Artigo Is severe visceral leishmaniasis a systemic inflammatory response syndrome ? – A case control study

1. Laboratory of Leishmaniasis, Tropical Diseases Institute Natan Portella, Teresina, PI, Brazil. 2. Department of Community Medicine, Federal University of Piauí, Teresina, PI, Brazil. 3. National School of Public Health Sérgio Arouca, Oswaldo Cruz Foundation, Rio de Janeiro, RJ, Brazil. 4. Federal University of Rio de Janeiro, Rio de Janeiro, RJ, Brazil. 5. Federal University of Minas Gerais, Belo Horizonte, MG, Brazil. 6. Department of Mother and Child Health, Federal University of Piauí, Teresina, PI, Brazil. 7. Federal University of Piauí, Teresina, PI, Brazil. Partially presented at the 13th International Congress of Immunology, Rio de Janeiro, August 21-27, 2007. Address to: Dr. Carlos Henrique Nery Costa. Instituto de Doenças Tropicais Natan Portella. R. Artur de Vasconcelos 151-Sul, 64001-450 Teresina, PI, Brasil. Tel: 55 86 3221-3413; Fax: 55 86 3222-3248 e-mail: chncosta@gmail.com Received in 10/11/2009 Accepted in 16/04/2010 ABSTRACT Introduction: The objective of the study is to identify the main risk factors for death by New World visceral leishmaniasis and establish a coherent pathogenic substrate of severe disease based on clinical findings. Methods: Seventy-six deceased inpatients and 320 successfully treated inpatients with VL were studied in a case control study. Results: Bacterial infection and bleeding were mutually exclusive events leading to death. Five risk factors were unique for death by bacterial infection (malnutrition, pulmonary rales, severe anemia, severe absolute neutropenia and higher neutrophil count), while another six were unique for death by bleeding (jaundice, severe relative neutropenia, severe thrombocytopenia, liver injury, kidney failure, higher bone marrow parasite load). Bacterial infection, bleeding, severe anemia, diarrhea, dyspnea, edema, jaundice and bone marrow parasite load were the main syndromes of visceral leishmaniasis among successfully treated patients. Conclusions: The data support the idea that bacterial infections are due to immune paralysis. Broad organ and system involvement is plausibly due to the high production of proinflammatory cytokines, whose actions fit well with visceral leishmaniasis. The syndromes and causative mediators are typical of a slowly developing systemic inflammatory response syndrome. Key-words: Visceral leishmaniasis. Kala-azar. Severe inflammatory response syndrome. AIDS. Bleeding.

Visceral leishmaniasis (VL), or kala-azar, is a parasitic disease endemic to many temperate and tropical countries.The clinical picture consists of a protracted course of fever, pallor, wasting, hepatosplenomegaly and pancytopenia.Thousands of deaths have occurred in lethal epidemics and the disease is an important opportunistic infection in patients with HIV 1 .The agent of the anthroponotic disease is Leishmania donovani on the Indian subcontinent and in East Africa.Leishmania infantum (=Leishmania chagasi) has a zoonotic cycle involving humans, dogs and wild canines in Asia, the Mediterranean area and the New World.The organisms are obligate intracellular protozoa, mostly found in the spleen, bone marrow, liver and lymph nodes, and the disease is transmitted by sand flies.Treatment is based on pentavalent antimonials and amphotericin B 1 .
There is no known effective strategy to reduce the transmission of VL, although vaccines are still a hope.Even with treatment, 5-10% of patients die 2 .In Brazil, urbanization of the disease has been reported since the early 1980s, progressively spreading throughout the country.VL leads to about 240 deaths each year in Brazil, a number much higher than that of malaria and dengue hemorrhagic fever 3 .Guidelines for managing severe disease have been developed 4 , but fatality rates remain high.Previous studies demonstrated that extremes of age, malnutrition, length of fever and the presence of vomiting, diarrhea, severe anemia, jaundice, enlarged spleen and thrombocytopenia are risk factors for death 2,[5][6][7] .Although bacterial infections and bleeding account for most fatal complications 5,[7][8][9] , little is known about their pathogenic pathways.However, disseminated intravascular coagulation 10 , liver failure 11 and hemophagocytic syndrome have been implicated.
A case-control study was therefore designed to investigate the nature of VL deaths, especially those associated with bacterial infections and hemorrhage.The independence of these two outcomes was evaluated, the risk factors associated with them were

METHODS
listed, and a coherent network of clinical and laboratory related data was identified and discussed so that a logical hypothesis explaining the pathophysiology of death from VL could be formulated.

Study population
At a referral hospital for infectious diseases in Teresina, northeastern Brazil, medical records from deceased inpatients and survivors with VL from 1996 to 2005 were randomly selected.Only patients with typical symptoms plus a positive direct bone marrow (BM) smear, BM culture or indirect immunofluorescence test (Biomanguinhos, Rio de Janeiro, Brazil) at dilution ≥ 1:80 were included.Records without proper data entries were excluded.The Institutional Review Board of the Natan Portella Institute of Tropical Diseases (Instituto de Doenças Tropicais) approved this retrospective medical record review.

Exposure and outcome measurement
Only signs and symptoms present at hospital admission were considered, as well as the initial laboratory results.Weighing of patients was infrequently performed.The weight-for-age z-score was calculated; patients ≤18 years-old were classified as malnourished if the z-score was less than -1.95.The presence of bacterial infection was determined mostly on clinical and radiological grounds and a positive culture was not required.A positive BM smear defined higher BM parasite load.The records of deceased patients with VL were independently reviewed by two observers to determine the causes of death.

Data analysis
Risk factors for all deaths were identified by univariate and multivariate logistic regression and associations between variables were expressed as odds ratios (OR) and respective 95% confidence intervals (95%CI).Multivariate analysis was performed using a step-by-step backward selection procedure and variables remained in the final model if the association with the outcome was significant at the 5% level.
The binomial test was applied to check whether the number of dead patients with both bacterial infection and bleeding was independent.Comparisons of the ORs between variables associated with death by bacterial infection and by hemorrhage were performed using seemingly unrelated estimation procedures.Among survivors, risk factors for the main syndromes were investigated and Spearman correlation coefficients between the quantitative laboratory tests were calculated.Analysis were conducted in Stata 9.0 (SataCorp, College Station, TX).

Ethical
The Ethical Review Board of Federal University of Piauí granted approval to the study.

Causes of death
Thirty-one patients died from severe bacterial infection, 29 died from bleeding and three died with simultaneous hemorrhage and evidence of infection.The observed number of concurrent deaths by bleeding and infection was significantly less than the expected number (EN) of deaths by the two events if they were statistically independent (EN = 14.3, p <0.001) (Table 2).

Symptoms and signs associated with death by all causes
In the univariate analysis, the risk of death was higher among infants <1 year and among patients >40 years-old.Since only 60 persons had their weight measured, the well-known effect of malnutrition on kala-azar mortality could not be determined with certainty.However, malnutrition was remarkably associated with death by bacterial infection (OR = 6.2).The following variables were associated with death: vomiting, diarrhea, respiratory distress (dyspnea and pulmonary rales), edema, severe anemia, hepatic involvement (jaundice and elevated alanine amino transferase), kidney syndromes (kidney failure and proteinuria), neutrophil count abnormalities (absolute number ≥ 5,000/mm 3 and relative count ≤10%), severe thrombocytopenia (≤50,000/mm 3 ), low serum albumin, high BM parasite load, history of any sign of hemorrhage, bacterial infection and HIV/AIDS (Table 3).

Symptoms and signs associated with death by bacterial infection and by bleeding disorders in univariate analysis
Variables similarly associated with death by bacterial infection and by hemorrhage: age <1 year-old and >40 years-old, vomiting, diarrhea, dyspnea, edema, any bleeding, HIV/AIDS and relative neutrophilia >70%.Variables associated with death by bacterial infection but not by hemorrhage: pulmonary rales, bacterial infection at admission, severe anemia, severe absolute neutropenia and higher neutrophil count.

Variables associated with death by hemorrhage but not by bacterial infection:
jaundice, severe relative neutropenia, liver injury, kidney failure, severe thrombocytopenia and high BM parasite load (Table 4).

DISCUSSION
Age and comorbidities, but not sex, were relevant to death by VL in this study.Children <1 year-old and adults aged over 40 yearsold were at an increased risk of death.Analysis of patient medical records revealed that surviving very young children had a higher risk of bacterial infections, diarrhea and severe anemia; these findings may be relevant for therapeutic considerations.Analysis also revealed more males with jaundice.Malnutrition also was not significantly associated with death in general, but it showed a remarkable effect on death by bacterial infection, though not on bleeding.Despite the small number of patients with HIV infection, it was a serious threat to the survival of patients with VL.However, it was not clear why coinfection increases fatality: whether it worsens the risks associated with the disease itself or if it leads to death by other opportunistic infections.In any case, the data indicate that patients with VL should be routinely tested for HIV and special attention should be provided for those infected.
Aside from bacterial infections and bleeding manifestations, several distinct clinical syndromes were indicative of death: a digestive syndrome composed of vomiting and diarrhea; a respiratory syndrome with rales and dyspnea; an edematous syndrome; a hepatic syndrome with jaundice and alterations in liver enzymes, a renal syndrome with failure and proteinuria; a complex hematologic syndrome with bipolar associations of neutrophils, anemia, thrombocytopenia and hemorrhages; and an immune suppressive syndrome with opportunistic bacterial infections and high BM parasite load.Therefore, the key question is to determine whether any common denominator exists in this spectrum of events that could serve as the target for interventions to reduce lethality by VL.
Bacterial infections and bleeding were the two most prominent outcomes associated with death by VL.Among the variables determined as connected with death solely by bacterial infection, anemia, pulmonary rales and neutrophilia were merely signs of current infection.Therefore, aside from infrequent severe neutropenia, there was no clinical clue predicting or explaining susceptibility to bacterial infections.However, similar to that which occurs in sepsis 12 , generalized cytokine-mediated immune paralysis could also occur in VL and this hypothesis should be investigated.In contrast, since other variables were determined to be associated with death by both infection and bleeding, the data seem to suggest that immune suppression coexists with inflammation, with the balance between them determining patient outcome: immune paralysis or severe inflammation.On the other hand, the association of certain variables only with bleeding provides support for the idea that hemorrhaging is part of the expression of a systemic inflammatory response.Thrombocytopenia may be only an adjuvant factor for bleeding in VL and not its cause, since it seldom reached levels that would trigger bleeding on its own; indeed, platelet count has been shown to be inversely correlated with megakaryocytic hyperplasia in the bone marrow 13 , which denotes consumption.The apparent cause of bleeding is disseminated intravascular coagulation (DIC), as previously described in VL 10 .It is likely that DIC follows the same pathway as in sepsis: inflammation causes DIC by activation of coagulation and fibrinolysis triggered by the proinflammatory response 14 .
The hematological data were the most complex to interpret.Analysis showed that severe anemia was associated with higher BM parasite load and that it predicted death by bacterial infection but not by bleeding.Moreover, it was related to inflammatory signs, such as vomiting, edema and lung inflammation, suggesting that bacterial infections and increasing parasite load might trigger anemia via inflammation.Finally, it seems that bleeding was actually not the cause of anemia in VL, because most patients with anemia did not present hemorrhagic events.Even modest neutrophilia and extreme neutropenia were determined to be associated with lethal infections, together with severe anemia, which suggests that a prompt search for infection should be conducted and empirical antibiotic usage should be considered in patients with these blood counts.However, since severe relative neutropenia was also strongly associated with bleeding, neutropenia may be a consequence of inflammation.The contrast of BM hypercellularity with neutropenia observed in patients with VL 15 suggests that neutrophils are produced and then sequestered from the peripheral circulation.Lastly, higher numbers of neutrophils apparently reduced the BM parasite load.
Analysis of other syndromes that composed the clinical picture of severe VL also offered interesting insights.The association of edema with severe VL might be a consequence of three mechanisms acting together: a) capillary leakage due to systemic endothelial activation; b) low oncotic pressure due to low serum albumin, a feature of acute phase reaction observed in VL; and c) increased hydrostatic pressure led by glomerulonephritis, as detected by hematuria.Analysis verified that serum creatinine was a risk factor only for death by bleeding and that it was correlated with elevated alkaline phosphatase, suggesting the presence of interstitial nephritis, as previously described 16 , and that it is part of the general inflammation status.Moreover, the presence of proteinuria should alert clinicians to the additional risk of glomerulonephritis to previously established interstitial nephritis.Dyspnea was associated with lethal bacterial infections and bleeding.Respiratory injury was demonstrated by the presence of lung sounds, coughing and x-ray infiltrates.The association with bleeding and edema, evidence of inflammation, suggests that acute respiratory distress syndrome following interstitial pneumonia occurs in severe human VL 17 .The questions that arise from the association of liver involvement with patients who died from bleeding include why hepatitis occurs in VL and what the nature of its relationship with hemorrhaging is.Since no hepatocyte parasitism occurs in VL 18 , pathogenesis is probably also due to inflammation.The relationship with bleeding might give the impression that absolute liver failure is frequent in VL.However, although hepatitis is common in VL, death by fulminant hepatitis is unusual 19 .Thus, the role of the liver in bleeding in VL is probably restricted to a relative deficiency of coagulation factors during established DIC.Finally, analysis of the data showed that diarrhea and vomiting are interconnected.Of note is the negative association of splenomegaly with diarrhea.
Certain other findings offer additional clues concerning the evolution of severe disease.Interestingly, BM parasite load was associated with fatal bleeding and severe anemia.Additionally, the longer the disease was active, the more the survivors bled, which is an indication of progressive inflammation in VL.This last finding indicates that bacterial infections follow a different pathogenic pathway compared to bleeding, though leading to it, and that negative feedback mediators might exist that oppose the two outcomes.Some weaknesses concerning data collection in this retrospective hospital-based case-control study should be highlighted: the quality of data recorded on medical charts may not have been uniform throughout the study period; bacterial infections were diagnosed by clinical suspicion as a result of early antibiotic usage; BM parasite load was coded as a binary variable.However, these problems mostly likely led only to random misclassification bias, which may actually have contributed to underestimate some of the associations.
This diversity of pathogenic associations with VL severity is the most striking result of this study.These findings are similar to sepsis and malaria, suggesting the same underlying mechanism, which is systemic inflammatory response with multiorgan failure 20 .The main difference is time; while these are acute diseases, VL develops over a protracted course and possibly because of this, patients do not present with acute hemodynamic changes.L. chagasi is an intracellular parasite restricted to the macrophage lineage, dendritic cells, neutrophils, eosinophils and fibroblasts 21 , no parasitism of hepatocytes, pneumocytes, glomerular or tubular cells or epithelial or glandular intestinal cells occurs 17,18,22,23 , and no toxins are released by Leishmania.Therefore, the absence of parasite actions or products that would harm the host cells or tissues is one more indication that the systemic pathogenicity of VL is dependent on the host response.Although inflammatory tests were not evaluated in this study, inflammation is an ordinary feature of VL, as indicated by increases in erythrocyte sedimentation rate and acute phase proteins [24][25][26] .Moreover, the proinflammatory cytokines interferon-γ, TNF and IL-6 and the chemokine IL-8 are elevated in VL as part of a broad and sustained innate response 27,32,33 .Indeed, the clinical actions of these cytokines in humans fit very well with the status seen in VL in this study: wasting, vomiting, fever, thrombocytopenia, diarrhea, hepatotoxicity, coagulopathy, alterations in leukocyte trafficking and release and acute phase response [34][35][36][37] .Therefore, the hypothesis that these cytokines mediate the pathogenesis of severe VL is very consistent and should be considered for future investigations.

Costa
CHN et al -Severe visceral leishmaniasis