Abstract
Objective. Rheumatoid arthritis (RA) is a chronic inflammatory disease accompanied by many complications, and serious infections are associated with many of the advanced therapeutics used to treat it. We assessed serum procalcitonin (PCT) levels to distinguish bacterial infection from other complications in patients with RA.
Methods. One hundred eighteen patients experiencing an RA flare, noninfectious complication of RA or its treatment, nonbacterial infection, or bacterial infection were studied. Serum PCT concentrations were determined with a chemiluminescent enzyme immunoassay.
Results. All patients experiencing an RA flare showed negative PCT levels (≤ 0.1 ng/ml; n = 18). The PCT level was higher in the bacterial infection group (25.8% had levels ≥ 0.5 ng/ml) than in the other 3 groups (0.0–4.3% had levels ≥ 0.5 ng/ml) and the difference was significant among groups (p = 0.003). Conversely, no statistically significant difference was observed among the groups with C-reactive protein (CRP) concentration ≥ 0.3 mg/dl (p = 0.513), white blood cell (WBC) count > 8500/mm3 (p = 0.053), or erythrocyte sedimentation rate (ESR) > 15 mm/h (p = 0.328). The OR of high PCT level (≥ 0.5 ng/ml) for detection of bacterial infection was 19.13 (95% CI 2.44–149.78, p = 0.005). Specificity and positive likelihood ratio of PCT ≥ 0.5 ng/ml were highest (98.2% and 14.33, respectively) for detection of bacterial infection, although the sensitivity was low (25.8%).
Conclusion. Serum PCT level is a more specific marker for detection of bacterial infection than either CRP, ESR, or WBC count in patients with RA. High PCT levels (≥ 0.5 ng/ml) strongly suggest bacterial infection. However, PCT < 0.5 ng/ml, even if < 0.2 ng/ml, does not rule out bacterial infection and physicians should treat appropriately.
Rheumatoid arthritis (RA) is an inflammatory disease that mainly affects synovial tissue. Several inflammatory cytokines, such as tumor necrosis factor (TNF), interleukin 1 (IL-1), and IL-6, are associated with RA, and cytokine blockade drugs have recently become an important treatment option1,2. The prognosis of RA has improved considerably because of developments in drug therapy, but susceptibility to infections remains a major problem. Distinguishing infection from inflammation or other noninfectious complications is important in clinical practice. Identification of severe bacterial infections is a major concern because it influences whether antibiotics will be prescribed. For subjects with systemic inflammation caused by bacterial infections, immediate administration of antibiotics should be considered.
Procalcitonin (PCT) is a 13-kDa precursor protein of calcitonin composed of 116 amino acids3. PCT is usually produced and cleaved to calcitonin in the C cells of the thyroid gland and is not detectable in normal control serum4. When bacterial infection accompanied by systemic inflammation occurs, PCT is produced in the liver, lung, kidney, adipocyte, and muscle, and levels in serum increase5. The cytokines IL-1ß and TNF then elevate levels of PCT5.
PCT levels in patients with arthritis are useful to discriminate bacterial arthritis and noninfectious arthritis such as RA, reactive arthritis, and crystal arthritis6,7,8. However, the efficacy of using PCT levels to identify bacterial complications in patients with RA has not been well investigated. Our study clarified whether serum PCT levels on admission can predict the final diagnosis of bacterial infection assessed using all clinical findings by physicians who were blind to the PCT levels in patients with RA.
MATERIALS AND METHODS
Subjects
The study subjects were 118 consecutive patients with RA admitted to the Niigata Rheumatic Centre between February 2008 and March 2009 because of either RA flare or infectious/noninfectious complications. Patients were excluded if they had RA without complications or flares. The study patients were divided into 4 groups: those with RA flare, noninfectious complication of RA or its treatment, RA with nonbacterial infection, or RA with bacterial infection. Bacterial infections included general bacterial and mycobacterium infections. Nonbacterial infections were viral or fungal infections including pneumocystis pneumonia. The “gold standard” for diagnosis of bacterial infection was final diagnosis based on the symptoms, bacterial culture tests, imaging studies, and response to antibiotic therapy. Because bacterial cultures are not always positive in all bacterial infections and false positives (colonization) can occur, physicians should comprehensively make the final diagnosis using all clinical findings. Our study aimed to reveal that PCT levels at admission can predict the final diagnosis of bacterial infection. Thus the final diagnosis in this study took into consideration all clinical features observed throughout hospitalization, by physicians who were blind to the PCT levels. All blood samples and cultures were obtained before antibiotic therapy was initiated and blood samples were stored at −80°C. The serum PCT concentration was measured after the final diagnosis had been made by each physician. Each patient met the 1987 American Rheumatism Association criteria for RA9. Anatomical stage and functional class were determined using the system proposed by Steinbrocker, et al10.
The study protocol was approved by the ethics committee of Niigata Rheumatic Centre and executed according to the Declaration of Helsinki. Written informed consent was obtained from all participants.
Biochemical assays of PCT
The PCT concentration was determined using serum samples collected before antibiotic therapy and stored at −80°C by a chemiluminescent enzyme immunoassay (CLEIA; SphereLight PCT®; Wako Pure Chemical Industries, Osaka, Japan). For the study, 2 cutoff values were determined to identify bacterial infections, ≥ 0.2 ng/ml and ≥ 0.5 ng/ml, because the minimum value was ≤ 0.1 ng/ml and 0.2 ng/ml was the first positive value of PCT, and the recommended cutoff value for bacterial sepsis was ≥ 0.5 ng/ml11.
Other inflammatory markers
We used C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and white blood cell (WBC) count as inflammatory markers. Cutoff criteria were determined as ≥ 0.3 mg/dl for CRP, > 15 mm/h for ESR, and > 8500 /mm3 for WBC count, which are the values considered abnormal at our clinic.
Statistical analysis
To compare baseline characteristics among the 4 groups — RA flare, noninfectious complication of RA or its treatment, RA with nonbacterial infection, and RA with bacterial infection — 1-way analysis of variance was used for continuous variables and chi-square test was applied for categorical variables. Distributions of inflammatory markers were described using the median and interquartile range, and the difference between groups was assessed by the Kruskal-Wallis test. Prevalence of abnormal inflammatory markers was compared using a chi-square test. For prevalence, 95% CI were calculated based on binominal distribution. To assess the strength of association between abnormal values of inflammatory markers and bacterial infection, OR and CI were calculated using unadjusted logistic regression analysis. The screening potential of these cutoff criteria for bacterial infection was compared using sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios. Statistical analyses were performed with SPSS v19 (SPSS Inc., Chicago, IL, USA). The 95% CI for prevalence were calculated using Microsoft Excel 2007 (Microsoft Inc., Redmond, WA, USA). P < 0.05 was considered statistically significant.
RESULTS
Characteristics of study subjects
The characteristics of subjects are listed in Table 1. Age, sex, body temperature, hemoglobin level, platelet count, and use of drugs for RA were not significantly different among the 4 study groups.
Characteristics of the study subjects. Data are mean ± SD or number (%).
Distribution of inflammatory markers and positive prevalence
Distributions of PCT levels and WBC counts were significantly different among the 4 study groups and the median value was higher in the bacterial infection group than the other 3 groups. However, median values of CRP or ESR were not (Table 2). No patient with RA flare showed PCT ≥ 0.2 ng/ml and only 1 patient with noninfectious complications showed PCT ≥ 0.5 ng/ml. Using standard cutoff criteria, the prevalence of PCT ≥ 0.5 and ≥ 0.2 ng/ml was significantly different between groups (p = 0.003, p = 0.013, respectively) and was most frequent in the bacterial infection group. However, the prevalence of abnormal CRP, WBC count, and ESR was not significantly different (p = 0.513, p = 0.053, p = 0.328, respectively; Table 2).
Distribution of inflammatory markers and positive prevalence.
Diagnosis of complications in patients with RA
Table 3 compares diagnosis of complications and PCT levels. Among patients with RA who had complications other than bacterial infection, only 1, with exacerbation of interstitial pneumonia associated with RA (RA lung), showed PCT ≥ 0.5 ng/ml (0.7 ng/ml). Slightly elevated PCT levels (≥ 0.2 ng/ml) were observed in all of 3 patients with secondary amyloidosis associated with RA (0.2, 0.2, and 0.3 ng/ml, respectively), and in those with meningitis due to Cryptococcus and panperitonitis due to Candida albicans (0.2 and 0.3 ng/ml, respectively). Although the prevalence of elevated PCT was high in RA complicated by bacterial infection, some patients with general bacterial pneumonia (21/27), urinary tract infections (3/7), acute bronchitis (2/3), infectious colitis (3/4), infectious arthritis (1/2), and wound infection (3/4), and all with cellulitis (6/6) and nontuberculous mycobacterial pneumonia (2/2) did not satisfy even the low threshold of elevated PCT (≥ 0.2 ng/ml).
Diagnosis of complications in patients with rheumatoid arthritis (RA).
Of 62 patients with bacterial infection, 22 had positive cultures. Of the remaining 40 patients, 27 cases were diagnosed based on typical image findings (patients with pneumonia, acute necrotizing pancreatitis, or acute cholangitis, and some patients with wound infections), and the other 13 patients were diagnosed based on clinical manifestations and response to antibiotic therapy (patients with cellulitis or febrile neutropenia, some patients with infectious colitis or urinary tract infection, etc.). Only 7 (31.8%) of the 22 patients with positive cultures had PCT ≥ 0.5 ng/ml (none showed PCT level between 0.2 and 0.5 ng/ml), and 2 of those had a positive result on blood culture; 1 was urinary tract infection with Mycoplasma hominis (PCT 1.3 ng/ml) and the other was infectious aneurysm with Staphylococcus aureus (PCT 1.8 ng/ml). No significant correlation was observed between PCT and CRP levels in patients with bacterial infection whose PCT levels were ≥ 0.2 ng/ml (Spearman’s rho = 0.204, p = 0.375; n = 21).
Among patients with bacterial infections, 10 used biologics (TNF inhibitor in 7 patients and anti-IL-6 receptor antagonist in 3). Positive rates of PCT (≥ 0.2 and ≥ 0.5 ng/ml) did not differ between patients using biologic therapies and all patients with bacterial infections [4/10 (40.0%) vs 17/52 (32.7%) and 2/10 (20.0%) vs 14/52 (26.9%), respectively]. For patients treated with IL-6 receptor antagonist, the positive rate of PCT (≥ 0.2 and ≥ 0.5 ng/ml) was not low — 2/3 (60.0%) and 1/3 (33.3%), respectively.
Screening potential for detecting bacterial infections
The screening potential of abnormal inflammatory markers for identifying bacterial infections among patients with RA was assessed (Table 4). PCT levels ≥ 0.2 and ≥ 0.5 ng/ml were significantly associated with bacterial infection, and the OR for PCT ≥ 0.5 ng/ml was very high (OR 19.13, 95% CI 2.44–149.78). Using PCT ≥ 0.5 ng/ml, although the sensitivity was low (25.8%), the specificity (98.2%), positive predictive value (94.1), and positive likelihood ratio (14.33) were higher than those of other abnormal inflammatory markers: only 1 of 17 patients (5.9%) who had RA without bacterial infection showed a false-positive. The sensitivity remained low (33.9%) even when the lower value, ≥ 0.2 ng/ml, was used as the positive determinant of elevated PCT. WBC count > 8500/mm3 was also significantly associated with bacterial infection (OR 2.80, 95% CI 1.33–5.92). However, WBC count > 8500/mm3 did not identify 3 of 16 patients with bacterial infection having PCT ≥ 0.5 ng/ml. Further, WBC count > 8500/mm3 was more frequently observed in cases of RA flare, complications due to RA or RA treatment, and nonbacterial infections than was PCT ≥ 0.5 ng/ml. Thus, the specificity and positive likelihood ratio of the WBC count > 8500/mm3 (58.9% and 1.61, respectively) were far lower than those of the PCT ≥ 0.5 ng/ml, showing that PCT ≥ 0.5 ng/ml is a much more specific marker of bacterial infection than WBC count > 8500/mm3. Also, combining PCT ≥ 0.5 ng/ml and WBC count > 8500/mm3 did not improve any screening indices for bacterial infection compared with using only PCT ≥ 0.5 ng/ml.
The screening potential of serum procalcitonin, C-reactive protein, white blood cell count, and erythrocyte sedimentation rate for detecting bacterial infections and strictly bacterial infections in patients with rheumatoid arthritis. Bacterial infection was finally diagnosed based on symptoms, bacterial culture tests, imaging studies, and response to antibiotic therapy. Strictly bacterial infection was diagnosed based on bacterial culture tests and/or imaging studies. OR, 95% CI, and p values were calculated by nonadjusted logistic regression analysis.
When bacterial infection was defined strictly as diagnosed by positive cultures and/or typical image findings (n = 49, 41.5%), PCT level ≥ 0.5 ng/ml and WBC count > 8500/mm3 were significantly associated with strictly bacterial infection (OR 4.15, 95% CI 1.36–12.71, p = 0.013, and OR 2.18, 95% CI 1.02–4.63, p = 0.043, respectively). Five patients with PCT ≥ 0.5 ng/ml were false-positive and a positive likelihood ratio of PCT ≥ 0.5 ng/ml decreased lower (3.38) than that for detection of bacterial infection (14.33); however, PCT ≥ 0.5 ng/ml had the highest OR, specificity, and positive likelihood ratio, compared with other inflammation markers (Table 4).
DISCUSSION
The results of our study indicate clearly that the serum PCT level is a more specific marker for detection of bacterial infection than other inflammatory markers, such as CRP, ESR, or WBC count, in patients with RA. PCT level ≥ 0.5 ng/ml showed a very high OR for detection of bacterial infection and also showed few false positives. Thus, PCT ≥ 0.5 ng/ml strongly suggests the presence of bacterial infections and indicates requirement for immediate antibiotic therapy.
PCT has been described as a marker of systemic infection such as sepsis and septic shock11,12. In our study, elevated PCT was observed in patients with cholecystitis, acute cholangitis, acute necrotizing pancreatitis, febrile neutropenia, infectious endocarditis, or infectious aneurysm. These severe bacterial infections should be immediately treated with antibiotics, thus measurement of PCT levels is meaningful in patients with RA who have signs suggesting infectious complications.
No patient with RA flare in our study showed elevated PCT levels, similar to previous reports6,7,8. Several studies have reported elevated PCT without infection in patients with autoimmune diseases13,14,15,16,17,18,19,20,21 such as adult-onset Still’s disease18, granulomatosis with polyangiitis (Wegener granulomatosis)15,20, and Goodpasture syndrome16. These autoimmune diseases often cause a sepsis-like high fever. Massive production of inflammatory cytokines such as TNF and IL-6 in such sepsis-like conditions22,23,24,25,26 may stimulate the production of PCT. Although elevation of these serum cytokines has also been reported in patients with RA27,28, this may be unusual because sepsis-like high fever is rare in chronic inflammatory diseases such as RA. In our study, PCT remained elevated despite treatment with biologics. However, as the number of patients using biologics was small and the types of bacterial infections differed, further study is needed to investigate the possible effects of biologics on PCT.
Christ-Crain, et al have proposed PCT-guided use of antibiotic agents in patients with lower respiratory tract infections29. According to their stratification criteria, antibiotic therapies are discouraged when PCT levels are < 0.25 ng/ml. Using this therapeutic guideline, they were able to reduce use of antibiotics without worsening prognosis. However, their guidelines should not be applied to patients with RA. In our study, many cases of bacterial infection were found even when serum PCT levels were < 0.2 ng/ml. Regarding general bacterial pneumonia, 21 of 27 cases had serum PCT levels < 0.2 ng/ml. Normal PCT levels have frequently been observed in patients with RA who had bacterial infections19. Further, our findings together with some previous reports19,30 suggest that PCT levels are not elevated in patients with local infections, such as wound infection or cellulitis. Because patients with RA are usually immunocompromised as a result of their RA treatment, immediate treatment of bacterial infection is of critical importance. Therefore, low serum PCT levels should not prevent the use of antibiotic therapy when signs of bacterial infection are present.
In our study, WBC counts > 8500/mm3 were also significantly associated with bacterial infection. This association was less specific than PCT ≥ 0.2 ng/ml. Therefore, bacterial infections should be carefully considered in patients with WBC > 8500/mm3 despite a normal PCT finding. However, antibiotic therapy should be initiated in any patient with PCT ≥ 0.5 ng/ml with normal WBC, because the WBC count is not always > 8500/mm3 in bacterial infection.
We note that patients with RA amyloidosis showed slightly elevated PCT levels (0.2 to < 0.5 ng/ml). All had severe diarrhea due to secondary amyloidosis associated with RA. In such patients, latent infection through the damaged intestinal mucosa might be present31. Amyloidosis itself also causes severe inflammation32. When these infectious and noninfectious inflammations exist simultaneously, production of PCT may result. Thus, the slightly elevated PCT in patients with RA with diarrhea might be useful as a marker of secondary amyloidosis associated with RA. Larger studies are needed to elucidate this hypothesis.
We used the CLEIA method to determine PCT levels, which is a quantitative analysis. Recently, a semiquantitative method for PCT has been developed. Since semiquantitative methods require neither specialized instrumentation nor techniques, use of this method has become commonplace. However, rheumatoid factor and human anti-mouse antibodies can affect the results of this semiquantitative method33,34; thus we selected a quantitative method for our study. A semiquantitative method for PCT should be interpreted carefully, especially in patients with RA.
The gold standard for bacterial infection in our study was a clinical diagnosis based on symptoms, bacterial cultures, image studies, and response to antibiotic therapy by physicians blind to the PCT levels. Theoretically, the gold standard should have been positive culture testing, but in this study, only 22/62 patients with bacterial infection had a positive culture result; this is a major limitation of our study. However, not all bacterial species can be detected on cultures, and the results should be interpreted carefully because of the possibility for false positives. Further, when diagnosis of bacterial infection was defined strictly by positive cultures and/or typical image findings, PCT ≥ 0.5 ng/ml was most significantly associated with bacterial infection, as compared to other inflammatory markers. The main purpose of our study was to minimize the interval from admission of patients with complicated RA to confirmation of the diagnosis of bacterial infection in order to start appropriate antibiotic therapy as early as possible. For this purpose, our gold standard would be reasonable.
Another limitation is that patients with various complications were included in order to determine that elevated PCT is rare in nonbacterial complications. Because study subjects must be chosen at random, we included consecutive patients admitted as a result of complicated RA. Consequently, patients with localized bacterial infections were recruited, although they were not expected to have elevated PCT levels. Nevertheless, elevated PCT was observed to be strongly associated with bacterial infection. Further studies are needed to clarify whether PCT can differentiate bacterial infections in specific situations: for example, for differential diagnosis of bacterial pneumonia, nonbacterial infectious pneumonia, RA lung, and drug-induced pneumonia.
A serum PCT level ≥ 0.5 ng/ml strongly suggests the presence of bacterial infections in patients hospitalized because of complicated RA, and immediate antibiotic therapy is recommended for such patients showing any signs of infection. However, physicians should be aware that a PCT finding < 0.2 ng/ml does not rule out the presence of bacterial infection.
Footnotes
-
Supported by a grant from the Ministry of Health, Labor and Welfare, Research Committee for Early Diagnosis to Prevent Severe Complications of Rheumatoid Arthritis in Japan.
- Accepted for publication April 26, 2012.
