Vol. 31, No. 10JOURNAL OF CLINICAL MICROBIOLOGY, OCt. 1993, p. 2745-27500095-1137/93/102745-06$02.00/0Copyright © 1993, American Society for Microbiology

Comparison of Polymerase Chain Reaction, Culture, andWestern Immunoblot Serology for Diagnosis of

Bordetella pertussis InfectionEMMANUEL GRIMPREL,l 2t* PIERRE BEGUE,1 ISAM ANJAK,1"2 FOTINI BETSOU,2

AND NICOLE GUISO2Consultation de Pe6diatrie, Pathologie Infectieuse et Tropicale, Urgences Pediatriques,

H6pital Armand-Trousseau, 75571 Paris Cede 12, 1 and Unite de BacteriologieMoleculaire et Medicale, Institut Pasteur, 75724 Paris cedex 15,2 France

Received 29 March 1993/Returned for modification 18 May 1993/Accepted 2 July 1993

Polymerase chain reaction (PCR) amplification of the pertussis toxin promoter region was used to detectBordeteUa pertussis infection in nasopharyngeal aspirates collected from 24 infants and children infected withpertussis and 13 adult contacts during an epidemiological study. The sensitivity of this PCR assay wasapproximately one bacterium, and the assay was specific for B. pertussis in tests with other BordeteUa speciesand other respiratory pathogens. The pertussis case definition required a cough with a duration of more than21 days for infants and children and laboratory confirmation by serology as the primary detection method forinfants, children, and adults. The sensitivity ofPCR and culture on Bordet-Gengou agar medium was assessedwith regard to the case definitions. In the group of infants and children (index cases), the sensitivities of theculture and the PCR were 54.1% (13 of 24) and 95.8% (23 of 24), respectively. In the adult group (householdcontacts), the sensitivities of the two methods were 15.4% (2 of 13) and 61.5% (8 of 13), respectively. PCRcombined with pertussis-specific serology appears to be a useful tool for diagnosis of pertussis especially inepidemiological studies.

Pertussis (whooping cough) is an infectious respiratorydisease caused by the bacterial species Bordetella pertussis.The widespread use of a classic whole-cell pertussis vaccineled to a spectacular decrease in pertussis morbidity andmortality in the United States (8) and in some Europeancountries, such as France (16). However, several reportshave shown an unexpected resurgence of pertussis in thesecountries (4). One hypothesis commonly accepted is thatpostvaccination immunity wanes with time (9), in part be-cause of the absence of a late vaccination booster or anatural booster. Thus, young children and adults, despiteprevious immunization, may become susceptible to B. per-tussis infection (21-23) and may disseminate the infectioninside their family. As a consequence, unprotected younginfants who are not yet immunized because of their age maybecome infected. These infants are particularly at risk forsevere pertussis and complications (8).

This new epidemiological situation underscores the needfor sensitive and rapid methods for diagnosis of pertussis.Diagnosis of pertussis is still difficult because the classicfeature of whooping cough (i.e., the whoops) is rarely foundin adults and young infants (4). B. pertussis infection gener-ally presents as a prolonged cough in this population, andparoxysms are often missing, especially in adults and school-age children. Furthermore, laboratory confirmation is alsodifficult. Serological testing by enzyme-linked immunosor-bent assay (ELISA) (26) or Western blot (immunoblot) (2)with specific B. pertussis antigens, i.e., pertussis toxin (PT),is very sensitive and specific, but a definitive confirmation

* Corresponding author.t Present address: Consultation de Pediatrie, Pathologie Infec-

tieuse et Tropicale, Urgences Pediatriques, H6pital Armand-Trous-seau, 26 avenue du Dr. Arnold Netter, 75571 Paris Cedex 12,France.

generally requires the comparison of two serum samplescollected at 4-week intervals, and such a delay is incompat-ible with an early diagnosis. Bacterial identification byculture from nasopharyngeal secretions is still considered tobe a reference method. However, culture is not sensitive inmany routine laboratories, and sensitivity decreases rapidlyafter the onset of paroxysms (11, 28). The polymerase chainreaction (PCR) has recently been developed for direct iden-tification of fastidious culture-growing microorganisms andhas already been used to improve laboratory diagnosis ofmany infectious diseases caused by such different pathogensas Mycobacterium tuberculosis (13), Towoplasma gondii (6),Treponema pallidum subsp. pallidum (12), Mycoplasmapneumoniae (5), Legionella spp. (19), and Leptospira spp.(20). Thus, PCR, which is rapid and specific, could representa useful complementary tool in pertussis diagnostic strategy.Two different PCR assays were recently developed for directidentification of B. pertussis (10, 15). Glare et al. (10) used arepetitive sequence from B. pertussis and concluded thattheir assay was rapid, sensitive, and specific and could beused for diagnosis of pertussis. However, the choice of sucha repetitive sequence can be questioned. (i) The number ofthe repetitive sequences contained in a single bacterium mayvary within strains (10). (ii) Some genes coding for B.pertussis virulence factors also contain repetitive sequencesthat can vary between Bordetella species (7). (iii) Therepetitive sequence used by Glare et al. (10) is also present ina single copy in some B. bronchiseptica isolates and there-fore cannot be considered totally specific for B. pertussis.For these reasons, we chose another sequence, the PTpromoter, previously used by Houard et al. (15). Thissequence is present in all B. pertussis isolates so far tested(3, 17). Furthermore, the three Bordetella species B. pertus-sis, B. parapertussis, and B. bronchiseptica possess the PToperon (pt), but B. parapertussis and B. bronchiseptica do

2745

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from

2746 GRIMPREL ET AL.

not express the toxin because of multiple mutations locatedwithin the pt promoter (17). Thus, this promoter appears tobe a good candidate for a B. pertussis-specific target forPCR. The amplification of a 191-bp sequence of the ptpromoter has been described by Houard et al. (15), who usedtwo oligonucleotide primers, pPT-1 and pPT-2, located on

sequences containing several intraspecies mutations (17).Some clinical specimens collected from infants suspected ofbeing infected with B. pertussis have been successfullytested with this method (15), but no correlation was madewith clinical, bacteriological, and/or serological data.The aim of this study was to compare the sensitivity and

specificity of amplification of the 191-bp portion of the ptpromoter region ofB. pertussis with those of nasopharyngealculture on Bordet-Gengou agar medium in clinical samplesduring confirmed cases of pertussis.

MATERIALS AND METHODS

Patients. Informed consent was obtained from the patientsor their parents or guardians. The study design followed theHuman Experimentation Guidelines of the Comite Consul-tatif pour la Protection des Personnes dans la RechercheBiomedicale according to present French legislation. Twogroups of patients were studied: a group of infants andchildren with pertussis and a group of adults who hadcontact with these children. Two pertussis case definitionswere used in our study. For the children and infants, thestrict pertussis case definition used by P. Olin (25) during a

Swedish vaccine trial in 1990 and later recommended by theWorld Health Organization for vaccine studies was chosenfor its specificity. These pediatric cases of infection requireda cough lasting more than 21 days (total duration) andconfirmation by either culture and/or serum detection ofanti-PT antibodies or culture and/or serologically confirmedhousehold exposure. For the adult contacts, no clinicalsymptom was required, but family exposure to one of thepediatric cases and confirmation by culture and/or serumdetection of anti-PT antibodies were required. Thirty-sevenpatients (24 infants and children and 13 adults) satisfied thecase definitions chosen for the study (Tables 1 and 2). All buttwo of them had detectable anti-PT antibodies in serum;thus, serology can be considered to be the primary labora-tory method for confirmation of the diagnosis in our study.

All 24 infants and children coughed for a duration of longerthan 21 days, and all had paroxysms during the course of thedisease. The mean duration of coughing at the first visit was16.4 days (range 8 to 30 days) (Table 1). The ages rangedfrom 3 Weeks to 12 years (19 of 24 patients were aged lessthan 6 months), and the sex distribution was 13 boys and 11girls. Only five of them had previously been immunized witha tetravalent combined diphtheria-tetanus-poliomyelitis-per-tussis whole-cell vaccine; one had received four injections(no. 3), one had received three injections (no. 14), and threehad received only one injection (no. 4, 5, and 21).

All infants and children underwent a physical examina-tion, blood sampling, and nasopharyngeal aspiration forbacterial culture and PCR during the first visit. Two patients(no. 3 and 24) had only one early serum sample collected inwhich no anti-PT antibody was detected, and no other serumsample from them was available for comparison. Child 3 hada positive nasopharyngeal culture and infant 24 had beenexposed to a serologically confirmed primary case of infec-tion (Table 2, no. 29) inside the family and thus satisfied thecase definition criteria. The 22 other children had detectableanti-PT antibodies in serum. Three of them (no. 10, 12, and

TABLE 1. Clinical data and results of nasopharyngeal culture,PCR, and detection of anti-PT antibodies in serum samples

from infants and children

Patient Agea Duration of cough NPCb PCR Anti-PTno. at first visit (days) antibodies

1 15wk 10 + + +2 18wk 19 + + +3 9yr 8 + + -4 13wk 16 + + +5 18wk 17 + + +6 8wk 15 + + +7 lOwk 21 + + +8 3wk 10 + + +9 5wk 10 + + +10 3wk 20 + + +11 9wk 15 + + +12 24wk 15 + + +13 36wk 13 + + +14 12yr 21 - + +15 3.5 yr 30 - + +16 ilyr 18 - + +17 7wk 27 - + +18 3wk 20 - + +19 8wk 10 - + +20 12wk 30 - + +21 lOwk 10 - + +22 Swk 14 - + +23 l5wk 10 - + +24 4wk 15 - - -

a Age at the beginning of the symptoms.b NPC, nasopharyngeal culture.

14) each had only one serum sample collected in whichanti-PT antibodies were already detected, and no otherserum sample was available for comparison. The 19 otherchildren each had two serum samples collected at 4- to6-week intervals for pertussis serology. The first serumsample was collected early in the disease, during the firstvisit, and the second serum sample was collected at follow-up, 4 to 6 weeks later. When available, for young infantsaged less than 6 months, a prepartum serum sample, col-lected from the mother during pregnancy (for rubella or

TABLE 2. Clinical data and results of nasopharyngeal culture,PCR, and detection of anti-PT antibodies in serum samples

from adult contacts

Patient Index Age IMM Cough Duration of NPCb PCR Anti-PTno. case no. (yr) cough (days) antibodies

25 1 30 U + 30 + + +26 2 23 0 + 10 + + +27 19 34 0 + 12 - + +28 4 45 0 + 30 - + +29 24 19 U + 15 - + +30 7 24 4c + 60 - + +31 6 35 P + 30 - - +32 23 35 0 + 30 - - +33 5 30 U + 25 - - +34 7 22 U + 30 - - +35 4 23 5 - - + +36 4 22 4 - - + +37 7 29 5 - - - +

a Abbreviations: IMM, immune status regarding pertussis; U, unknown; P,past history of pertussis infection.

b NPC, nasopharyngeal culture.c Number of pertussis immunizations.

J. CLIN. MICROBIOL.

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from

METHODS FOR DIAGNOSING B. PERTUSSIS INFECTION COMPARED 2747

toxoplasmosis serology), was tested retrospectively for per-tussis serology and compared with the first serum sample.Eighteen of them (no. 1, 2, 4 to 9, 11, 13, 15 to 20, 22, and 23)had two serum samples collected in which seroconversionwith anti-PT antibodies could be demonstrated (absence ofanti-PT antibodies in the first serum sample and positivedetection in the second serum sample or significant increasein anti-PT antibodies in both serum samples). For theremaining patient (no. 21), the seroconversion could bedemonstrated by comparing the first serum sample, whichalready contained anti-PT antibodies, with the prepartumserum sample collected from the mother.Among the household, 13 adult contacts (10 mothers, 2

fathers, and 1 grandmother) were tested by serology and hada nasopharyngeal aspiration cultured and tested by PCR.Their relationships to the infants and children with indexcases of infection are shown in Table 2. Ten of themcoughed, and 7 of 10 (no. 25, 28, and 30 to 34) startedcoughing before the child (Table 2). All 13 adults hadpositive detection of anti-PT antibodies in their serum sam-ples. Seroconversion was observed in 10 cases of infection(no. 25 to 27 and 31 to 37), either by comparing the twoserum samples collected during the disease or by comparingthe first serum sample with a prepartum serum sample (whenavailable). The three remaining adults (no. 28 to 30) hadanti-PT antibodies strongly detectable in the first serumsample, and no preceding serum sample was available forcomparison.

Sixteen infants and 4 adults with coughs lasting for lessthan 10 days and with culture-negative nasopharyngeal as-pirates and absence of anti-PT antibodies in two serumsamples collected at 4- to 6-week intervals were considerednot infected by B. pertussis and were tested by PCR asnegative controls in the study.

Bacterial strains. Tohama B. pertussis strain (Collection del'Institut Pasteur [CIP] 8132) was used as a positive controlfor the development and optimization of the PCR assay andas a PCR target for elaboration of the DNA probe used in thestudy. Two B. pertussis mutants, BPRA (1) and BP-TOX 6(27), six B. pertussis clinical isolates, 2 B. parapertussisisolates (CIP 63.2 and CIP 64.11), 10 B. bronchisepticaisolates (CIP 9.73 and 9 clinical isolates), and 1 B. aviumclinical isolate were tested by PCR. All Bordetella strainswere cultured for 3 days at 37°C on fresh Bordet-Gengouagar supplemented with 15% defibrinated sheep blood (BGS)without antibiotic and were resuspended in sterile distilledwater. The number of CFU was determined by opticaldensity at 650 nm (optical density = 1, corresponding to 3 x109 CFUJ/ml).

Biological samples from patients. The nasopharyngeal as-pirates were collected in a dry sterile tube during the firstexamination and transported within 4 h at room temperatureto the laboratory for culture. Aliquots of the samples weresimultaneously plated on fresh BGS and cultured at 36°C for72 h and then were inoculated in Stainer-Scholte medium for20 h at 36°C before reinoculation on BGS. The remainingsamples were stored at -80°C until PCR analysis was carriedout. Identifications of all isolates were biochemically verifiedby testing catalase, oxidase, urea, nitrate reduction, andpigmentation on tyrosine agar after growth on blood agar.

Preparation of the samples for PCR. Frozen nasopharyn-geal aspirates were thawed at room temperature, and 50 ,ul ofeach sample was digested for 1 h at 650C with Proteinase K(0.2 ng/ml) in 12 mM Tris hydrochloride (pH 7.6) in a finalvolume of 100 p,l, according to the technique described byGlare et al. (10). Proteinase K was then heat inactivated at

100°C for 20 min. Part of the extract (30 ,ul) was used for PCRamplification. Each experiment contained 8 to 10 samples,including two reagent negative control tubes (one containingTris hydrochloride and proteinase K and one containingsterile water added through the mineral oil) which were usedto assess spurious contamination during the procedure.PCR assay. The oligonucleotide primers, PTp-1 and

PTp-2, derived from nucleotides 307 to 332 and 469 to 497 ofthe respective sense and antisense strands of the PT operon(15), were synthesized on an Applied Biosystems DNAsynthesizer (model 391). DNA amplifications were per-formed in a final volume of 50 p1. The reaction mixtureconsisted of 50 mM KCl, 10 mM Tris hydrochloride (pH8.4), 6 mM MgCl2, 1 mg of gelatin per ml, 200 puM eachdeoxyribonucleotide, 1 p.M each primer, and 0.5 U of Taqpolymerase (Beckman, Fullerton, Calif.). Reactions wereperformed with a PHC-3 thermal cycler (Techne, Cam-bridge, United Kingdom) for 40 cycles with the followingparameters: 1 min of denaturation at 95°C, 1 min 30 s ofannealing at 66°C, and 1 min 30 s of extension at 72°C. Thepositive control tube contained 10 pul (equivalent to 103 cells)of a boiled suspension of purified B. pertussis DNA (Tohamastrain).

Detection of the PCR products. The amplification productswere run in parallel with molecular weight markers in a 1%agarose gel stained with ethidium bromide. For greatersensitivity and specificity, the analysis of the PCR productswas routinely performed by hybridization with a probespecific for B. pertussis pt promoter. The same oligonucle-otide primers (pPT-1 and pPT-2) were used to synthesize thisprobe in a PCR experiment in which dTTP was replaced byfluorescein-n-dUTP (Amersham, Little Chalfont, UnitedKingdom). The PCR products (10 p.l) were added to 40 ,ul ofTris-EDTA buffer and heat denaturated for 10 min. Thesamples were then applied to a nylon membrane (Hybond-N;Amersham) by using a Minifold I apparatus (Schleicher andSchuell, Dassel, Germany). The filters were dried, baked for2 h at 80°C, and hybridized with the B. pertussis probeaccording to the manufacturer's instructions. The detectionof the fluorescein-labelled probe was performed by use ofhorseradish peroxidase-labelled anti-fluorescein antibodies(Amersham) with the Amersham enhanced-chemilumines-cence system.

Serological assays. The detection of anti-PT antibodies inserum was performed with purified PT antigen and Westernblot as previously described (lOa). The immunochemicaldetection was performed with peroxidase-labelled anti-hu-man immunoglobulins with the Amersham enhanced-chemiluminescence system, which has several advantages,including high sensitivity (the first serum dilution that can beused is 2 x 10-3 and not 1 x 10-1 or 1 x 10-2) and rapidity(maximum of 30 min until revelation of the reaction). Withthis detection system, blackening of the X-ray film is pro-portional to the sample light emission. In order to comparethe results obtained with paired serum samples, the treatedmembranes were exposed to X-ray films for three differenttimes: 6 s, 1 min, and 10 min. Detection of the immunecomplex was classified as ++ + after 6 s, + + after 1 min,and + after 10 min, no detection after 10 min was classifiedas -. An increase in the signal detection from + to + + +was considered significant.

RESULTS

Sensitivity and specificity of the PCR amplification. PCRwas used to detect B. pertussis by using the pt promoter

VOL. 31, 1993

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from

2748 GRIMPREL ET AL.

region as a target sequence. We first analyzed six B. pertus-sis clinical isolates by PCR. Each strain produced a positivesignal (a 191-bp band in agarose gel electrophoresis). The B.pertussis BPRA mutant carrying a deletion of the -whole ptoperon, including the pt promoter (1), did not produce anysignal. The B. pertussis BP-TOX 6 mutant, also carrying adeletion of the pt operon but not of the pt promoter region(27), produced a positive signal. The sensitivity of the PCRreaction was determined by using a boiled suspension of B.pertussis Tohama strain, serially diluted to obtain decreasingconcentrations of bacteria (from 106 organisms to 100). Bythis method, we were able to detect less than 102 organismson the agarose gel and 101 organisms after dot blot hybrid-ization with a fluorescein-labelled probe, which is in agree-ment with Houard et al., who used a 32P-labelled probe (15).To determine the specificity of the oligonucleotide primersfor B. pertussis, 2 B. parapertussis strains, 10 B. bronchi-septica strains, and 1 B. avium strain were tested by usingthese primers and the same hybridization conditions forPCR. We confirmed the data reported by Houard et al. (15):all B. pertussis strains possessing the pt promoter werepositive by PCR and all B. parapertussis, B. bronchiseptica,and B. avium strains were negative. Furthermore, we testedthe specificity of the primers with purified DNA obtainedfrom different bacterial and fungal strains liable to contami-nate nasopharyngeal aspirates upon collection (Corynebac-terium spp., Staphylococcus aureus, Mycoplasma pneumo-niae, Legionella pneumophila, Aspergillus spp., andCandida albicans). None of these DNA samples demon-strated positive amplification of the pt promoter region byPCR.

Bacterial culture and PCR results. Thirty-seven patients(24 infants and children with cases of pertussis [Table 1] and13 adult contacts [Table 2]) belonging to 23 different families(patients 3 and 14 were sisters) each had a nasopharyngealaspirate cultured and tested by PCR.Among the 24 pertussis-infected infants and children,

54.1% (13 of 24) had a nasopharyngeal-positive culture, and95.8% (23 of 24) were positive by PCR (Table 1). All of the13 samples positive by culture were also positive by PCR(Table 1). Among the 24 infants and children, 10 of 17 hadreceived oral antibiotics prior to the nasopharyngeal aspiratecollection-5 of 9 were culture positive (4 macrolide and 1cephalosporin), and 5 of 8 were culture negative (3 macrolideand 2 amoxicillin).Among the 13 adults with pertussis identified within the

household, only 15.4% (2 of 13) were positive by culture and61.5% (8 of 13) were positive by PCR (Table 2). Here again,the two samples positive by culture were positive by PCR.However, it has to be noted that two adults (no. 35 and 36)were positive by PCR and had detectable anti-PT antibodiesin serum but did not exhibit any clinical symptoms (Table 2).As shown in Table 2, four symptomatic and serologicallyconfirmed cases of pertussis (no. 31 to 34) were detectedneither by culture nor by PCR. No data concerning antibiotictreatments were available for these adults.

All of the 20 negative controls tested in this study werenegative by culture and by PCR, and no antibodies weredetected in their serum samples.

DISCUSSION

We voluntarily limited our study to B. pertussis infections.Thus, we chose a single antigen, PT, for serology and the PTpromoter sequence as the PCR target because of their highspecificity to B. pertussis. We also chose a strict case

definition of pertussis for this first epidemiological PCRstudy in order to validate the assay. All of the childrencoughed for more than 21 days, all but 2 of our patients haddetectable anti-PT antibodies in serum, and a significantincrease of anti-PT antibodies was found for most of ourpatients (29 of 37). This may have increased the sensitivity ofour PCR assay. However, the choice of a single copysequence and the absence of internal control for amplifica-tion may, in turn, have limited this sensitivity.The overall sensitivity (adults, infants, and children com-

bined) of the nasopharyngeal culture (40.5%) was compara-ble to those in many pertussis studies (11, 26). The sensitiv-ity of the nasopharyngeal culture was higher in the group ofinfants and children (54.1%) than in the group of adults(15.4%). Lower sensitivity of culture in adults comparedwith children has been reported (28). This result can beattributed to the particularly long duration of the cough inour adult series. No relationship between antibiotic therapyand culture result could be demonstrated, but few data wereavailable. In comparison, the PCR method had greatersensitivity both in infants and children and in adults (95.8and 58%, respectively). In the adult group, four symptomaticand serologically confirmed cases of pertussis were notdetected by culture and PCR. Our choice of a single copysequence as the PCR target may be responsible for reducedsensitivity of the PCR for samples containing few bacteria.This result demonstrates, however, that although it is verysensitive, PCR cannot detect all cases of pertussis. Pertus-sis-specific serology presently remains the reference tech-nique for pertussis diagnosis in epidemiological studies.Ten of 13 adults had a prolonged cough and detectable

anti-PT antibodies in their serum samples, and were consid-ered as having cases of pertussis according to the definitionused in this study. Two of them had a positive nasopharyn-geal culture, four others had detectable anti-PT antibodies inthe second serum sample but not in the first serum sample,and two others had a significant increase in the level of theanti-PT antibodies when the first and second serum sampleswere compared. These results can be considered to beevidence of recent pertussis infection. The remaining twoadults, however, each had only one serum sample availablein which anti-PT antibodies were detected with a strongsignal. The hypothesis that the presence of these antibodiescould be the consequence of past immunization or infectionis unlikely. Indeed, preliminary data collected from a sero-logical study recently conducted in the French populationshowed that less than 20% of the individuals aged more than15 years possessed detectable anti-PT antibodies despiteprevious pertussis immunization (unpublished data) or infec-tion. This observation corroborates the fact that anti-PTantibody levels decrease rapidly in the absence of boosterimmunization.Three adults had a positive detection of anti-PT antibodies

in serum and did not cough despite an intrafamilial contactwith a confirmed case of pertussis. One of them had detect-able anti-PT antibodies in the second serum sample but notin the first serum sample and was culture negative and PCRnegative. The two others had serological evidence of pertus-sis, and the PCR result was positive but the culture wasnegative. These two cases of infection are difficult to inter-pret, considering that these adults did not cough. Twohypotheses can be suggested to explain this result. First,these adults were simply colonized, and a booster effect wasobserved. Such an observation has been previously reportedin a study in which adult contacts with cases of pertussis hada positive nasopharyngeal culture and serologic evidence of

J. CLIN. MICROBIOL.

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from

METHODS FOR DIAGNOSING B. PERTUSSIS INFECTION COMPARED 2749

pertussis infection but no clinical symptoms (18). Second,contamination might have occurred during specimen collec-tion or processing. Such a hypothesis can never definitivelybe ruled out in a PCR study. However, the absence offalse-positive PCR results in the control group gives someguarantees concerning the quality of our results.

Since we started this study, two other studies using PCRin clinical samples from pertussis-infected patients havebeen reported (14, 24). Both studies used the repetitivesequence described by Glare et al. (10) and suggested thatthis sequence was specific for B. pertussis. However, only afew B. bronchiseptica strains were tested for specificity. Theinterpretation of the results reported by Olcen et al. (24) ishindered by the absence of clear definition of the cases ofpertussis and by the lack of correlation with serological data.In the study by He et al. (14), the PCR results werecorrelated with clinical data (duration of cough), culture, andserology by using an enzyme immunoassay with three puri-fied antigens (PT, filamentous hemagglutinin, and pertactin).A significant increase in only two of the three correspondingantibodies was considered to be evidence of a recent case ofpertussis. This suggests that some patients with no anti-PTantibodies were considered to be infected with B. pertussis.These patients could as well have been infected by B.parapertussis and B. bronchiseptica, two Bordetella specieswhich do not synthesize PT. This may in part explain thelower rates of positivity by culture and PCR observed intheir study, in addition to the fact that their patients wereolder (children and adults) and had been previously immu-nized.Although limited to a small number of patients, especially

adults, our results are encouraging and suggest that PCRcould represent an interesting complementary tool for directdiagnosis of pertussis. A larger prospective clinical studywill be necessary to confirm these results. In our study,rapidity was an advantage with PCR because the completeprocedure took 3 days compared with 6 to 8 days forconventional culture. However, because of the necessity formultiple precautions to avoid bacterial and DNA contami-nation, PCR requires a well-trained operator using differentPCR-dedicated rooms. For this reason, the PCR methodpresently does not seem to be a routine procedure and thenasopharyngeal culture is still the reference diagnosticmethod for routine diagnosis of pertussis. However, becauseof its high sensitivity and specificity, the PCR technique, inassociation with serology, promises to be a valuable diag-nostic tool for epidemiological studies.

ACKNOWLEDGMENTS

We are grateful to Isabelle Saint Girons and Guy Baranton forconstant interest in this work and critical reading of the manuscriptand to Iain Old for editorial assistance with English usage.We thank M. J. Quentin-Millet (Pasteur Merieux Serum et Vac-

cins) for providing purified PT, A. Rappuoli and C. Locht forproviding the BP-TOX6 and BPRA B. pernussis mutant strains, N.El Solh for providing S. aureus purified DNA, J. P. Carlier forproviding the Corynebactenum spp. bacterial suspension, D. Dus-soix for providing the Mycoplasma pneumoniae purified DNA, C.Tram for providing the L. pneumophila purified DNA, J. P. Latgefor providing the Aspergillus spp. purified DNA, and P. Boiron forproviding the C. albicans purified DNA.

This work was supported by funds from the Institut PasteurFondation.

REFERENCES1. Antoine, R., and C. Locht. 1990. Roles of the disulfide bond and

the carboxy-terminal region of the S1 subunit in the assembly

and biosynthesis of pertussis toxin. Infect. Immun. 58:1518-1526.

2. Arciniega, J. L., E. L. Hewlett, F. D. Johnson, A. Deforest,S. G. F. Wassilak, I. M. Oronato, C. R. Manclark, and D. L.Burns. 1991. Human serologic response to envelope-associatedproteins and adenylate cyclase toxin of Bordetella pertussis. J.Infect. Dis. 163:135-142.

3. Arico, B., R. Gross, J. Smida, and R. Rappuoli. 1987. Evolu-tionary relationships in the genus Bordetella. Mol. Microbiol.1:301-308.

4. Bass, J. W., and S. R. Stephenson. 1987. The return of pertussis.Pediatr. Infect. Dis. J. 6:141-144.

5. Bernet, C., M. Garret, B. de Barbeyrac, C. Bebear, and J.Bonnet. 1989. Detection of Mycoplasma pneumoniae by usingthe polymerase chain reaction. J. Clin. Microbiol. 27:2492-2496.

6. Burg, J. L., C. M. Grover, P. Pouletty, and J. C. Boothroyd.1989. Direct and sensitive detection of a pathogenic protozoan,Towoplasma gondii, by polymerase chain reaction. J. Clin.Microbiol. 27:1787-1792.

7. Charles, I. G., L. J. Li, R. Strugnell, K. Beesley, M. Romanos,G. Dougan, P. Novotny, I. Heron, M. A. Jensen, C. R. Man-clarck, M. J. Brennan, and N. F. Fairweather. 1990. Repeatsequence motifs constitute the immunodominant regions of theP.69 protein, pertactin, from Bordetella pertussis: comparisonwith repeat sequences from Bordetella parapertussis and Bor-detella bronchiseptica, p. 136-140. In C. R. Manclark (ed.),Proceedings of the Sixth International Symposium on Pertussis.Department of Health and Human Services publication no.90-1164. Public Health Service, Bethesda, Md.

8. Cherry, J. D., P. A. Brunell, G. S. Golden, and D. T. Karzon.1988. Report of the task force on pertussis and pertussisimmunization. Pediatrics 81(Suppl.):930-984.

9. Edwards, K. M., and D. T. Karzon. 1990. Pertussis vaccines.Pediatr. Clin. N. Am. 37:549-566.

10. Glare, E. M., J. C. Paton, R. R. Premier, A. J. Lawrence, andI. T. Nisbet. 1990. Analysis of a repetitive DNA sequence fromBordetella pertussis and its application to the diagnosis ofpertussis using the polymerase chain reaction. J. Clin. Micro-biol. 28:1982-1987.

11. Granstrom, G., B. Wretlind, and M. Granstrom. 1991. Diagnos-tic value of clinical and bacteriological findings in pertussis. J.Infect. 22:17-26.

12. Grimprel, E., P. J. Sanchez, G. D. Wendel, J. M. Burstain, G. H.McCracken, Jr., J. D. Radolf, and M. V. Norgard. 1991. Use ofpolymerase chain reaction and rabbit infectivity testing to detectTreponema pallidum in amniotic fluid, fetal and neonatal sera,and cerebrospinal fluid. J. Clin. Microbiol. 29:1711-1718.

12a.Guiso, N., E. Grimprel, I. Anjak, and P. Begue. Eur. J. Clin.Microbiol. Infect. Dis., in press.

13. Hance, A. J., B. Granchamp, V. Levy-Frebault, D. Lecossier, J.Rauzier, D. Bocart, and B. Gicquel. 1989. Detection and identi-fication of mycobacterial DNA. Mol. Microbiol. 3:843-849.

14. He, Q., J. Mertsola, H. Soini, M. Skurnik, 0. Ruuskanen, andM. K. Viljanen. 1993. Comparison of polymerase chain reactionwith culture and enzyme immunoassay for diagnosis of pertus-sis. J. Clin. Microbiol. 31:642--645.

15. Houard, S., C. Hackel, A. Herzog, and A. Bollen. 1989. Specificidentification of Bordetella pertussis by the polymerase chainreaction. Res. Microbiol. 140:477-487.

16. Karsenty, J. Y., C. Roure, and G. Vidal-Trecan. 1990. Le pointsur: la coqueluche en France. Bull. Epidemiol. Habdomadaire19:81-82.

17. Locht, C., and J. M. Keith. 1986. Pertussis toxin gene: nucle-otide sequence and genetic organization. Science 232:1258-1264.

18. Long, S. S., C. J. Welkon, and J. L. Clark. 1990. Widespreadsilent transmission of pertussis in families: antibody correlatesof infection and symptomatology. J. Infect. Dis. 161:480-486.

19. Mahbubani, M. H., A. K. Bej, R. Miller, L. Haff, J. DiCesare,and R. M. Atlas. 1990. Detection of Legionella with polymerasechain reaction and gene probe methods. Mol. Cell. Probes4:175-187.

20. Merien, F., P. Amouriaux, P. Perolat, G. Baranton, and I. Saint

VOL. 31, 1993

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from

J. CLIN. MICROBIOL.

Girons. 1992. Polymerase chain reaction for detection of Lep-tospira spp. in clinical samples. J. Clin. Microbiol. 30:2219-2224.

21. Mertsola, J., 0. Ruuskanen, E. Eerola, and M. K. Viljanen.1983. Intrafamilial spread of pertussis. J. Pediatr. 103:359-363.

22. Mortimer, E. A. 1990. Pertussis and its prevention: a familyaffair. J. Infect. Dis. 161:473-479.

23. Nelson, J. D. 1978. The changing epidemiology of pertussis inyoung infants. Am. J. Dis. Child. 132:371-373.

24. Olcen, P., A. Bickman, B. Johansson, E. Esbjorner, E. Torn-qvist, J. Bygraves, and W. L. McPheat. 1992. Amplification ofDNA by the polymerase chain reaction for the efficient diagno-sis of pertussis. Scand. J. Infect. Dis. 24:339-345.

25. Olin, P. 1990. New conclusions and lessons learned from thevaccine trial in Sweden, p. 299-302. In C. R. Manclark (ed.)Proceedings of the Sixth International Symposium on Pertussis.

Department of Health and Human Services, publication no.

90-1164. Public Health Service, Bethesda, Md.26. Onorato, I. M., and S. G. F. WassilaL 1987. Laboratory

diagnosis of pertussis: the state of the art. Pediatr. Infect. Dis.J. 6:145-151.

27. Relman, D. A., M. Domenighini, E. Tuomanen, and R. Rappuoli.1989. Filamentous hemagglutinin of Bordetella pertussis: nucle-otide sequence and crucial role in adherence. Proc. Natl. Acad.Sci. USA 86:2637-2641.

28. Wirsing Von Konig, C. H., J. E. Hoppe, A. Tacken, and H.Finger. 1990. Detection of Bordetella pertussis in clinical spec-

imens, p. 315-321. In C. R. Manclarck (ed.), Proceedings of theSixth International Symposium on Pertussis. Department ofHealth and Human Services publication no. 90-1164. PublicHealth Service, Bethesda, Md.

2750 GRIMPREL ET AL.

on April 23, 2021 by guest

http://jcm.asm

.org/D

ownloaded from