JOURNAL OF CLINICAL MICROBIOLOGY, June 1989. p. 1222-12290095-1137/89/061222-08$02.00/0Copyright © 1989, American Society for MiCrobiology

Antilipopolysaccharide Antibodies and Differential Diagnosis ofChronic Pseudomonas aeruginosa Lung Infection in Cystic FibrosisANDERS FOMSGAARD,lt* BO DINESEN,2 GEOFFREY H. SHAND,3 TANJA PRESSLER,3 AND NIELS H0IBY3

Endotoxin Laboratory' and Department of Clinical Microbiology,3 Rigshospitalet, DK-2200 Copenhagen, andNordisk Gentofte A/S, DK-2820 Gentofte,2 Denmark

Received 8 December 1988/Accepted 23 February 1989

Chronic lung infection in cystic fibrosis is characteristically associated with polyagglutinable, serum-sensitive, mucoid strains of Pseudomonas aeruginosa. Enzyme-linked immunosorbent assay (ELISA) methodsfor standard-free quantitation of immunoglobulin G (IgG) and IgM antibodies to P. aeruginosa lipopolysac-charides (LPSs) have been developed. We now report the development of assays for quantitation of monomerand dimer total IgA and IgA anti-LPS antibodies. Use of these methods in diagnosis of early chronic P.aeruginosa lung infection was assessed. IgG and IgA anti-LPS levels increased significantly at the onset ofchronic infection and continued to increase to very high levels in the later stages of infection. IgM anti-LPSlevels also rose at the onset of chronic infection but did not increase further. The function of true- andfalse-positive rates was illustrated by using various concentrations of IgG, IgA, and IgM anti-LPS fordiscrimination of patients. Values that gave optimum separations were used for statistical evaluation of thediagnostic sensitivities and specificities of anti-LPS antibody concentrations. The results obtained in theseassays were compared with a diagnosis, based on the number of precipitins in crossed immunoelectrophoresis,of serum samples from cystic fibrosis patients. In 64 paired serum samples taken before and immediately afterthe onset of chronic infection, as defined by crossed inimunoelectrophoresis precipitins, the predictive valuesof a positive ELISA were 86% for IgG and 89% for IgA. The predictive values for a negative ELISA were 98%for IgG and 97% for IgA. Results of the IgM anti-LPS ELISA had a lower predictive value. Immunoblottingand absorption studies showed that IgG anti-LPS antibodies were directed specifically against LPS of P.aeruginosa. ELISAs were developed to determine the specific IgG subclasses involved. The increase in IgGanti-LPS involved all four subclasses. Highest anti-LPS titers were seen with IgGl and IgG4, but the largestrelative increases were seen with IgG2 and IgG3.

Chronic pulmonary infection with Pseudomonas aerugi-nosa is characteristic of cystic fibrosis (CF) patients and isthe major factor determining the severity of illness andmortality (28). After a period of relatively harmless coloni-zation of the mucosal lining by P. aeruginosa, the clinicalcondition begins to deteriorate and symptoms of respiratoryinfection increase. Although long-term eradication of P.aeruginosa, once the infection has become established, isseldom achieved, exacerbations can be brought into remis-sion by regular intensive antibiotic treatment. Early andactive treatment regimens improve the lung function, well-being, and long-term survival of patients (20). It is thereforeimportant to identify the early stages of infection and theappropriate time to start intensive antimicrobial therapy.The establishment of chronic invasive, tissue-destroying

infection is associated with characteristic changes in thebacteria. Thus, P. aeruginosa isolated from sputum of CFpatients with chronic pulmonary infection are often mucoid,serum sensitive, and polyagglutinable (14). We have previ-ously shown that the lipopolysaccharide (LPS) (O antigen;endotoxin) of polyagglutinating clinical isolates is deficient inthe O-polysaccharide chain but conserved in the core oli-gosaccharide and lipid A part of the molecule (5). Theprevalence of 0-deficient strains in which common antigenicsites are accessible to antibodies makes the LPS of suchstrains potentially useful in diagnostic assays of chronic lung

* Corresponding author.t Present address: Laboratory of Infectious Diseases, National

Institute of Allergy and Infectious Diseases, 12441 Parklawn Drive,Rockville, MD 20852.

infection. Serological assays determining antibodies to P.aeruginosa (1, 2, 19, 23) and P. aeruginosa products (12, 16,21) have previously been found to be of value in monitoringboth the course of infection and treatment. However, noneof the tests using single antigens has been able to detect theantibody response in all P. aeruginosa-infected CF patients.The number of different kinds of precipitating antibodies(precipitins) against P. aeruginosa whole-cell sonic extractsrises during the infection and is used in combination withbacteriological criteria to diagnose the onset of chronic P.aeruginosa lung infection in Danish CF patients (15). How-ever, more sensitive assays detecting antibodies of differentclasses and subclasses specific for defined antigens would beuseful for characterization of early specific antibody re-sponses in these patients. Information about specific anti-body responses could be especially useful in the earlydiagnosis of cases in which there is a rapid development ofthe lung infection.

In a longitudinal study of concentrations in serum ofantibodies to Pseudomonas LPSs during the course ofchronic lung infection in CF patients, we found a goodpositive correlation between immunoglobulin (IgG) antibod-ies to LPS from a clinical isolate of P. aeruginosa 0:3/9 andthe number of precipitins to Pseudomonas sonic extracts (8).An increase in the number of precipitins, in turn, reflects theseverity of the disease (8, 15). In the study reported here, wehave further characterized the immune response to LPS of apolyagglutinable P. aeruginosa strain involving IgG, IgGsubclasses 1 to 4, IgM, and IgA at the onset of chronic P.aeruginosa lung infection in CF patients and evaluated thediagnostic usefulness of these antibodies.

1222

Vol. 27, No. 6

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

DIAGNOSIS OF CHRONIC P. AERUGINOSA INFECTION IN CF

MATERIALS AND METHODS

Normal values. To establish normal values for concentra-tions of anti-LPS antibodies, serum samples from 25 infants(0 to 1 year of age), 15 children (2 to 14 years), and 60 adults(15 to 40 years) with no history of P. aeruginosa infectionwere investigated.CF patients. CF patients were monitored regularly by

clinical and bacteriological examinations at the Danish CFCenter at Rigshospitalet. The diagnosis of CF was based onaccepted criteria, including abnormal sweat electrolyte lev-els, exocrine pancreatic insufficiency, and altered pulmonaryfunction (28). Noncolonized CF patients were defined as CFpatients with no growth of P. aeruginosa in sputum atmonthly examinations and with no or one precipitin incrossed immunoelectrophoresis (CIE). The onset of chronicP. aeruginosa lung infection was defined as the point atwhich the bacteria had been found in every sputum samplefor at least 6 months and serum contained two or more CIEprecipitins to P. aeruginosa (15). Serum samples obtainedwhile the patients were noncolonized (CFO) were availablefrom 64 patients. These were paired with the first serumsamples taken after the onset of chronic P. aeruginosa lunginfection (CF+) for enzyme-linked immunosorbent assay(ELISA) determination of IgM, IgG, and IgA anti-Pseudo-monas LPS antibody concentrations. Serum samples from10 CF patients with lung infection for 8 to 16 years whoshowed consistent growth of P. aeruginosa in sputum were

examined for precipitins and anti-LPS antibodies and usedas examples of samples from patients with late chronicinfection (CF+ +).LPS antigens. A polyagglutinating mucoid P. aeruginosa

170 0-group 3/9 bacteriophage type 21,48,68,109,352,1214,F3,M4,colll+ was isolated from sputum of a CF patientwith chronic pulmonary infection. LPS was extracted by thephenol-chloroform-petroleum ether method (10). The LPSwas purified by successive ultracentrifugation steps, purgedof associated inorganic cations and low-molecular-weightbasic amines by electrodialysis (9), and converted to itsuniform triethylamine salt form. The detailed chemical com-position and immunochemical characterization of this O-polysaccharide-deficient LPS are reported elsewhere (5).Pseudomonas standard antigens. Water-soluble antigens

from P. aeruginosa strains representing the 17 serotypes ofthe International Antigen Typing Scheme were prepared as

previously described (15) and used as a standard antigen inCIE.Anti-LPS ELISAs. For ELISAs, 96-well polystyrene mi-

crotest plates (Immunoplate I; Nunc A/S, Roskilde, Den-mark) were coated with 1 ,ug of P. aeruginosa 0:3/9 LPS per100 ,ui in phosphate-buffered saline containing 0.02 MMgCi2. The coated plates were covered and stored at 4°C,with the coating solution remaining in the wells for at least 72h, and were usable for at least 6 months. Before use andbetween subsequent incubation steps, the plates werewashed with 0.15 M NaCI-0.1% Tween 20 (washing solu-tion). The test sera were diluted 1:50 and 1:500 in phosphate-buffered saline-0.1% Tween 20 and incubated in the LPS-coated wells (100 p1i per well) overnight at 22°C. Patient sera

containing high concentrations of anti-LPS antibodies werediluted further (1:5,000 and 1:50,000) and reassayed. Ai-though the serum incubation was sufficiently completewithin 2 h, overnight incubation was used for practicalreasons. After washing, incubation proceeded for 1 h, using100 ,ul of peroxidase-conjugated rabbit anti-human IgG (1:15,000; Dakopatts, Glostrup, Denmark), rabbit anti-human

IgM (1:2,000; Dakopatts), or rabbit anti-human IgA (1:5,000;Dakopatts) diluted in phosphate-buffered saline-.1%Tween 20 per well. Color was developed by addition to eachwell of 100 ,ul of enzyme substrate consisting of 14 mg ofo-phenylenediamine and 7.5 ,ul of hydrogen peroxide (30%,vol/vol) in 18 ml of distilled water. The reaction was termi-nated after 30 min by addition of 150 ,ul of 2.5 M sulfuric acidper well. A492 was measured with a photometer (modelSLT-210; Kontron, Zurich, Switzerland). Quantitative re-sults of IgG, IgM, and IgA anti-LPS antibodies were ob-tained by comparison with results for a standard CF serumpool. Concentrations of specific anti-LPS antibodies in thestandard CF serum pool were determined by standard-freequantitation (6). Briefly, double-antibody sandwich ELISAsfor total IgG, IgM, and IgA were performed simultaneouslyon a separate part of the ELISA plate, using the sameconjugate, buffers, and assay conditions. The concentrationsof specific IgG, IgM, and IgA anti-Pseudomonas LPS anti-bodies in the CF serum pool were then read on the respec-tive standard curves for total IgG, IgM, and IgA. Resultswere corrected for differences in antibody binding as de-scribed previously (6).The double-sandwich ELISA for quantitation of IgG and

IgM has been described elsewhere (3, 7). A description ofthe ELISA for total IgA is given below.ELISA for total IgA in serum. Each well of polystyrene

microtest plates was coated with 100 ,ui of rabbit anti-humanIgA (Dakopatts) diluted 1:5,000 in 0.1 M carbonate buffer(pH 9.8). Plates were stored at 4°C for at least 72 h and usedwithin 3 months. After four washes with washing solution,the plates were incubated overnight at 22°C with serialdilutions from 5- to 200-,ug/liter concentrations of an IgAstandard human serum (Seronorm 103; Nycomed, Oslo,Norway) in phosphate-buffered saline-O.1% Tween 20-0.5%bovine serum albumin (incubation buffer). After beingwashed, the wells were incubated for 1 h with peroxidase-conjugated rabbit anti-human IgA (a-chain specific; Dako-patts) diluted 1:10,000 in the incubation buffer. Color wasdeveloped as described above. Quantitative results wereobtained from a calibration graph (3, 6, 7). Measurement ofIgA in serum and colostrum by this ELISA was comparedwith results obtained by conventional rocket immunoelec-trophoresis (17).ELISA for anti-P. aeruginosa LPS IgG subclasses. ELISAs

were developed to investigate the distribution and develop-ment of specific anti-P. aeruginosa 0:3/9 LPS IgG sub-classes. Serum samples from healthy donors were used tooptimize the incubation conditions. Use of sera from healthypersons ensured the development of an assay sensitiveenough to measure anti-LPS antibodies in healthy controlsand in noncolonized CF patients. The optimal concentra-tions for the conjugates were determined as previouslydescribed (3). For assay of anti-LPS IgG subclasses, perox-idase-conjugated mouse monoclonal antibodies specific forhuman IgGl (1:5,000), IgG2 (1:1,000), IgG3 (1:500), or IgG4(1:1,000) in PBS-Tween 20 (Janssen Biochemica, Beerse,Belgium) were used. Coating of microtest plates, incubationtimes, and experimental conditions were as described above.Anti-LPS IgGl to -4 titers were expressed as the reciprocaldilutions of samples corresponding to an A492 of 0.500.Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

and immunoblotting. LPS preparations from P. aeruginosaF-D1 (0:6) and clinical isolates 1118 (0:3), 174 (0:9), 170(0:3/9), 15704/83 (0:3/6/9/10), and 14580/83 (0:3/6/9/10) (5)and from Shigella sonnei, Klebsiella pneumoniae, Salmo-nella minnesota R60 (Ra), Escherichia coli Ra, E. coli

VOL. 27, 1989 1223

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

1224 FOMSGAARD ET AL.

ELISAIgA/g xrI

14

10

6-

2-

2 6 10 14IgA/g xl-

Rocket immunoelectrophoresis

FIG. 1. Comparison of IgA concentrations in 99 serum samples,using the new quantitative ELISA and rocket immunoelectrophore-sis. The regression line ( ) was computed by means of linearleast squares.

0111:B4, and E. coli 055:B5 were treated for 5 min at 100°Cin 0.05 M Tris hydrochloride (pH 6.8)-2% sodium dodecylsulfate-10% sucrose-0.01% bromphenol blue and separatedon a sodium dodecyl sulfate-polyacrylamide gel containing 5and 14% acrylamide in the stacking and separating gels,respectively. Gels were silver stained (26) or the sampleswere transferred to nitrocellulose paper as previously de-scribed (8). After blocking in 1% gelatin-Tris-buffered saline(TSG), the nitrocellulose paper was incubated for 2 h withCF serum diluted 1:500 in TSG. The nitrocellulose paper wasthen washed three times in Tris-buffered saline and incu-bated for 1 h with peroxidase-conjugated rabbit antibodies tohuman IgG, IgM, or IgA (Dakopatts) diluted 1:200 in TSG.Color was developed in 0.003% (wt/vol) 4-chloro-1-naph-thol-0.05% (vol/vol) H202.

Precipitating antibodies against P. aeruginosa. P. aerugi-nosa precipitins were determined by CIE as describedearlier (15). Briefly, 2 ,ud of the standard antigen preparation(22 ,ug/ml of protein) was applied to a well and separated inthe first dimension (10 V/cm, 1 h) in 1% agarose gels.Second-dimension separation was performed for 18 h at 1 to2 V/cm into 1% agarose containing 15 ,ul of patient serum per

cm2. The immunoprecipitates were stained with Coomassiebrilliant blue, and the number of precipitin peaks wascounted.

RESULTSELISA for total IgA and IgA anti-LPS. A comparison of

IgA concentrations in serum as measured by conventional

Dilution factor

FIG. 2. Dose-response curves of IgA in serum and colostrum as

measured by the IgA ELISA. The dose-response curve of IgAanti-P. aeruginosa LPS antibodies in the CF serum pool was

assayed simultaneously to confirm parallelism with the extendedIgA standard curve (*) used for quantitation of specific IgA anti-LPS.

rocket immunoelectrophoresis and the IgA ELISA is shownin Fig. 1. The correlation of the two assays (excluding thetwo extreme values [rocket, ELISA] = [<0.01, 0.002] and[10.9, 14.2]) can be expressed as ELISA = 0.06 + 0.93 xrocket (n, 97, r, 0.96). The dose-response curves of IgA inserum (predominantly monomer IgA) and colostrum (pre-dominantly secretory dimer IgA) as measured by the IgAELISA are shown in Fig. 2. Identical maximal bound en-

zyme activity as well as parallel dilution curves of serum andcolostrum IgA indicate that the a-chain was detected withequal efficiency regardless of molecular configuration.The dose-response curve of IgA anti-P. aeruginosa LPS in

the standard CF serum pool was assayed simultaneously toconfirm parallelism with the IgA standard curve (Fig. 2). Thedetection limit of IgA bound by anti-IgA or LPS, defined asthe lowest concentration of IgA significantly different from 0at the 95% confidence limit, was below 8.4 ,ug/liter (in 15assays). The relative standard intra- and interassay deviationand the plate-to-plate variation on different days for assay ofIgA anti-P. aeruginosa LPS (range, 0.5 to 100 ,ug/ml) in 15serum samples of low (5 samples), medium (5 samples), andhigh (5 samples) antibody concentrations were 4, 8, and10%, respectively.Anti-Pseudomonas LPS antibody concentrations. Table 1

TABLE 1. Concentrations of anti-P. aeruginosa 0:3/9 LPS and numbers of precipitins against P. aeruginosa cellsin CF patients and healthy controls

Patient group No. Median anti-LPS concn (,ug/ml) No. of(age [yr]) tested IgM IgG IgA CIE-Pb

ControlsInfants (0-1) 25 6 (0.5-16)c 4 (1-12) 0.5 (0.2-3) 0-1Children (2-14) 15 15 (5-34) 6 (2-13) 1 (0.5-4) 0-1Adults (15-40) 60 11 (10-40) 15 (4-30) 4 (1-9) 0-1

CFO (2-11) 64 29 (1-110) 19 (1-100) 3 (1-18) 0-1CF+ (3-12) 64 54 (19-250) 125 (50-700) 25 (7-160) 2-7CF++ (12-27) 10 30 (15-110) 986 (458-1,898) 160 (67-456) 15-48

a CFO, CF patients without precipitins detectable in CIE; CF+, same CF patients when chronic P. aeruginosa lung infection was diagnosed; CF++, CFpatients at late stage of chronic lung infection (3 to 15 years of infection).

b Number of precipitins to P. aeruginosa standard antigens (0:1-17) in CIE.Shown in parentheses are 95% intervals.

J. CLIN. MICROBIOL.

/-l-

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

DIAGNOSIS OF CHRONIC P. AERUGINOSA INFECTION IN CF

Numbers Numbers4b CF+ 40 CF+

n =64201 2]

O2O 020 >70

Numbers40 CF+

20

0

60 6

40 CF 3 40n -64

22

80160 ,240 '320>320 20

100-

80-

60-

40-

20-

0-

100o

80-

60-

Normals 40-

n =100

20-

CFO

* 4'0 ' 60 ' >70

Normals

40-

2]

8

6

4

2

80 160 240 320>320 20 40 60 >7

IgG anti-PA LPS (mg/l) IgA anti-PA LPS (mg/t)

CF O

' i '>14 40

Normals

IgM anti-PA LPS (mg/l)FIG. 3. Distribution of IgG, IgA, and IgM anti-P. aeruginosa LPS concentrations in normal and CF patient sera before (CF0) and

immediately after (CF+) the onset of chronic P. aeriuginosa lung infection.

shows the number of precipitins in CIE and the concentra-tions of anti-LPS antibodies in healthy subjects and CFpatients. The number of precipitins in normal subjects waszero to one. In contrast, the IgM, IgG, and IgA anti-P.aeruginosa LPS antibody concentrations in normal subjectswere age related. The IgG and IgA anti-LPS antibodyconcentrations reached the highest levels in adults, whereasIgM anti-P. aeruginosa LPS concentrations reached thehighest levels in children. The median IgA anti-P. aerugi-nosa LPS concentrations were the same for healthy subjectsand CF patients before the onset of chronic lung infection(CFO) but increased significantly (P < 0.001, paired t test) inCF patients at the onset of chronic infection (CF+). IgG andIgA anti-LPS concentrations and the number of precipitinscontinued to rise to very high levels in later stages of theinfection (Table 1). In contrast, the median concentrations ofIgM anti-P. aeruginosa LPS antibodies increased from nor-

mals, via CFO to CF+ patients, after which no furtherincrease was found. In fact, there was a small decrease inmedian IgM anti-P. aeruginosa LPS in early (CF+) and late(CF+ +) chronically infected patients (Table 1). Figure 3shows the distribution of IgG, IgA, and IgM anti-P. aerugi-nosa LPS antibodies among healthy subjects and 64 CFpatients before and immediately after diagnosis of chronic P.aeruginosa lung infection. Significant differences were ob-served in the distribution of IgG, IgA, and IgM anti-LPSantibodies among the three groups (normal subjects, CFOpatients, and CF+ patients; P < 0.001 in all three cases,chi-square test). There was, however, some overlap in IgGand IgA anti-LPS concentrations between CFO and CF+patients (Fig. 3).

The nosographic sensitivities and specificities of the anti-P. aeruginosa LPS ELISAs were evaluated in CF+ and CFOsamples defined as infected and noninfected, respectively,by the CIE test (Table 2). The predictive value of a positiveanti-LPS ELISA (PVPS), used to detect early P. aeruginosainfection, was the diagnostic specificity, and the predictive

TABLE 2. Distribution of CF0 and CF+ patients and diagnosticvalues of anti-LPS ELISAs

Anti-LPS assay with given No. of patientsconcn (,ug/ml) of: CF+ CFO

IgG"0-40 1 54>40 63 10

IgAh0-8 2 56>8 62 8

IgM'0-33 17 38>33 47 26

Significant difference of CF+ versus CFO, P < 0.0005 (chi-square test);nosographic specificity, 54/64 = 0.84; nosographic sensitivity, 63/64 = 0.98;PVp,.;, 63/73 = 0.86; PVneg, 54/55 = 0.98.

b Significant difference of CF+ versus CFO. P < 0.0005 (chi-square test);nosographic specificity, 56/64 = 0.88; nosographic sensitivity, 62/64 = 0.97;PVPO,' 62/70 = 0.89; PVneg, 56/58 = 0.97.

Significant difference of CF+ versus CFO, P < 0.0005 (chi-square test);nosographic specificity, 38/64 = 0.59; nosographic sensitivity, 47/64 = 0.73;PVro, 47/73 = 0.64; PVnC,5 38/55 = 0.69.

VOL. 27, 1989 1225

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

1226 FOMSGAARD ET AL.

CE

F PR

FIG. 4. Discrimination ability of the diagnostic anti-P. aerugi-nosa LPS ELISAs, illustrated by means of receiver-operatingcharacteristic curves (27) describing the relationship between thetrue-positive rate (TPR; rate of positives among patients withchronic infection) and the false-positive rate (FPR; rate of positivesamong patients without infection) when the criterion of positivevaries over the range of anti-LPS concentrations. The more thiscurve is removed from the line of no discrimination, the better is theperformance. Symbols: (A, CF+ values compared with normalvalues, using IgG anti-LPS; *, CF+ values compared with CFOvalues, using IgG anti-LPS; O, CF+ values compared with CFOvalues, using IgA anti-LPS; x, CF+ values compared with CFOvalues, using IgM anti-LPS.

value of a negative test (PVneg), used to rule out chronicinfection, was the diagnostic sensitivity (Table 2). PVps andPVneg of the IgG and IgA anti-P. aeruginosa LPS ELISAswere high in both cases (Table 2). Higher values were

obtained if a combination of elevated concentrations in bothtests was used: PVP,, and PVneg of IgG anti-LPS at >40jtg/ml and IgA anti-LPS at >8 ,ug/ml were 95% (61 of 64).However, PVPO, and PVneg for elevated IgG or IgA anti-LPSconcentrations or both were 81 (64/79) and 100% (49/49). Incomparison, the sensitivity and specificity of the IgM anti-LPS ELISA were lower (Table 2).The values of the IgG and IgA anti-LPS concentrations

used to distinguish chronically infected CF patients fromuninfected subjects may, however, be chosen differently andare a balance of false-positive and false-negative results. Thesuccessive overlap between CF+ and healthy subjects as

well as between CF+ and CFO patients can be illustrated forthe IgG, IgA, and IgM anti-LPS ELISAs by means ofreceiver-operating characteristic curves (27), i.e., the rela-tionship between the true-positive rate (rate of positivesamong patients with infection) and the false-positive rate(rate of positives among patients without infection), whenthe criterion of positive varies over the range of anti-LPSconcentrations (Fig. 4). By using this method, the concen-

trations of anti-LPS antibodies resulting in the minimaloverlap of CF+ and CFO patients, as defined by CIEprecipitins, was found and used to evaluate the diagnosticvalues presented above. Figure 4 shows equally good sepa-

ration of CF+ and CFO patients by the IgG and the IgAanti-P. aeruginosa LPS ELISAs. The IgM anti-LPS curve

was always lower than the IgG and IgA anti-LPS curves,

demonstrating less separation of patients by IgM anti-LPSregardless of the concentrations chosen. Figure 4 also dem-

B

FIG. 5. (A) Immunoblot of a control serum pool (1:50) of 40normal donors selected for high (>40 p.g/ml) concentrations of IgGantibodies to various LPSs (6) to LPSs from (left to right) P.aeruginosa F-Dl, 1118, 174, 170, 15704/83, and 14580/83, Shigellasonnei, K. pneumoniae, S. minnesota R60, and E. coli Ra, 0111:B4,and 055:B5. (B) Corresponding immunoblot of a typical example ofthe reaction of IgG anti-LPS in a CF serum sample (1:500) withelevated lgG anti-P. aeruginosa LPS antibodies. The sample shownis from a CFO patient with elevated IgG anti-P. aeruginosa LPS.

onstrates better separation of CF+ patients from normalsubjects than from CFO patients by IgG anti-LPS concentra-tions.

Specificity of anti-LPS antibodies. The specificities of theanti-P. aeruginosa LPS antibodies in sera with elevatedanti-LPS concentrations were investigated further. The anti-P. aeruginosa LPS antibodies could in each case be ab-sorbed by P. aeruginosa 0:3/9 LPS coated onto sheeperythrocytes (11). None of the 10 CFO serum samples withelevated IgG anti-P. aeruginosa LPS antibody concentra-tions had elevated (>40 ptg/ml) concentrations of antibodiesto other LPSs (including strains of E. coli and of Salmonellaand Serratia species) as measured by ELISA (6). In immu-noblotting, IgG anti-LPS antibodies in the 10 CFO serumsamples with elevated IgG anti-P. aeruginosa LPS concen-trations reacted to P. aeruginosa 0:3/9 LPS and also toLPSs from other clinical isolates of P. aeruginosa but onlyvery weakly to LPSs of other gram-negative bacteria (Fig.5).

Anti-P. aeruginosa IgG subclasses. Figure 6 shows thedistribution of the IgG subclass specificity of the anti-LPSantibodies of normal subjects and CFO and CF+ patients.The increase in IgG anti-LPS from CFO to CF+ involved ailfour subclasses. Highest anti-LPS titers were seen in IgGland IgG4. The largest relative increase within an anti-LPSIgG subclass was found in IgG3 and IgG2.

J. CLIN. MICROBIOL.

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

DIAGNOSIS OF CHRONIC P. AERUGINOSA INFECTION IN CF

CI)

IgG3 anti-LPS IgG4 anti-LPS

8- 200

4- 100

o0N CFO CF+ N CFO CF+

FIG. 6. Distribution of the IgG subclass specificity (mediantiters) of the anti-P. aeruginosa LPS antibodies in 100 normal and 64CF serum samples before (CFO) and just after (CF+) chronic lunginfection. ELISA titer is expressed as the reciprocal of the dilutionat a net absorption (A492) of 0.500.

DISCUSSION

A sandwich ELISA for quantitation of total IgA was

developed by using microtest plates and commercially avail-able reagents. The quantitative results of IgA levels in sera

obtained with this ELISA correlated well (r, 0.96) with thoseobtained by traditional rocket immunoelectrophoresis (Fig.1). The high sensitivity and demonstrated detection of a-

chains in both monomer and secretory dimer forms makesthis ELISA convenient for measuring total IgA in sera andsecretions. In comparison with rocket immunoelectrophore-sis, the ELISA for total IgA has a higher capacity, is moresensitive, and is simpler, faster, and less expensive.

In this study, ELISA for total IgA was used together withan indirect ELISA for IgA anti-LPS antibodies to quantitatespecific IgA anti-P. aeruginosa LPS antibodies in a com-bined design (Fig. 2). By reading the anti-LPS concentrationon the simultaneously performed standard curve for totalIgA, the amounts of specific anti-LPS antibodies could beconveniently estimated in the CF serum pool as describedearlier (6). The same principle was used for quantitation ofthe IgG and IgM anti-P. aeruginosa antibodies in normalsubjects and CF patients.The results showed that there was no significant difference

in the number of precipitins and the median IgA anti-P.aeruginosa LPS antibody concentrations between healthysubjects and CF0 patients. In contrast, the median concen-tration of IgM anti-P. aeruginosa LPS antibodies and, to alesser extent, IgG anti-LPS, increased from levels in age-matched normals, via CFO to CF+ patients. No furtherincrease in IgM anti-LPS was then found. In fact, there wasa small decrease in median IgM anti-P. aeruginosa LPS fromearly (CF+) and late (CF+ +) chronically infected patients(Table 2). These findings are in agreement with earlier resultsand are consistent with IgM being an early and IgG a laterimmune response to bacterial infections (4, 8). The lesspronounced and persistent IgM anti-LPS response is alsoconsistent with LPS being a T-cell-independent antigen (4).

Establishment of a diagnostic test defining the onset ofchronic lung infection is complicated because the increase inconcentrations of anti-P. aeruginosa LPS antibodies other

than IgM is a continuum toward extremely high values in thelater stages of infection (Table 1; 8). A useful definition of theonset of chronic P. aeruginosa lung infection is the demon-stration of two or more precipitins against P. aeruginosawhole-cell sonic extracts in CIE (15). A comparison wastherefore made of the diagnostic values of the anti-LPSELISAs and CIE. By these criteria, the IgM anti-P. aerug-inosa LPS concentration in serum was of less diagnosticsignificance than the concentrations of anti-LPS IgG and IgA(Fig. 3). The nosographic specificities and sensitivities of theIgG and IgA anti-P. aeruginosa LPS ELISAs were higherand showed both high PVPOS (86 and 89%) and high PVneg (98and 97%) (Table 2). The concentrations of IgG and IgAanti-P. aeruginosa LPS antibodies chosen for evaluation ofthe anti-LPS ELISA as a diagnostic test are the points ofminimal overlap between the groups tested (CF+ versusCFO) (Fig. 4). However, a different diagnostic value ofanti-LPS concentrations could be chosen on the basis ofother specific clinical and practical considerations. Thelower nosographic specificity of the IgG anti-LPS ELISA incomparison with CIE was due to 10 of the 64 CFO patientshaving elevated anti-LPS concentrations (Table 3). How-ever, 7 of these 10 patients did not have elevated IgA anti-P.aeruginosa LPS antibody levels. Similarly, five of the eightCFO patients with elevated IgA anti-P. aeruginosa LPSantibodies did not have elevated specific IgG anti-LPSantibody levels. Thus, the combination of both elevated IgGand IgA anti-P. aeruginosa LPS antibodies may be charac-teristic of chronically infected CF patients. The elevation ofIgA or IgG anti-P. aeruginosa LPS concentrations alonemay be an earlier sign of chronic lung infection.The IgG anti-P. aeruginosa LPS antibodies in serum

samples from the 10 CFO patients with elevated antibodyconcentrations were found to be specific for P. aeruginosaLPS in a series of experiments, including immunoblotting. Itis interesting that in immunoblotting (Fig. 5), most of thesamples from CF patients reacted with LPS from clinicalisolates of P. aeruginosa other than the 0:3/9 strain. Thisresult could be attributable to the similar chemical compo-sitions of these LPSs, which differ only in the molar ratios ofrhamnose, glucose, heptose, and alanine (5).The differences noted between the anti-LPS ELISA and

the CIE precipitins may reflect differences in the sensitivityof the two tests and suggest an earlier onset of the anti-Pseudomonas antibody response than is currently recog-nized. The anti-LPS ELISA method may be more sensitivethan the CIE precipitin method for detecting early lunginfection and distinguishing early infection from coloniza-tion. The higher sensitivity of the anti-LPS ELISAs is veryuseful for early detection of rapidly developing lung infec-tions.The anti-LPS ELISA described here is a diagnostic tool

comparable in sensitivity and specificity to CIE. Determina-tion of precipitins is time consuming and requires a labora-tory that routinely performs CIE, whereas the ELISAmethod has the advantage of easier handling of a largenumber of samples and rapid. quantitation of results by anELISA plate reader that can also be computerized. Anotheradvantage of the anti-LPS ELISA is that it uses a singlepurified antigen (a chemically defined LPS molecule) in aquantitative method capable of distinguishing different anti-body immunoglobulin classes and subclasses.An increase in median titers from CFO to CF+ was found

in all four IgG subclasses (Fig. 5), as previously reported byMoss et al. (18). However, substantial individual variationswere seen, and no single subclass was consistently elevated

VOL. 27, 1989 1227

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

1228 FOMSGAARD ET AL.

or depressed despite the overall increase in IgG response toP. aeruginosa LPS. The median titer of IgG2 anti-P. aerug-inosa LPS in CFO sera tended to be lower than in normalsera; however, this titer increased at the onset of chronicinfection (Fig. 4). Similar results have been reported byMoss et al. (18) and suggest a relative deficiency in IgG2antibody levels to polysaccharide antigens (25). Among theearly chronically infected CF patients (CF+), the highesttiters were seen in IgGl and IgG4. The different concentra-tions of conjugated antibodies used for detection of the IgGsubclasses makes direct comparison of the titers difficult.However, since the higher concentrations of conjugatedantibodies were used to improve sensitivity in order todetect CF sera of low anti-LPS immunoreactivity, it isreasonable to establish the range of immunoreactivity ofanti-P. aeruginosa IgG in early chronically infected CFpatients as IgGl > IgG4 > IgG2 > IgG3. The highestpercentage increase in median titers was, however, observedin IgG3 and IgG2 anti-LPS, both known to be primarilyantisaccharide antibodies (25). An increase in IgG3 and IgG4anti-P. aeruginosa LPS was also found by Shryock et al.(24). However, these authors failed to demonstrate any IgGlantibody response in CF sera. That finding is in contrast withthe results of this study and the generally held opinion thatopsonic IgGl constitutes the majority of the total IgG inserum and that saccharide antigens and LPS primarily in-duce IgGl antibodies (13). Moreover, the CF patients weexamined at later stages of chronic lung infection hadmanyfold-higher titers of all anti-P. aeruginosa LPS IgGsubclasses (data not shown). The reason for this discrepancyis not clear but could be partly explained by individualvariations in the 10 patients. Alternatively, there could betechnical differences in the sensitivity of the anti-LPSELISA, the LPS preparation, or the monoclonal anti-IgGlconjugate used. Whether the distribution among specific IgGsubclasses changes during the course of the chronic infectionas suggested by Moss et al. (18) is currently being investi-gated in a longitudinal study.The early increase in specific IgG and IgA anti-LPS

antibodies and the extremely high concentrations in laterstages of the infection suggest that P. aeruginosa LPS is ahighly relevant antigen to investigate in CF patients. Thisphenomenon is probably due to the pronounced biologicalactivity and toxicity of the LPS endotoxin molecule (22). IgGand IgM anti-LPS antibodies have been suggested to protectagainst the deleterious effects of endotoxinemia. However,continuous high concentrations of IgG and IgA anti-LPSantibodies may contribute to the formation of immune com-plexes and may activate complement and other mediators ofinflammation locally in the lungs, thus contributing to lungtissue damage. Although IgA is principally an antibody ofmucosal defense, the IgA anti-LPS levels in serum paralleledthe increase in IgG anti-LPS and CIE precipitins. The highconcentrations of IgA anti-P. aeruginosa LPS in the serumof CF patients would be expected because of the mucosalinflammation in the upper respiratory tract and could serveto eliminate LPS antigens by excretion via the bile.We recommend the quantitation of specific IgG and IgA

antibodies to P. aeruginosa LPS by ELISA as a diagnostictool for CF to distinguish between superficial colonizationand early or persistent infection caused by P. aeruginosa.The latter conditions would serve to initiate anti-P. aerugi-nosa therapy in our CF center.

ACKNOWLEDGMENTSWe thank Susanne Jensen for excellent technical assistance.

This study was supported by grant 12-7932 from the DanishMedical Research Council and by the Danish National Associationfor Prevention of Lung Disease.

LITERATURE CITED

1. Borowski, R. S., L. M. Stock, and N. L. Schiller. 1984. Devel-opment of an enzyme-linked immunosorbent assay for studyingPseudomonas aeruginosa cell surface antigens. J. Clin. Micro-biol. 19:736-741.

2. Brett, M. M., A. T. M. Ghonheim, and J. M. Littlewood. 1986.Serum antibodies to Pseudomonas aeruginosa in cystic fibrosis.Arch. Dis. Child. 61:1114-1120.

3. Dinesen, B. 1986. Immunoglobulin M, p. 71-82. In H. V.Bergmeyer (ed.), Methods of enzymatic analysis, vol. 10. VCHVerlagsgesellschaft, Weinheim, Federal Republic of Germany.

4. Feldmann, M., and S. Kontiaiuen. 1981. The role of antigen-specific T-cell factors in the immune response. Lymfokines2:87-123.

5. Fomsgaard, A., R. S. Conrad, C. Galanos, G. H. Shand, and N.H0iby. 1988. Comparative immunochemistry of lipopolysaccha-rides from typable and polyagglutinable Pseudomonas aerugi-nosa strains isolated from patients with cystic fibrosis. J. Clin.Microbiol. 26:821-826.

6. Fomsgaard, A., and B. Dinesen. 1987. ELISA for human lgGand IgM anti-lipopolysaccharide antibodies with indirect stan-dardization. J. Immunoassay 8:333-350.

7. Fomsgaard, A., B. Feldt-Rasmussen, M. Deckert, and B. Dine-sen. 1987. Micro-ELISA for the quantitation of human urinaryIgG. Scand. J. Clin. Lab. Invest. 47:195-198.

8. Fomsgaard, A., N. H0iby, G. H. Shand, R. S. Conrad, and C.Galanos. 1988. Longitudinal study of antibody response tolipopolysaccharides during chronic Pseudomonas aeruginosalung infection in cystic fibrosis. Infect. Immun. 56:2270-2278.

9. Galanos, C., and O. Luderitz. 1975. Electrodialysis of lipopoly-saccharides and their conversion to uniform salt forms. Eur. J.Biochem. 54:603-610.

10. Galanos, C., O. Luderitz, and O. Westphal. 1969. A new methodfor the extraction of R lipopolysaccharides. Eur. J. Biochem.9:245-249.

11. Galanos, C., O. Luderitz, and O. Westphal. 1971. Preparationand properties of antisera against the lipid A component ofbacterial lipopolysaccharides. Eur. J. Biochem. 24:116-122.

12. Granstrom, M., A. Ericsson, B. Strandvik, B. Wretlind, O. R.Pavlovskis, R. Berka, and M. L. Vasil. 1984. Relation betweenantibody response to Pseudomonas aeruginosa exoproteins andcolonization/infection in patients with cystic fibrosis. Acta Pae-diatr. Scand. 73:772-777.

13. Hammerstrom, C., R. A. Insel, M. A. A. Persson, and C. I. E.Smith. 1988. The IgG 1 subclass preference of selected naturallyoccurring antipolysaccharide antibodies. Monogr. Allergy 23:18-26.

14. Hancock, R. E., L. M. Mutharia, L. Chan, R. P. Darveau, D. P.Speert, and G. B. Pier. 1983. Pseudomonas aeruginosa isolatesfrom patients with cystic fibrosis: a class of serum-sensitive,nontypable strains deficient in lipopolysaccharide O side chains.Infect. Immun. 42:170-177.

15. H0iby, N. 1977. Pseudomonas aeruginosa infection in cysticfibrosis. Diagnostic and prognostic significance ofPseudomonasaeruginosa precipitins determined by means of crossed-immu-noelectrophoresis. A survey. Acta Pathol. Microbiol. Scand.Sect. C Suppl. 262:3-96.

16. H0iby, N., J. B. Hertz, and D. Sompolinsky. 1980. Antibodyresponse in patients with Pseudomonas aeruginosa infection toa "common antigen" from P. aeruginosa analysed by means ofquantitative immunoelectrophoretic methods. Acta Pathol. Mi-crobiol. Scand. Sect. C 88:149-154.

17. Laurel, C.-B. 1966. Quantitative estimation of proteins byelectrophoresis in agarose gel containing antibodies. Anal. Bio-chem. 15:45-52.

18. Moss, R. B., Y.-P. Hsu, M. M. Sullivan, and N. J. Lewiston.1986. Altered antibody isotype in cystic fibrosis: possible role in

J. CLIN. MICROBIOL.

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from

DIAGNOSIS OF CHRONIC P. AERUGINOSA INFECTION IN CF

opsonic deficiency. Pediatr. Res. 20:453-459.19. Pedersen, S. S., F. Espersen, and N. H0iby. 1987. Diagnosis of

chronic Pseudomonas aeruginosa infection in cystic fibrosis byenzyme-linked immunosorbent assay. J. Clin. Microbiol. 25:1830-1836.

20. Pedersen, S. S., T. Jensen, N. H0iby, C. Kock, and C. W.Flensborg. 1987. Management of Pseudomonas aeruginosa lunginfection in Danish cystic fibrosis patients. Acta Paediatr.Scand. 75:840-845.

21. Pitt, T. L., H. C. Todd, C. A. Mackintosh, and S. W. K. Im.

1985. Evaluation of three serological tests for detection ofantibody to Pseudomonas aeruginosa in human sera. Eur. J.Clin. Microbiol. 4:190-196.

22. Rietschel, E. T., C. Galanos, O. Luderitz, and O. Westphal.1982. Chemical structure, physiological function and biologicalactivity of lipopolysaccharides and their lipid A component, p.

183-229. In D. R. Webb (ed.), Immunopharmacology and theregulation of leukocyte function. Marcel Dekker, Inc., NewYork.

23. Shand, G. H., S. S. Pedersen, R. Tilling, M. R. W. Brown, and

N. H0iby. 1988. Use of immunoblot detection of serum antibod-ies in the diagnosis of chronic Pseudomonas aeruginosa lunginfection in cystic fibrosis. J. Med. Microbiol. 27:169-178.

24. Shryock, T. R., J. S. Mollé, J. D. Klinger, and M. J. Thomassen.1986. Association with phagocytic inhibition of anti-Pseudo-monas aeruginosa immunoglobulin G antibody subclass levelsin serum from patients with cystic fibrosis. J. Clin. Microbiol.23:513-516.

25. Siber, G. R., P. H. Schur, A. C. Aisenberg, S. A. Weitzman, andG. Schiffman. 1980. Correlation between serum IgG-2 concen-

trations and the antibody response to bacterial polysaccharideantigens. N. Engl. J. Med. 303:178-182.

26. Tsai, C. M., and C. E. Frash. 1982. A sensitive silver stain fordetecting lipopolysaccharides in polyacrylamide gels. Anal.Biochem. 119:115-119.

27. Weinstein, M. C., and H. V. Fineberg. 1980. Clinical decisionanalysis, p. 114-120. The W. B. Saunders Co., Philadelphia.

28. Wood, R. E., T. F. Boat, and C. F. Doershuk. 1976. State of theart: cystic fibrosis. Am. Rev. Respir. Dis. 113:833-878.

VOL. 27, 1989 1229

on June 28, 2018 by guesthttp://jcm

.asm.org/

Dow

nloaded from