Dynamics of Peripheral Blood Lymphocyte Subpopulations in the Acute and Subacute Phase of Legionnaires’ Disease

Study Objective Absolute lymphocytopenia is recognised as an important hallmark of the immune response to severe infection and observed in patients with Legionnaires’ disease. To explore the immune response, we studied the dynamics of peripheral blood lymphocyte subpopulations in the acute and subacute phase of LD. Methods and Results EDTA-anticoagulated blood was obtained from eight patients on the day the diagnosis was made through detection of L. pneumophila serogroup 1 antigen in urine. A second blood sample was obtained in the subacute phase. Multiparametric flow cytometry was used to calculate lymphocyte counts and values for B-cells, T-cells, NK cells, CD4+ and CD8+ T-cells. Expression of activation markers was analysed. The values obtained in the subacute phase were compared with an age and gender matched control group. Absolute lymphocyte count (×109/l, median and range) significantly increased from 0.8 (0.4–1.6) in the acute phase to 1.4 (0.8–3.4) in the subacute phase. B-cell count showed no significant change, while T-cell count (×106/l, median and range) significantly increased in the subacute phase (495 (182–1024) versus 979 (507–2708), p = 0.012) as a result of significant increases in both CD4+ and CD8+ T-cell counts (374 (146–629) versus 763 (400–1507), p = 0.012 and 119 (29–328) versus 224 (107–862), p = 0.012). In the subacute phase of LD, significant increases were observed in absolute counts of activated CD4+ T-cells, naïve CD4+ T-cells and memory CD4+ T-cells. In the CD8+ T-cell compartment, activated CD8+ T-cells, naïve CD8+ T-cell and memory CD8+ T-cells were significantly increased (p<0.05). Conclusion The acute phase of LD is characterized by absolute lymphocytopenia, which recovers in the subacute phase with an increase in absolute T-cells and re-emergence of activated CD4+ and CD8+ T cells. These observations are in line with the suggested role for T-cell activation in the immune response to LD.


Introduction
Legionnaires' disease (LD) is a potentially fatal pneumonia mainly caused by Legionella pneumophila serogroup 1. These Gramnegative bacteria cause an intracellular infection of alveolar macrophages and are an important cause of community-and hospital-acquired pneumonia. Infection with Legionella species ranks among the three most common causes of severe community-acquired pneumonia (CAP) [1]. In the Netherlands, around 300 patients are admitted annually to the hospital with CAP due to L. pneumophila [2]. If not asymptomatic, the clinical presentation of LD may be variable, from relatively mild disease to severe pneumonia requiring Intensive Care Unit (ICU) admission with high mortality rates depending, among others, on the presence of multi-organ involvement and effectiveness of initial therapy [3,4,5]. Currently, immunochromatographic assays for rapid qualitative detection of L. pneumophila serogroup 1 antigen in urine specimens and polymerase chain reaction tests for detection of L. pneumophila DNA in serum and respiratory specimens allow for rapid diagnosis of LD [6,7]. Yet, despite these advances in recognition of L. pneumophila infection, its clinical course remains unpredictable [8].
Absolute lymphocytopenia (peripheral blood lymphocyte count below 1.0610 9 /l) is increasingly recognised as an important hallmark of the immune response to various microorganisms [9,10]. This phenomenon has been observed in several forms of CAP, especially in the acute phase, and is probably limited to Tcells and T-cell subpopulations [11]. The severity of pneumonia may be correlated with the degree of lymphocytopenia in the acute stage as is observed in patients with Mycoplasma pneumoniae pneumonia [12,13].
It has been hypothesized that depression of absolute peripheral blood T-cell counts in CAP patients represents shift of these cells towards the lung in order to be sequestered in protective mechanisms [14,15]. In sepsis, proposed mechanisms of lymphocytopenia are margination and redistribution of lymphocytes within the lymphatic system and marked accelerated apoptosis [16,17]. In apoptosis, selected cell populations are actively deleted from certain tissues and in animal sepsis models this has been shown a mechanism of lymphocyte death [18,19,20]. In septic patients, extensive apoptosis was noted in circulating lymphocytes occurring by both death receptor-and mitochondrial-mediated pathways [21].
Among laboratory findings on hospital admission in patients with LD, absolute lymphocytopenia was observed in the majority of patients [22,23]. Furthermore, LD is characterized by accumulation of activated T-cells in the lungs [24]. This could well be a reflection of early recruitment of activated Tcells at the site of the infection. Increased serum levels of the Th1 cytokines interferon-c and interleukin-12 suggest a predominance of the cellular immune response in patients with LD in the resolution of the primary infection [25]. Furthermore, animal models show that control of the infection and clearance of L. pneumophila depend on recruitment and function of CD4 + and CD8 + T-cells [26].
Cell-mediated immunity appears to be the primairy host defence mechanism against Legionella pneumophila infection, although exact immunopathogenesis in LD remains unknown.
To further explore the cellular immune response in patients with LD, we studied the proportions and dynamics of peripheral blood lymphocyte subpopulations in the acute phase, characterized by lymphocytopenia, and subacute phase of LD.

Patients
Adult patients (age 18 years and older) admitted over a 15month period (2006-2007) with CAP caused by L. pneumophila serogroup 1 were studied. Community-acquired pneumonia (CAP) was defined as the presence of symptoms of lower respiratory tract infection (new cough, sputum production, dyspnoea, hypo-or hyperthermia, altered breath sounds upon physical examination) in the presence of a new infiltrate on plain chest radiography. Sputum culture, aerobic and anaerobic blood cultures, serologic analysis and immunochromatographic urinary antigen detection tests were routinely performed according to standard microbiological methods to identify potential pathogenic microorganisms such as among others Streptococcus pneumoniae, Haemophilus influenzae, Moraxella catarrhalis, Staphylococcus aureus, Pseudomonas aeruginosa, Enterobacteriaceae, Mycoplasma pneumonia, Chlamydia psittaci and Legionella pneumophila.
Patients were admitted to the Jeroen Bosch Hospital, an 800bed teaching hospital in 's-Hertogenbosch, the Netherlands. Ethylenediaminetetraacetic acid (EDTA)-anticoagulated blood from eight patients with LD was obtained on the day the diagnosis was made (acute phase) through detection of L. pneumophila serogroup 1 antigen in urine using an immunochromatographic assay (BinaxNOW Legionella, Binax, Inc., Scarborough, ME, USA). The case definition proposed by the European Working Group for Legionella Infections for confirmed cases was used in this study [27]. The validated CURB-65 score was calculated in all patients upon admission. The purpose of the CURB-65 score is to calculate the probability of mortality in patients with CAP [28]. Subsequent antibiotic treatment of patients was consistent with national guidelines and consisted of second (ciprofloxacin) or fourth generation (moxifloxacin) fluoroquinolones [29]. A second blood sample was obtained one week later or on the day of discharge from the hospital (subacute phase). Eight age and sex matched healthy volunteers who had no existing comorbidities were included as a control group.

Ethics Statement
Individual patient consent was not obtained since blood samples used in this study were all drawn for routine hematological analysis by order of the treating physician. The Internal Review Board of the Jeroen Bosch Hospital approves anonymous use of discarded blood for scientific purposes. All patients who donate blood are informed of this possibility with right of refusal. The Internal Review Board of the Jeroen Bosch Hospital specifically approved the study and waived the need for informed consent.

Infection Markers
Data on infection markers were retrieved from the hospital's laboratory information system. C-reactive protein (CRP) levels were measured with a fully automated enzyme-linked immunoassay using an Aeroset 2.0 analyzer (Abbott Diagnostics, Santa Clara, CA, USA). White blood cell (WBC) counts and leucocyte differentiation were determined on a Sysmex XE-2100 hematology analyser (Sysmex Corporation, Kobe, Japan).

Statistical Analysis
We first judged for fit to the normal distribution by using stem-and-leaf plots and quantile-quantile plots. As our data did not follow a normal distribution, Wilcoxon signed rank tests were performed for the comparison of CRP levels, WBC counts, absolute neutrophil counts, absolute lymphocyte and lymphocyte subpopulations counts in the acute and subacute phase of LD. To test the differences in CRP levels, WBC counts, absolute neutrophil counts, absolute lymphocyte and lymphocyte subpopulations counts in the subacute phase of LD and healthy controls, Mann Whitney U tests were performed. Finally, we performed Wilcoxon signed rank tests to assess the differences between the relative increase of the absolute lymphocyte count and the relative decrease/increase of the various lymphocyte subpopulations (relative expansion). A pvalue of less than 0.05 was considered statistically significant. All reported p-values are two sided. Statistical analyses were performed using SPSS software (SPSS, version 19, IBM, Chicago, Il, USA).

Patients
During the study period ten patients with CAP due to L. pneumophila serogroup 1 were hospitalized. Two patients were not included in the analysis because subacute phase blood samples were not obtained. Thus, the study population consisted of eight patients with a median age of 56 years (range 31-93). None of the patients was known for a condition that could otherwise explain lymphocytopenia upon presentation (e.g. haematological disease, chemotherapy, malnutrition). None of the patients used glucocorticoid medication upon admission or received glucocorticoid treatment. In none of the patients a coinfection with another pathogen was detected. None of the patients were immunocompromised. The median hospital stay for all patients was seven days (range 5-14) days. Timely adequate antibiotic treatment (within 12 hours after admission) was achieved in 7/8 (87.5%) patients [31]. All patients survived their disease episode. The base-line characteristics of the study patients are shown in table 1. Eight age and sex matched healthy volunteers who had no existing comorbidities were included as a control group (median age 55, range 31-93).

Flow Cytometric Analysis and Kinetics of Lymphocyte Subpopulations
Flow cytometric immunophenotyping of lymphocytes was performed on blood samples obtained in the acute and subacute phase of LD and from healthy controls (table 2). In the subacute phase of LD, significant increases compared to the acute phase were observed in CD3 + T-cell counts and CD3 2 CD16/56 + natural killer cell counts, but not in the number of CD2 2 CD19 + B-cells. Likewise, within the T-cells, significant increases were noted in both the CD3 + CD4 + T-cell and CD3 + CD8 + T-cell subpopulations. The CD4/CD8 ratio did not differ between the two time points (not shown).
Among the CD4 + T-cells, a significant increase in the absolute number of activated CD3 + CD4 + CD38 + HLA-DR + T-cells was observed in the subacute phase of LD compared to the acute phase. Likewise, a significant increase was noted for the number of activated CD3 + CD8 + CD38 + HLA-DR + T-cells. An example of the re-emergence of activated CD4 + T-cells and CD8 + T-cells in a single LD patient is shown in figure 1. In the CD4 + T-cell compartment, significant increases were furthermore observed for both the number of naïve CD3 + CD4 + CD45RA + T-cells and memory CD3 + CD4 + CD45RA 2 T-cells. Accordingly, among CD8 + T-cells, significant increases were noted in numbers of naïve CD3 + CD8 + CD45RA + CD27 + CD8 + T-cells and memory CD3 + CD8 + CD45RA 2 T-cells, while effector CD3 + CD8 + CD45RA + CD27 2 T-cells did not show a significant difference between the acute and subacute phase of LD.
Finally, we compared the decrease or increase of the different lymphocyte subpopulations on the individual patient level in the acute versus the subacute phase to the relative change (increase) of the absolute lymphocyte count in the same period (acute versus subacute). In figure 2 this relative expansion of the different subpopulations is shown.

Discussion
Lymphocytopenia has been observed in patients presenting with community-acquired LD and has recently been identified as a possible key diagnostic marker in this disease [22,23]. Here, we confirm that the acute phase of LD is characterized by absolute lymphocytopenia. Furthermore, we show that lymphocyte counts recovered in the subacute phase as the clinical condition of the LD patients improved. The one patient with an adverse clinical course (intensive care unit admission, mechanical ventilation and longest length of stay) showed a delayed recovery of the lymphocyte count compared to the other LD patients (not shown).
Lymphocytopenia is observed in several infectious emergencies and can be used in the prediction of bacteremia in infectious emergency admissions [9,10,32]. The decrease in absolute lymphocyte count observed in critically ill patients is probably related to two factors: recruitment of lymphocytes from the bloodstream to peripheral tissues and accelerated apoptosis [16,18,33,34,35]. Although relatively unknown as a marker of disease severity or prognosis, lymphocytopenia has been described in several forms of CAP [11,36]. Depression of absolute peripheral blood T-cell and T-cell subpopulation counts was observed in the majority of patients with respiratory infections. However, abnormalities in T-cell profiles did not predict outcome nor did it correlate with disease severity [11,14]. Regarding CAP patients, it has been hypothesized that depression of absolute peripheral blood lymphocyte counts and lymphocyte T-cell counts reflects shifts of these cells towards the lungs, to be used in protective functions against the causative organism [11,14,15]. Absolute lymphocytopenia may result from T-cell migration to lung lesions Table 1. Baseline characteristics upon hospitalization of Legionnaires' disease patients (n = 8). Hypothetically, immune cells, including T-cells, may control inflammatory substances that are released from the initially infected site. The immune reaction to these substances, in case of affinity to lung cells, may be responsible for lung inflammation. During this process, non-specific T-cells against these substances may display increased apoptosis [37,38]. Significant alterations in absolute peripheral blood CD4 + and CD8 + T-cell counts have been previously described in CAP patients and attributed to the severity of the underlying infection [11,39]. To our knowledge the kinetics of lymphocyte subpopulations during LD have not been previously studied in detail. While the absolute B-cell count did not significantly change during the course of LD, we observed significant increases in absolute numbers of CD4 + and CD8 + T-cells. This is in agreement with the concept that cellular immunity is presumably essential for the resolution of L. pneumophila infection whereas humoral immunity plays a role only as a second line of defence by reducing the intrapulmonary bacterial multiplication [40,41].
A/J mice inoculated intratracheally with L. pneumophila showed an increased number of circulating CD4 + and CD8 + T-cells in the second phase of infection when circulating cytokine levels and bacterial counts already declined [26,42,43]. T-cell depleted A/J mice showed impaired control of infection and increased lethality [26]. Furthermore, significant increases in the levels of the Th1 cytokines interferon-c and interleukin-12 have been observed in the acute phase of LD emphasizing the importance of cellmediated immunity in response to L. pneumophila infection [25].
Both activated CD4 + and activated CD8 + T-cell numbers were significantly decreased in the acute phase compared to the subacute phase of LD. Bronchoalveolar lavage findings of patients upon presentation with severe LD indicated a marked intrapulmonal increase of activated T-cells, defined as co-expression of HLA-DR and CD25, in combination with severe peripheral blood lymphocytopenia (mean absolute count 0.33610 9 /l) [24]. Together, these observations illustrate that recruitment of (activated) lymphocytes from the bloodstream to the lung tissue could possibly explain the observed lymphocytopenia in the acute phase of LD. The recovery of absolute lymphocyte counts in the subacute phase of LD illustrates the reconstitution of the circulating lymphocyte pool. The relative expansion of lymphocyte subpopulations during LD was most prominent in the activated CD4 + T-cells, followed by the activated CD8 + T-cells. Further analysis also showed significant relative expansion of both memory CD4 + and memory CD8 + T-cell subpopulations but not of the naïve CD4 + T-cells subpopulation and both the naïve CD8 + and effector CD8 + T-cell subpopulations.
In the subacute phase of LD, most lymphocyte subpopulation counts showed no significant differences compared to matched healthy controls illustrating that patients in the subacute phase had significantly recovered. Correspondingly, CRP levels in the subacute phase were also clearly decreased compared to the acute phase. The numbers of activated CD4 + and activated CD8 + Tcells in the subacute phase of LD even exceeded values observed in matched healthy controls, reaching significance in case of activated CD4 + T-cells. Presumably, this observation reflects an overshoot of the lymphoproliferative response during recovery. The increased number of activated CD4 + T-cells in the subacute phase compared to values obtained from healthy controls is probably due to the inflammatory response and the subsequent activation of these lymphocytes, which might be important for clearing of the infection or formation of long-term memory as it has been suggested that effector cells seed the memory pool [44].
This study has several limitations. First, we analysed a relatively small number of patients, which may affect statistical significance as some parameters have a wide range of values such as WBC Table 2. CRP levels, WBC counts, absolute neutrophil -, absolute lymphocyte -and lymphocyte subpopulation counts in the acute and subacute phase of Legionnares' disease (n = 8) and in age and gender matched healthy controls (n = 8).

Lymphocyte subpopulation markers
Acute phase of LD NA, not applicable; LD, Legionnaires' disease; CRP, C-reactive protein (mg/l); WBC, white blood cell (10 9 /l); absolute neutrophil count expressed as 10 9 /l; NK, natural killer; absolute lymphocyte count expressed as 10 9 /l; lymphocyte subpopulation counts expressed as 10 6 /l; all data presented as median and range; *p-value Wilcoxon Signed Rank tests for differences in CRP levels, WBC counts, absolute neutrophil, absolute lymphocyte and lymphocyte subpopulation counts between the acute and subacute phase of LD; # p-value Mann Whitney U tests for differences between the subacute phase of LD and healthy controls. doi:10.1371/journal.pone.0062265.t002 count and differential. Second, we used a matched control group and although statistically there is a good association between distribution of lymphocyte subpopulations in the subacute phase of LD and the control patients, this may have introduced bias. Third, although the values of the peripheral blood lymphocyte subpopulations were comparable to reference values for peripheral blood lymphocyte phenotypes in healthy adults, differences in severity of illness and age between patients may have had impact on the degree of lymphocyte apoptosis and our results should be validated in a larger sample size [45]. Fourth, the subacute phase has been arbitrarily set on 7 days after the initial presentation of the patients. Three patients recovered earlier and therefore were discharged before the 7 th day. Blood samples on the day of discharge were used in these patients to assess the lymphocyte subpopulations. It is obvious that these patients showed better clinical improvement (also reflected by the decline of their CRP levels). Nevertheless, the different time-point of assessment of the lymphocyte subpopulations in the subacute phase could have introduced bias. Fifth, based upon our research, the exact role for the T-cell activation in LD remains to be delineated. Whether the observed changes in lymphocyte subpopulation numbers in our patients are attributed to redistribution, apoptosis or a combination of both remains to be determined. Neutrophils, monocytes and serum cytokine levels were not measured and both major pathways involved in the initiation of apoptosis, i.e., receptor initiated caspase-8-mediated pathway and mitochondrial-initiated caspase-9-mediated pathway were not analysed [46,47].

Conclusions
The acute phase of LD is characterized by absolute lymphocytopenia. This recovers in the subacute phase with an increase in absolute T-cell count and re-emergence of activated and memory CD4 + and CD8 + T-cells, following the possible recruitment of