Identification rate of Legionella species in non-purulent sputum culture is comparable to that in purulent sputum culture in Legionella pneumonia

ABSTRACT Many Legionella pneumonia patients do not produce sputum, and it is unknown whether purulent sputum is required for the identification of Legionella species. This study aimed to evaluate the identification rate of Legionella species based on sputum quality and the factors predictive of Legionella infection. This study included Legionella pneumonia patients at Kurashiki Central Hospital from November 2000 to December 2022. Sputum quality, based on gram staining, was classified as the following: Geckler 1/2, 3/6 and 4/5. Geckler 4/5 was defined as purulent sputum. The sputa of 104 of 124 Legionella pneumonia patients were cultured. Fifty-four patients (51.9%) were identified with Legionella species, most of which were Legionella pneumophila serogroup 1 (81.5%). The identification rates of Legionella species according to sputum quality were 57.1% (16/28) in Geckler 1/2 sputum, 50.0% (34/68) in Geckler 3/6 sputum, and 50.0% (4/8) in Geckler 4/5 sputum, which were not significantly different (P = 0.86). On multivariate analysis, pre-culture treatment with anti-Legionella antimicrobials (odds ratio [OR] 0.26, 95% confidence interval [CI] 0.06–0.91), Pneumonia Severity Index class ≥IV (OR 2.57 [95% CI 1.02–6.71]), and intensive care unit admission (OR 3.08, 95% CI 1.06–10.09) correlated with the ability to identify Legionella species, but sputum quality did not (OR 0.88, 95% CI 0.17–4.41). The identification rate of Legionella species in non-purulent sputum was similar to that in purulent sputum. For the diagnosis of Legionella pneumonia, sputum should be collected before administering anti-Legionella antibiotics and cultured regardless of sputum quality.

sputum, is needed for detecting all SGs of L. pneumophila, including SG1 and other Legionella species in cases with both positive and negative UAT results.
Although culture of Legionella species from a respiratory specimen is the diagnostic gold standard (15), it is known to have certain drawbacks, including the fact that it needs a specific culture medium, such as buffered charcoal yeast extract-α (BCYE-α) and Wadowsky-Yee-Okuda-α (WYO-α), and takes about 3-5 days to identify the Legionella species (10).In addition, previous reports showed that 50% to 70% of Legionella pneumonia patients do not produce sputum (16)(17)(18).We have also often experienced the inability to collect purulent sputum from Legionella pneumonia patients.For identification of the causative microorganisms in sputum culture, appropriate collection and culture of purulent sputum is essential, as far as possible.However, since there is much variability in the production of sputum as a symptom and the quality of sputum, whether purulent, in Legionella pneumonia cases, the detection power of respective sputum purulence in Legionella pneumonia is unknown.
The aim of the present study was to evaluate the occurrence of sputum purulence and detection ability of Legionella species based on sputum quality, in addition to factors predictive of the identification of Legionella species in Legionella pneumonia patients.

Study population
Legionella pneumonia patients diagnosed at Kurashiki Central Hospital, a 1,166-bed tertiary hospital, from November 2000 to December 2022 were retrospectively enrol led in this study.Patients who had abnormal shadows on radiologic examinations, along with at least one symptom, such as fever, cough, sputum, chest pain, and general malaise, and one clinical finding (abnormal auscultation findings or an increased inflammatory reaction) were diagnosed with pneumonia (19).Patients who were <15 years old and those with hospital-acquired pneumonia were excluded.This study was approved by the institutional review board of our hospital (IRB number 4168).The IRB waived the need for obtaining patient informed consent due to the retrospective nature of the study.

Study design
We investigated the clinical characteristics of Legionella pneumonia patients, including age, sex, comorbidities, symptoms, pre-culture antibiotic treatment before admission, and severity of pneumonia, including CURB-65 score [confusion, urea >7 mmol/L, respiratory rate ≥30 breaths per minute, low blood pressure (systolic <90 mmHg or diastolic ≤60 mmHg), and age ≥65 years] (20), Pneumonia Severity Index (PSI) [calculated using the total score of age, sex, resident in nursing home or not, five comorbidities, five physical examination findings, six laboratory findings, and one radiological finding] (21), A-DROP score [age ≥70 y in men and ≥75 y in women, dehydration or blood urea nitrogen ≥21 mg/dl, SpO 2 ≤90%, disturbance in orientation, and systolic blood pressure ≤90 mmHg] (22), and prognosis, in addition to sputum quality and culture results.Legionella pneumonia was diagnosed in patients who satisfied at least one of the following criteria: positive results of Legionella UATs, identification of Legionella species in lower respiratory tract sputum samples, positive results of Legionella gene tests, or 4-fold increase in paired serum antibody testing.In clinical practice at our hospital, Binax NOW Legionella (Abbott Diagnostics Medical, Lake Forest, CA, USA) (Binax) was used from December 2004 to July 2016, and Immunocatch Legionella (Eiken Kagaku Corporation, Tokyo, Japan) has been used since July 2016 for the diagnosis of Legionella infection.Since June 2019, Libotest Legionella (Kyokutou Corporation, Tokyo, Japan) (Libotest) has also been used in addition to Immunocatch Legionella.
In daily clinical practice, sputum culture using WYO-α, gene testing by loop-mediated isothermal amplification (LAMP) (Eiken Kagaku Corporation, Tokyo, Japan) of sputum, and serum antibody tests were performed at the discretion of the attending physicians.

Study outcomes
The primary outcome was the identification rate of Legionella species in each sputum quality class according to Geckler's classification.The secondary outcome was the identification rate of Legionella species in terms of pneumonia severity, based on CURB-65, PSI, and A-DROP scores.CURB-65 ≥3 points, PSI ≥ class IV, and A-DROP ≥3 points were defined as severe pneumonia (20)(21)(22).In addition, we evaluated factors predictive of the identification of Legionella species on sputum culture.

Statistical analysis
Continuous variables are expressed as medians and interquartile ranges, and cate gorical variables are expressed as numbers and percentages.Categorical variables were analyzed by Fisher's exact test, and continuous variables by the non-parametric Mann-Whitney U-test.To evaluate the identification ability according to sputum quality, we divided sputum into the following three quality groups: Geckler 1/2, 3/6, and 4/5.Univariate analysis was performed to identify predictive factors in the identification of Legionella species on sputum culture.Multivariate analysis using stepwise logistic regression analysis was conducted for all variables that were found to have a P value of ≤ 0.05 on univariate analysis, in addition to sputum quality and pre-culture treatment using anti-Legionella antibiotics.All tests were two-tailed, and a P value of < 0.05 was considered significant.All statistical analyses were performed using EZR statistical software (version 3.0.3,Vienna, Austria) (25).

Patients' baseline characteristics
A total of 124 Legionella pneumonia patients, including 119 hospitalized patients and 5 outpatients, were evaluated.Figure 1 shows the results of Legionella UAT, culture, sputum gene analyses by LAMP, and serum antibodies.Among the 124 Legionella pneumonia patients, sputum for culture was obtained from 104 patients (83.9%).The sputum culture positivity rate was 51.9% (54/104), and 14 among the 18 patients with negative UAT results were diagnosed by sputum culture (Fig. 1).The baseline clinical characteristics of Legionella pneumonia patients with and without the identification of Legionella species by sputum culture are shown in Table 1.Median patient age was 68 years, and 87.5% were male.The most common comorbidity was diabetes mellitus, followed by chronic heart disease.The most common symptom was fever (89.4%), followed by cough (39.4%) and relative bradycardia (39.4%).Forty-six patients (53.5%) without sputum as a symptom were identified with Legionella by sputum culture.LAMP was performed in 10 of the 104 patients, among whom 7 patients (70%) showed positive results.Regarding the positivity rate following LAMP in terms of sputum quality, the LAMP positive rate was 100% (1/1) in purulent sputum and 66.7% (6/9) in non-purulent sputum.

Distribution of Legionella species and Legionella pneumophila serogroups identified by sputum culture
The Legionella species identified by sputum culture is shown in Fig. 2. The most common Legionella species was L. pneumophila SG1 (81.5%), followed by L. pneumophila SG3 (11.1%).Non-L.pneumophila SG1 was identified in 18.5% of Legionella pneumonia patients.

Ability to identify Legionella species by sputum culture according to pneumo nia severity
Table 2 shows the identification rate of Legionella species in non-severe and severe pneumonia groups according to CURB-65, PSI, and A-DROP scores.There was a significant difference in the ability to identify Legionella according to the severity of pneumonia determined using PSI, but not with CURB-65 and A-DROP scores.
Regarding the correlation between the identification rate of Legionella species and interval between sputum collection and administration of anti-Legionella antimicrobials,  the identification rate was not inferior in patients in whom sputum was sampled within 24 h after administering anti-Legionella antibiotics compared with patients who did not receive antibiotic therapy for Legionella pneumonia [75.0%(6/8) vs 54.2% (47/87)].However, the identification rate of Legionella species decreased steeply and was significantly lower if the sputum was collected more than 24 h after administer ing anti-Legionella antibiotics, compared with patients without pre-treatment and with pre-treatment by anti-Legionella antibiotics within 24 h (55.8% vs 11.1%, p = 0.04) (Fig. 4).

DISCUSSION
The present study showed that most Legionella pneumonia patients had non-purulent sputum, and the identification rate of Legionella species by sputum culture did not differ significantly between purulent and non-purulent sputa.Regarding the correla tion between the identification rate of Legionella species and pneumonia severity, the identification rate was higher in severe pneumonia patients, defined by PSI, than in non-severe patients.In addition, antibiotic treatment for Legionella pneumonia before sputum sampling was a negative predictor, and severe pneumonia defined by PSI and ICU admission within 24 h after admission were positive predictors in the identification of Legionella species by sputum culture.
Many Legionella pneumonia patients have non-purulent sputum or sputum cannot be obtained for culture (4).Therefore, most attending physicians usually use UAT kits for the diagnosis of Legionellosis in Japan and many other countries (10)(11)(12)(13).The main disadvantage of the existing UAT kits is their inability to diagnose Legionella pneumonia due to non-Legionella pneumophila serogroup 1 (26).In February 2019, Libotest Legionella (Kyokutou Corporation, Tokyo, Japan), which can detect all Legionella pneumophila serogroups, was launched in Japan.We previously reported that Libotest Legionella had similar diagnostic ability compared with existing UAT kits, including BinaxNOW and Q line, in addition to the ability to detect all serogroups of Legionella pneumophila (18).However, since the sensitivity of Libotest Legionella for diagnosing Legionellosis due to non-L.pneumophila SG1 is unknown and there is a substantial need for surveillance of Legionella species in Legionella pneumonia patients, performance of sputum culture for the diagnosis of Legionella pneumonia will continue to be important.In the present study, sputum culture was performed in 83.9% of Legionella pneumonia patients, and Legionella species were identified in almost half of them, although only 17.5% of the patients had sputum as a symptom.In addition, the identification rate of Legionella species was not significantly different between purulent and non-purulent sputa (50.0%vs 52.1%, p = 1.00).Cunha et al. reported that the sputum of Legionella pneumonia patients had less neutrophils and was watery (10).According to these studies, including that of our study, efforts should be made to collect sputum for the diagnosis of Legionella pneumonia even in patients without sputum as a symptom.Two previous studies have evaluated the correlation between sputum quality and the culture results of Legionella species in Legionella pneumonia patients (27,28).Ingram et al. investigated the correlation between sputum quality and culture results in 19 patients with Legionella pneumonia due to L. pneumophila (27).They reported that L. pneumophila was identified the most in Geckler six sputum (n = 7), followed by Geckler one sputum (n = 4), and most patients (78.9%) had non-purulent sputum (Geckler 1, 2, 3, and 6) (27).Another study by Shakeshaft et al. also showed that 46 of the 72 culture-positive Legionella pneumonia patients (63.9%) had non-purulent sputum, as defined by Murray and Washington criteria (28).The authors of these two studies concluded that all sputum samples, including non-purulent sputum, should be submitted for Legionella culture when Legionella pneumonia is suspected.The results of their studies were very significant for diagnosing Legionella pneumonia, although there were some possible limitations.First, only patients with identification of Legionella species by sputum culture were included in both studies.Therefore, the distribution of sputum quality and identification rate of Legionella species in each sputum quality group of Legionella pneumonia patients were unknown.Second, Ingram's study included a relatively small number of patients (n = 19) (27), whereas most of the 72 culture-positive Legionella pneumonia patients included in Shakeshaft's study were due to L. longbeachae (65.3%) (28).Therefore, our study is the first to evaluate the distribution of sputum quality and identification rate of Legionella species, including L. pneumophila and other species, according to sputum quality in over a hundred Legionella pneumonia patients, including more than 90% of L. pneumophila cases.
Regarding the correlation between pneumonia severity and the identification rate of Legionella species, our study showed that the identification rate was significantly higher in severe cases than in non-severe cases, as defined by PSI.There are several possible reasons for this observation.First, a previous report indicated that the sensitivity of Legionella UATs is higher in severe disease cases than in non-severe cases (29).According to that study, severe patients might have a greater load of Legionella species in sputum compared with urine.Second, it is possible that the attending physicians made a greater effort to collect sputum in severe pneumonia patients.However, in the present study, the collection rate of sputum was high (83.9%),and the collection rate of sputum was almost the same in PSI I-III and IV-V cases (75.0%vs 88.8%, data not shown).Therefore, sputum collection bias is not likely to have affected the results.
We also reported that administration of anti-Legionella antimicrobials before sputum evaluation was a negative predictor of the identification of Legionella species, and severe pneumonia, defined as PSI class ≥IV and ICU admission within 24 h after admission, was a positive predictive factor in the identification of Legionella species by sputum culture.A previous report showed that the sensitivity of sputum culture in diagnosing pneumo coccal pneumonia decreased after antibiotic administration (93% without pre-culture antibiotics vs 74% with pre-culture antibiotics) (30).Another study by Mentasti et al. reported that the identification rate of Legionella species by sputum culture was higher in sputum collected within 2 days after admission compared with sputum collected more than 2 days after admission (79.6% vs 47.8%) (31).According to these reports, pre-cul ture antibiotic treatment with anti-Legionella antimicrobials might lead to a reduction in the amount of Legionella species observed in sputum.Indeed, our study showed that the identification rate of Legionella species significantly reduced from 24 h after administering anti-Legionella antibiotics.Musher et al. reported that the positive sputum culture rate for S. pneumoniae clearly decreased from 24 h after administering antibiotics compared with the rate within 24 h of antibiotic administration (28.6% in ≥24 h, 88.9% in 6-24 h, 77.8% in <6 h, and 93% without antibiotics) (30).In addition, despite a report of three cases of Legionnaires' disease showing the correlation between positivity in polymerase chain reaction (PCR) tests and antibiotic treatment, Korosec et al. reported that Legionella amplicon intensity was highest in sputum or bronchial aspirates collected at or before the start of antibiotic therapy and decreased markedly within 3 days of antibiotic therapy (32).Therefore, sputum collection for culture should preferably be performed before or within 24 h of anti-Legionella antimicrobial administration, although treatment delay for sputum collection is not recommended.In terms of the correlation between identification rate and pneumonia severity, previous studies reported that the identification rate was higher in severe pneumonia compared with non-severe pneumonia cases, as previously mentioned (29).This is the first study to show that the identification rate was significantly higher in severe pneumonia, including cases requiring ICU admission within 24 h, than in non-severe pneumonia using multivariate analysis.
Regarding the diagnosis of Legionella pneumonia, gene tests, including PCR and LAMP, are essential for diagnosing Legionella pneumonia in patients with negative UAT results in daily clinical practice.Indeed, a previous report showed that a larger num ber of Legionella pneumonia patients could be diagnosed using both LAMP and UAT (33).However, PCR and LAMP could not detect the details of Legionella pneumophila SG and Legionella species.Therefore, performing sputum culture for Legionella species identification is likely to be significant for surveillance in individual areas and countries.
There are some limitations to the present study.First, this study was conducted retrospectively at a single center in Japan.Patients' symptoms, including sputum, might have been underestimated due to its retrospective nature.Second, the attending physicians' efforts in collecting sputum might have affected the diagnosis of Legion ella pneumonia.Legionella pneumonia might have been underdiagnosed because the attending physicians did not collect sputum in pneumonia patients with negative Legionella UAT results because of the absence of sputum production as a symptom.Third, sputum samples were not obtained from 20 patients who were diagnosed with UATs.However, since the percentage of such patients was relatively small (16.1%), it is not likely to have affected the study results.Finally, since the prevalence of Legionella pneumonia and Legionella species varies between areas and countries, similar studies are needed in other countries as well.Additionally, since many Legionella pneumo nia patients might be underdiagnosed due to negative UAT results, we believe that collection and culture of sputum for diagnosing Legionella pneumonia is important even in places with a low prevalence of Legionella.A strength of the present study is that it was relatively large, including 124 Legionella pneumonia patients, among whom sputum was collected from 104 patients.In addition, sputum quality was examined by experienced microbiological test engineers, and this is the first study to evaluate the identification rate of Legionella species in terms of sputum quality.
In conclusion, most Legionella pneumonia patients in this study had non-purulent sputum, and the identification rate of Legionella species was comparable between non-purulent and purulent sputum.These results suggest that regardless of sputum quality, sputum should be collected and cultured within 24 h of the administration of anti-Legionella antimicrobials in cases of suspected Legionella pneumonia.

FIG 1 FIG 2
FIG 1 Study flowchart, the number of positive results, and the results of diagnostic tests, including urinary antigen tests, sputum cultures, gene tests, and serum antibodies, among the 124 Legionella pneumonia patients are shown.LAMP, loop-mediated isothermal amplification; SG, serogroup; UAT, urinary antigen test.

FIG 3
FIG 3 Identification rate of Legionella species according to sputum quality in the 104 Legionella pneumonia patients.Almost all the patients (92.3%) had non-purulent sputum, and a few patients (7.7%) had purulent sputum.The identification rate of

FIG 4
FIG4 Identification rate of Legionella species by sputum culture relative to time after the administration of anti-Legionella antibiotics The identification rate of Legionella species was 54.2% (47/87) in patients without pre-culture antibiotic treatment using anti-Legionella antibiotics, 75.0%(6/8) in patients within 24 h after the administration of anti-Legionella antibiotics, 16.7% (1/6) in patients within 24 to 48 h after the administration of anti-Legionella antibiotics, and 0% (0/3) in patients more than 48 h after the administration of anti-Legionella antibiotics.

TABLE 1
Baseline clinical characteristics of Legionella pneumonia patients with positive and negative sputum culture results b,c (Continued on next page)

TABLE 1
Baseline clinical characteristics of Legionella pneumonia patients with positive and negative sputum culture results b,c (Continued) a Late ICU admission means ICU admission more than 24 h after admission.b A-DROP: age ≥70 years in men or ≥75 years in women, blood urea nitrogen ≥21 mg/dL or dehydration, oxyhemoglobin saturation measured by pulse oximetry ≤90% or partial pressure of oxygen in arterial blood ≤60 mmHg, confusion, and systolic blood pressure ≤90 mmHg; CURB-65: confusion, urea >7 mmol/L, respiratory rate ≥30 breaths/min, low blood pressure (systolic <90 mmHg or diastolic ≤60 mmHg), and age ≥65 y; ICU: intensive care unit; NA: not assessed; PSI: Pneumonia Severity Index.c Data are shown as numbers (%) or medians and interquartile range.

TABLE 2
Identification rate of Legionella species by sputum culture in terms of pneumonia severity a,b

TABLE 3
Predictive factors in the identification of Legionella species by sputum culture a a CI, confidence interval; ICU, intensive care unit; NA, not assessed; PSI, Pneumonia Severity Index.