The definition of leptospirosis was made by clinical-epidemiologic criteria (exposure to soil and water contaminated with urine from typical leptospirosis reservoirs or exposure to flooding after rainfall and clinical features compatible with leptospirosis) plus any of the following diagnostic confirmations: serologic diagnosis by Immunoglobulin M (IgM) enzyme-linked immunosorbent assay (ELISA) or microscopic agglutination test (MAT) (one single sample >1:800; seroconversion or a 4-fold increase in two samples, one collected at the acute phase and another collected at convalescence); positive culture; typical pathological features of leptospirosis in the necropsies (interstitial nephritis, acute tubular necrosis, disorganization of the liver cell plates, pulmonary hemorrhage) accompanied or not by positive immunohistochemistry to the Leptospira antigen. All leptospirosis cases had to fulfill the criteria of septic shock (described below). We excluded cases of leptospirosis in which patients had acquired other bacterial infections during intensive care. Autopsies were carried with a median of 8 h after death (from 7–12 h).

Septic shock patients were assigned according to the septic shock definition valid at admission (19): presence of bacterial infection (proven or not) with a systemic inflammatory response syndrome (SIRS) associated with hypotension (mean arterial pressure < 70 mmHg) requiring vasoactive drug infusion. The cases included in this study also fulfill SOFA and quick SOFA criteria for sepsis definition (20). We excluded patients with immunosuppressive disease or taking immunosuppressive medication. Autopsies were carried out with a median of 8 h after death (from 7–11).

The control group was composed of healthy portions of the spleen of individuals who had abdominal trauma and had undergone laparotomy with splenectomy. We excluded patients who arrived at the emergency room with haemorrhagic shock, patients with immunosuppressive disease or taking immunosuppressive medication.

We collected the following data from charts of all patients: age, gender, APACHEII score, intensive care stay duration and peripheral lymphocyte count. For leptospirosis patients, we collected data for aspartate aminotransferase (AST), alanine transaminase (ALT), total serum bilirubin, platelet count, leukocyte count, serum potassium, creatinine and urea.

The APACHE II is a score system that predicts prognosis of patients in the general intensive care unit (ICU), by including age, previous health conditions and physiologic variables (i.e., vital signs and laboratory exams) and is calculated during the first 24 h of hospitalization in the ICU (21).

Sections 5 μm thick of paraffin-embedded spleen samples were stained with Haematoxylin-Eosin (HE) for morphological analysis.

The assay was carried out on 4 μm thick paraffin-embedded spleen samples according to Hsu et al. (22), following a procedure standardized by the Pathology Department of the University of São Paulo Medical School. Briefly, specimens were deparaffinized and hydrated in ethanol, the antigens were retrieved in TRIS/EDTA buffer pH 9.0, for 20 min, at 95°C. The primary antibodies were diluted in 1% bovine albumin solution and incubated overnight at 4°C. Afterwards, the biotinylated secondary antibody and streptavidin-peroxidase complex (Dako, Agilent, Santa Clara, CA, USA,) were applied and specimens were incubated for 30 min, at 37°C. Antibody information is provided in Table 2. 3,3-diaminobenzidine tetrahydroxychloride (Sigma-Aldrich/Merck, Darmstadt, Germany) was used as chromogen and the slides were counterstained with haematoxylin. All reactions were performed with positive and negative controls. Two antibodies (anti CD4 and CD8) required signal amplification using the CSAII kit (Dako, catalog number K1497), according to the manufacturer's instructions.

In the case of cytokine staining, the assay was modified as follows: before blocking, slices were treated with 0.1% saponin in PBS, for 10 min, and then rinsed with PBS. In addition, after secondary antibody incubation, another incubation with saponin was carried out, followed by rinsing and a second blocking of endogenous peroxidase.

The histological features of the spleens were analyzed as following: A—by the presence or absence (0 = absence, 1 = presence of the event) of: pericapsular/subcapsular hemorrhage, red pulp hemorrhage, red pulp necrosis, vessel necrosis and thrombosis, central artery of the white pulp with signs of endothelial activation, extra medullary hematopoiesis, depletion of T-zone and B-zone, atrophy or hyperplasia of follicles. B—red pulp congestion type: focal or diffuse. C—subjective semi-quantitative score (0 = normal, 1 = discrete, 2 = moderate, and 3 = intense) according to the presence of: red pulp congestion, haemosiderin pigment, red pulp reticular cells/macrophages, red pulp polymorphonuclear cells, red pulp plasma cells, red pulp lymphocytes. D—a quantitative analysis was done using a grid-scale with 10 × 10 subdivisions in an area of 0.0625 mm² to count the cell phenotypes and immune cells expressing cytokines in their cytoplasm. Ten fields were counted randomly at 400 times magnification in the two compartments of the spleen (red pulp and white pulp). Evaluations were carried out independently by two pathologists.

The categorical data are expressed as percentages and we calculated the Pearson Chi-Square and Fischer's Exact Test. The quantitative data are expressed as median and 25th or 75th percentiles of cells per mm². Differences between the medians of the three groups were analyzed using Kruskal-Wallis. Differences among two groups were analyzed by unpaired t-test and by nonparametric methods (Mann-Whitney test). Values of p < 0.05 were considered statistically significant.

Table 3 summarizes leptospirosis patient clinical and laboratorial characteristics. The group was composed mainly by men (82%) which took 4.5–7 days to present to the hospital after symptoms began and had a high APACHE II score (21 to 28). The median APACHE II score was 23.5, which corresponds to a survival rate of 30%, reflecting the severity of the leptospirosis cases (21). In seven cases, the microagglutination test was positive, and the main serovars were the Icterohaemorrhagiae (5 cases) and Copenhageni (3 cases).

Septic shock patient data (Table 4) are as follows: 10 individuals with a median age of 57 years (minimum 19, maximum 68; p = 0.02 compared to controls, not different from leptospirosis), 20% male (p = 0007 compared to leptospirosis; p = 0.001 compared to controls). Main comorbidities were: systemic arterial hypertension (80%) and diabetes mellitus (70%). The autopsy showed that sepsis cases had bronchopneumonia (60%), skin and soft tissue infections (40%), pyelonephritis (20%), and intra-abdominal infections (10%). Median intensive care stay was of 7.5 days (1–30 days). All patients were treated with vasoactive drugs and mechanical ventilation. The APACHE II score was high, but calculated in only two cases (25 and 28).

As for the control group, healthy portions of the spleen were used from trauma patients. There were 12 individuals, with a median age of 27 years (minimum 22, maximum 40), 92% male; 9 patients were admitted due to a motorbike or car collision with a lamppost and 2 cases resulted from falls.

Table 5 and Figures 1, 2 display the results of the histological parameters analyzed. The congestion in the red pulp was more intense and diffuse in the leptospirosis and sepsis groups than in the control group (Figures 1A,I). Pericapsular or sub capsular hemorrhages were more frequent in the control group than in the leptospirosis and sepsis groups. Hemorrhages in the red pulp occurred equally among the groups (Figure 1A).

In the leptospirosis and sepsis groups, the reticular cells showed pronounced hyperplasia and hypertrophy and there was an increased number of macrophages in the red pulp (Figure 1D).

In the cords, neutrophil, eosinophil, plasma cell and lymphocyte numbers were increased in leptospirosis and sepsis groups in comparison with the control group.

In response to stress, including bacterial infections, stromal haematopoietic stem cells around the sinusoids of the spleen can be activated, enabling extramedullary haematopoiesis to occur (7, 24). Although this observation is not directly associated with the immune response, we would like to mention that we describe, for the first time, that extramedullary haematopoiesis occurs in leptospirosis. It occurred more frequently in leptospirosis patients than in the sepsis group and was not observed in control patients, as expected (Figure 1C). We believe this observation may be associated with the Leptospira virulent factors or with the fact that leptospirosis patients were younger and may have had more responsive precursor cells.

The follicles of the leptospirosis and sepsis (Figure 1I) groups were atrophic, whereas those of the control group were normal (Figure 1E).

The T-cell and B-cell-dependent regions had a lower cellular density in the leptospirosis and sepsis groups as compared to the control group (Figures 1F,I).

The endothelium of the central artery showed signs of activation in the leptospirosis and sepsis groups in comparison with the control group (Figures 1F,I). Artery activation is characterized by: tumefaction of the endothelial cell (cubical aspect as compared to normal flat cells) and arterial wall oedema.

The Leptospira antigen was positive in the spleen of 8 out 11 of the leptospirosis patients (72.7%) (Figure 1G).

NK cells (CD57⁺) were present at lower levels (≤20% of control) in the leptospirosis and sepsis groups as compared to the control group (Figures 2A–C).

Dendritic S100⁺ cells levels were increased more than 10-fold in the red pulp of both leptospirosis and sepsis groups as compared to control and the levels found in leptospirosis tissues were higher than those in the septic tissues. However, in the white pulp, S100 cells were higher only in the leptospirosis group (Figures 2D–F). The changes in dendritic cell numbers were the most extreme findings regarding cell data in our study.

Cleaved caspase 3 positive (apoptotic) cell levels were lower in the leptospirosis and sepsis groups than in the control group in the white pulp (Figures 2M–O). Indeed, no active caspase 3-positive cells were found in the leptospirosis group (compared to 62.4 cells/mm² in the control group). Apoptosis was also significantly lower in the leptospirosis group compared to the sepsis group in the red pulp.

CD68⁺ cells were found at higher levels in the sepsis group as compared to leptospirosis and control groups.

The TCD4⁺ cells were in lower quantity in the leptospirosis and sepsis groups than in the control group (Figures 2G–I). More TCD4⁺ cells were found in the red pulp of the sepsis group than in that of the leptospirosis group.

TCD8⁺ cell numbers in the red and white pulps were lower in the sepsis group than in the leptospirosis (Figure 1H) and control groups.

Both leptospirosis and sepsis groups had more CD20⁺ cells than patients with control in the white pulp (Figures 2J–L). The leptospirosis group had more CD20⁺ cells than the sepsis and control groups in the red pulp.

There was a minimum expression of Th1 cytokines (IFN-γ, IL-1, IL-2, and IL-6) in all experimental groups, with no statistical difference. An exception was IL-12, with a significantly higher expression in the red pulp of the leptospirosis patients as compared to the control group. We have added immunohistochemistry positive controls presented in Supplementary Figure 1 to indicate that low expression was not a result of incomplete staining.

TNF-α levels were higher in the leptospirosis group as compared to sepsis and control groups in both the white and red pulps (Figures 2P–R). In fact, no TNF-α-positive cells were found in sepsis and control patients in the white pulp, whereas 32 cells/m² were found in leptospirosis samples.

Control and leptospirosis groups had similar distribution of IL-4 in both white and red pulps and this expression was higher than that of the septic group in the red pulp.

The number of cells expressing IL-10 was greater (more than 10-fold, p < 0.0005, one of the greatest changes found in our study regarding cytokines) in leptospirosis and sepsis groups as compared to control in both red and white pulps (Figures 2S–U).

As for TGF-β, its expression was similar in all three groups in the white pulp, whereas levels were higher in the sepsis group compared to the other two groups in the red pulp.

Leptospirosis patients have a very characteristic demographic: they are mostly young and male. We found it difficult to find corresponding sepsis patients, that were older and only 20% male, and controls, that were younger than the leptospirosis group. The most common serovars identified in the present cases were Icterohaemorrhagiae and Copenhageni, of the Leptospira interrogans serogroup, which are the most prevalent in the State of São Paulo and are associated with severe cases of leptospirosis (23).

Some authors believe that antibiotic treatment in severe cases of leptospirosis is unnecessary because, at this stage, it is likely that all bacteria have already been eliminated and that focus should be on treating organ damage due overreacting immune and inflammatory response (25). We describe that the Leptospira antigen is still present in the spleen at time of death, which implies that early antibiotic treatment probably could have saved some patients and should be prescribed in cases of severe leptospirosis. However, we should point out that the time until admission was found to have a median of 5 days. It is likely that this delay in treating patients with antibiotics (ceftriaxone) and intravenous hydration was an important factor in mortality. Our results indicate that health providers should be trained to recognize leptospirosis as early as possible and start antibiotic treatment to avoid sepsis.

Our results demonstrate a strong similarity in histological aspects between leptospirosis and septic spleens of patients, as opposed to controls, displaying a pattern of acute splenitis as found by other authors (26–28). The endothelial cells of the central artery had histological signs of activation in leptospirosis without necrosis. In contrast, the sepsis group had endothelial activation plus, in two cases, endothelial necrosis and thrombosis. Increased plasma levels of sE-selectin and Von Willebrand factor, possibly derived from endothelial cells, have been previously described in leptospirosis patients and higher sE-selectin levels were associated with lower mortality (29).

The Leptospira antigen showed strong labeling in a granular pattern widely distributed in the red and white pulps, both intracellularly (macrophages, reticular cells, endothelial cells) and extracellularly, indicating that these bacteria probably have an important role in the pathogeny of severe leptospirosis. This is the first study to show that the Leptospira antigen is present in the spleen of patients with leptospirosis, although this antigen has been previously found in the liver, kidneys, heart, lungs, and skeletal muscle (3, 30–32).

As discussed earlier, the presence of Leptospira and signs of immunosuppression in later and severe stages of infection corroborate that antibiotic treatment should occur at all stages of the disease as prescribed for other types of sepsis (33) and proven to be beneficial in animal models (34).

Although our study was carried out evaluating immune response in both red and white pulps, we consider that the red pulp is a transition zone and will focus our discussion on results obtained in the white pulp.

We report, for the first time, that patients with severe leptospirosis have a compromised innate immune response associated with a decrease in NK cells and low levels on IFN-γ. The decrease in NK levels was also found in sepsis. Low NK cell count was previously found in an ex vivo report for leptospirosis (35), although Leptospira are known to stimulate these cells in vitro; NK levels at early stages of the disease probably correlate with its outcome (36, 37), as found in sepsis (38).

We found a >10-fold increased number of dendritic cells in leptospirosis and in sepsis patients. In leptospirosis, an ex vivo study found that dendritic cells proliferate and secrete a high quantity of Th1 cytokines and a low quantity of IL-10 (39). Reports for sepsis (10, 40) show a loss of dendritic cells in the spleen due to apoptosis as opposed to our results. This can be explained by our different methodology: we employed a broad marker to detect dendritic cells—S100, while other authors used specific tissue markers for follicular cells (CD21 and FDC-M1) (10, 40). The S100 protein allows us to detect myeloid cells, including interdigitating dendritic cells, Langerhans cells and follicular dendritic cells (41). We chose this marker because of its availability and because it allows us to detect dendritic cells in both red and white pulp. We also believe a high number of dendritic cells is possible in the leptospirosis group and can be attributed to the virulence factors, young age and predominant male gender.

Regarding inflammation, we believe patients had suffered from an intense inflammatory response early during infection since both leptospirosis and sepsis patients presented with severe tissue lesions including nephritis, interstitial pneumonitis, pulmonary hemorrhage, hepatopathy, and muscular lesions. As mentioned above, histological analysis also indicated splenitis probably caused by an intensive inflammatory response to the infective agents.

Both leptospirosis and sepsis patients showed clear follicle atrophy in the white pulp, with low cellular density in the T- and B-cell dependent regions. This is a strong indication of the inhibition of the adaptive immune response. TCD4⁺ levels in the leptospirosis and sepsis groups were significantly lower as compared to controls. The depletion of TCD4⁺ cells in leptospirosis appears to be essential for the host's homeostasis (42, 43).

We have no data to indicate how TCD4⁺ cells are lost in leptospirosis, though it is likely by apoptosis, as found in animal models (44) and in spleens of patients with septic shock (8–10). However, we found a decreased level of apoptosis in the sepsis and leptospirosis patients as compared to controls. It is possible that our results are due to the methodological approach: we used antibodies against activated caspase-3, whereas other authors used anti-caspase 9 or the TUNEL assay, which may label necrotic cells (45). Another possibility is that there is a massive apoptotic event during earlier stages of infection, which is completed by the later stages—and cannot be detected—also suggesting that no new lymphocytes migrate to the spleen. Other forms of cell death or control of cell population by inhibition of proliferation are other possible explanations for the low amount of immune cells found in sepsis and leptospirosis, especially since we found low levels of proliferation stimulants (i.e., NK cells and cytokines) (46).

Regarding CD20⁺ B lymphocytes, they were found in increased numbers in both infectious groups as compared to control which may be associated with a humoral response to the infectious agents.

We showed that TNF-α levels are increased in leptospirosis in comparison with the sepsis and control groups, in spite of the lack of regulation of other Th1 cytokines. TNF-α is the most studied cytokine in leptospirosis and has been found to be a marker for clinical outcome (47, 48).

We found that IL-1, IL-2, IL-4, IL-6, IL-12, and IFN-γ levels were very low or undetectable, with no major changes among patient groups, confirming previous studies (48–53). IL-12 levels were expected to be high in leptospirosis, based on in vitro and animal studies (39, 51, 54). Our results showed elevated expression of IL-12 in leptospirosis in the red pulp, but since it was absent from the white pulp, the overall result is probably of immunosuppression. The low secretion of IFN-γ during the immunoparalysis phase of sepsis is a key aspect of this syndrome and is associated with its outcome (55).

We found increased levels of IL-10 in both sepsis and leptospirosis groups indicating an immunosuppressed environment and confirming previous studies in other tissues (51, 56, 57).

TGF-β levels were also increased in the spleens of leptospirosis and sepsis patients, as described in kidneys of hamsters infected with Leptospira (57).

We conclude from our data and its corroboration by other studies that there is immunoparalysis in leptospirosis as also occurs in sepsis. However, as our reviewers have pointed out, the data is restricted to a late time point (after death), a single tissue (spleen) and morphological analysis of preserved tissues—and therefore that it is hard to be certain of the immunological processes that occurred in the patients. Therefore, the following analysis should be viewed with caution.

The in situ decrease (or lack of expression) in Th1 pro-inflammatory cytokines (IL-6, IFN-y, IL-1β, IL-2r, and IL-12) and the increase in IL-10 indicate that there is an inhibition of the innate and acquired immunity during the acute and severe phases of leptospirosis and sepsis. Previous data obtained in vitro or in patient serum indicate that there is an initial Th1 response in leptospirosis (36, 37, 39, 47, 58–62) as opposed to our findings in later stages of the disease. This may indicate that in later stages of severe leptospirosis there is an immune system burnout or some sort of inhibition of immune response triggered by bacteria.

Human patients and animal models show that sepsis is associated with an immunosuppressive state (55, 63–65) corroborating that severe leptospirosis is similar to sepsis.

Overall, our study shows that there is an “immunoparalysis” in severe leptospirosis and sepsis: low quantity of NK and TCD4⁺ cells, low expression of IFN-γ, IL-12, IL-6, IL-1 and IL-2 and high expression of IL-10. The successful immune response to leptospirosis appears to depend on the secretion of Th1 pro-inflammatory cytokines (36, 37, 59, 61, 66). We believe that the host's response to the infectious agent needs to hit a sweet spot: not too high (such as with cytokine storms) and not too low, as in immunoparalysis (63, 64, 67). Immunoparalysis has been shown in sepsis, where it favors the development of new infections and morbidity (63, 68). Sepsis associated with septic shock has been correlated with a Th2 response (55, 64, 69). Indeed, in hamster models of Leptospira infection, it has been shown that there is an initial Th1 response followed by a predominance of high IL-10 at later stages of the infection (51, 56, 57).

We also found that there appears to be a lack of interaction between innate and adaptive immunity when comparing the increase in dendritic cells, but decrease in IL-2 and IL-12 expression—in dendritic cells. Had the levels of these cytokines increased, then they should have been able to stimulate other Th1 cells (70).

We hypothesized that severe leptospirosis with shock has some elements of immunosuppression-like septic shock as we have demonstrated in the spleen. Leptospirosis patients should be included in clinical trials for sepsis treatment and biomarker discovery as well as sepsis caused by Gram +/– bacteria (9, 71, 72).

Our results indicate that an immunomodulatory treatment may aid in patient recovery, controlling the initial immune response in order to prevent immune system burnout or stimulating the immune system at later stages (8, 12, 73). Another practical aspect of our results is that it suggests that immunoparalysis markers can be useful in patient diagnosis and to aid treatment (63).

Our study has some limitations: it was based on retrospective data recovered from patient files. This has drawbacks because some clinical data was not collected at the time of admission or during treatment (especially for several sepsis cases), which restricts our analysis. Also, because Brazilian law requires that autopsies be carried out at least 6 h after death—the median time until autopsy was 8 h in our study—some sample autolysis may have happened, with loss of antigens, although HE staining shows no signs of tissue degradation and staining controls were positive. Since we based our study on patients that died from leptospirosis or sepsis, we do not have spleen data for mild cases, where there was recovery. We also do not have spleen samples from severe cases before death (at earlier stages of the severe disease). Overall, we based our study on morphological aspects of the spleen and do not have data relating to systemic immune response, such as circulating immune cell types and circulating cytokines and other inflammatory markers. We also designed our study based on broad cell markers and not on more specific markers (for example, CD21 for dendritic follicular cells).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.