rickettsia slovaca and associated with Dermatocentor ticks. Clin. Infect. Dis. 34:1331-1336. [DOI] [PubMed] [Google Scholar]
- 24.Raoult, D., H. Tissot-Dupont, P. Caraco, P. Brouqui, M. Drancourt, and C. Charrel. 1992. Mediterranean spotted fever in Marseille: descriptive epidemiology and the influence of climatic factors. Eur. J. Epidemiol. 8:192-197. [DOI] [PubMed] [Google Scholar]
- 25.Roux, V., E. Rydkina, M. Eremeeva, and D. Raoult. 1997. Citrate synthase gene comparison, a new tool for phylogenetic analysis, and its application for the rickettsiae. Int. J. Syst. Bacteriol. 47:252-261. [DOI] [PubMed] [Google Scholar]
- 26.Shpynov, S., P. E. Fournier, N. Rudakov, M. Tankibaev, I. Tarasevich, and D. Raoult. 2004. Detection of a rickettsia closely related to Rickettsia aeschlimannii, Rickettsia heilongjiangensis, Rickettsia sp. RPA4, and Ehrlichia muris in ticks collected in Russia and Kazakhstan. J. Clin. Microbiol. 42:2221-2223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Teysseire, N., and D. Raoult. 1992. Comparison of Western immunoblotting and microimmunofluoresence for diagnosis of Mediterranean spotted fever. J. Clin. Microbiol. 30:455-460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680. [DOI] [PMC free article] [PubMed] [Google Scholar]