Immunologic Biomarkers, Morbidity and Mortality Among HIV Patients Hospitalized in a Tertiary Care Hospital in the Brazilian Amazon


 Background: While antiretroviral therapy (ART) has significantly improved survival rates of people living with HIV, some regions in Brazil still show a linear trend of growth in the opportunistic infections that cause HIV-associated mortality. We aimed to describe HIV-associated morbidity and mortality among hospitalized medical patients in a tertiary care hospital in Manaus, in the Brazilian Amazon, by investigating clinical data and immunologic biomarkers in order to assess predictive factors of mortality in this patient group. Methods: We prospectively measured concentrations of cytokines Th1/Th2/Th17 and soluble CD14 (sCD14) and reviewed the laboratory parameters and opportunistic infections in outcomes of either death or discharge of eighty-three HIV/AIDS patients who were admitted in 2017-2018 to the Fundação de Medicina Tropical Doutor Heitor Vieira Dourado (FMT-HVD) in Manaus. Results: The mortality in the sample studied was 20.5%. Tuberculosis (TB) showed a relative risk (RR) =1.86 (confidence interval (CI) 1.14 to 2.81: and p = 0.026), and weight loss was the symptom (RR=1.81; CI: 1.21 to 2.53 and p = 0.007) most highly associated with the death outcome in HIV/AIDS inpatients. Univariable analyses showed that the eosinophil count, platelet distribution width (PDW), and alanine aminotransferase were the only laboratory parameters that differed among patients who died. In relation to cytokines and sCD14 levels, no differences were found between those who died or were discharged. A multivariable logistic regression model was used to predict mortality and showed that individuals with no digestive syndrome (especially the absence of oropharyngeal candidiasis), nor TB are 63% to 76% less likely to die, respectively. In addition, increases in PDW values also decreased the probability of death. Curiously, patients who were discharged showed a trend towards a concomitant increase in PDW and mean platelet volume (MPV) in relation to those who died.Conclusions: Opportunistic infections continue to be major events in morbidity and mortality of HIV/AIDS patients, and the relationship between increased PDW and the likelihood of survival suggests the need for future studies on innate immune response of platelets in HIV/AIDS inpatients.


Background
Brazil has been successfully combating the HIV/AIDS epidemic by providing state-of-the-art antiretroviral therapy (ART) and a uni ed, universal and free public health system widely disseminated to the population 1,2 . The mortality rate due to HIV/AIDS in Brazil has fallen signi cantly in the last twenty years, especially in the southeastern, southern and mid-western regions. This has been principally in uenced by a government policy of linking HIV/AIDS treatment and medication dispensing to compulsory noti cation within the public health system (Sistema Único de Saúde/SUS) 1,3 . This revised system has been assisting decision-making at the individual and programmatic levels of different antiretroviral treatment regimens for all people living with HIV4.
Since the beginning of the epidemic in 1982, more than three hundred thousand deaths with HIV/AIDS being attributed as the cause have been reported. The number of new cases of HIV/AIDS has been gradually decreasing in recent years in Brazil, in uenced by the southeast, southern and central western regions, while the northern and northeastern regions still show a linear trend of growth in the detection of new cases 3 . As of 2018, more than 900,000 AIDS cases had been reported in Brazil and approximately 600,000 are on ART 1 . After a signi cant reversal in mortality due to the adoption of ART in 1996, deaths related to the HIV/AIDS epidemic have been showing a downward trend in Brazil as a whole, however this trend is still subtle in the northern and northeastern regions 1, 3,5 .
Located in the north of Brazil, the State of Amazonas has peculiar characteristics, such as geographic isolation and a lack of access to full health services, mainly in the interior of the State. Manaus is the capital and the largest city of the state, and the decentralized health units are located in the neighborhoods and semi-urban areas. These are generally more accessible and promote the highly active antiretroviral therapy (HAART) 2 . The Fundação de Medicina Tropical Doutor Heitor Vieira Dourado (FMT-HVD), located in the middle of the city, is a tertiary care hospital which performs HAART services and admits moderate to severely ill people living with HIV. This represents a contingent greater than 85% of all patients receiving ART, as well as concentrating data on incidence levels of AIDS, mortality rates and late diagnosis 2,5 . Studies on the HIV/AIDS epidemic in the region have indicated over time that the trend in the number of cases continues to increase principally in men, that tuberculosis (TB) is still the leading cause of death, and that respiratory failure is the main cause of hospital admission for these patients [5][6][7][8] .
The hallmark of HIV infection is the progressive depletion of CD4 T cells caused by an active replication of the virus in the lymph nodes and blood stream, which, over time, lead to a chronic HIV infection [9][10][11] . In addition, HIV cripples the immune system and allows reactivation of dormant pathogens or increases susceptibility to exogenous pathogens. These co-infections further fuel the immune activation that characterizes a strong predictor of disease progression in HIV infection 12,13 . Although morbidity and mortality caused by opportunistic diseases have decreased due to HAART, coinfections continue to be important factors for the pathogenesis of chronic HIV-infected patients. In our prospective study, in which HIV/AIDS inpatients were classi ed by either death or hospital discharge, we assessed which comorbidities, coinfections and concentrations of serological biomarkers were predictors of the risk of death.

Study type
This prospective study involved HIV/AIDS inpatients admitted to the Fundação de Medicina Tropical Doutor Heitor Vieira Dourado (FMT-HVD), the tertiary referral hospital for infectious diseases in Manaus. This is a prospective study, in which sampling was done for convenience. The target population consisted of HIV/AIDS patients of either sex who were admitted to the FMT-HVD in years 2017 and 2018. Eighty-three participants aged between 12-70 were enrolled in this study, the sampling was by convenience and we enrolled those patients that were admitted in FMT-HVD within than 72 hours. After signature of the informed consent form, the blood collection was performed. All had their CD4 count and viral load monitored. Plasma for the tests was stored at -80 °C.

Clinical data
During rst contact with the patients, socio-demographic data were collected, such as name, age, gender and use of ART. Through access to the electronic medical records (iDoctor®) of the patients, the following clinical data were collected: general health status, comorbidities, co-infections, treatment, clinical manifestations (weight loss, diarrhea, vomiting) and death.
Laboratory data were obtained at the hospital itself, which currently uses software to store medical records and electronic prescriptions. This instrument also provided laboratory data such as: blood count, serology (Cytomegalovirus -CMV, Toxoplasmosis, Epstein Barr -EBV, Herpes virus, hepatitis B virus -HBV, hepatitis C virus -HCV), immunological markers (viral load and T cell count CD4 + T and CD8 + T) and biochemical markers (bilirubin, creatinine, lactic dehydrogenase, GT range, albumin, alkaline phosphatase, GPT and OGT). chronic lymphocytic leukemia, Hodgkin's disease, aplastic anemia, and neurological disorders like multiple sclerosis and myasthenia gravis, were monitored.

Blood sample collection
On the same day of the patient's enrollment, after the interview and signing of the consent form, 5 mL of blood was collected by vacuum venipuncture. The samples, collected in dry tubes, were centrifuged at 3,500RPM for 5 minutes at 25 ºC to obtain the serum and aliquoted (1 mL) for analysis of in ammatory markers. The whole blood collected in an anticoagulant tube was homogenized, aliquoted (1 mL) and stored for future analysis. Both were stored at -80 ºC until use.
Recombinant human CD14 protein (R&D Systems) was used as a standard.

Statistical analysis
Data were tabulated in an Excel database created by the researchers and analyzed using the GraphPad Prism program, version 7. For data analysis, patients were classi ed in two groups: Death (those who died during hospitalization) and Survival (those who were discharged from hospital). Descriptive analysis was performed with mean and standard deviation for numerical variables, with normal and median distribution for the others. For comparison between categorical variables, the Chi-Square test was used to determine relative risk. Serum soluble CD14 and cytokine concentrations were compared between Death and Survival groups using the Mann Whitney Test. Logistic regression predictions were performed, the level of statistical signi cance de ned in both cases was p < 0.05.
Laboratory data were obtained from electronic medical records, followed by blood collection. As previously mentioned, almost all of the patients were classi ed as having AIDS due CD4 T counts being 350 cells/mL. The mortality in the sample studied was 20.5%. Univariable analyses with of HIV-RNA > 1000 copies and CD4 T < 350 cells/mL showed no relative risk for death (Table 1). Comorbidities were assessed to identify predictive factors of mortality. Among the clinical data, weight loss showed a relative risk (RR) of 1.81 with a con dence interval (CI) of between 1.18 and 2.97 (p = 0.013). The active coinfections were diagnosed by microbiological or serological tests. Tuberculosis was the only risk associated with death, with a RR of 3.44 (CI 2.33-5.08; p = 0.001). The IgG serological tests used for previous infections did not differ between HIV/AIDS patients who died or were discharged from hospital.  Table 1 shows the relative risk of anemia in HIV/AIDS patients and indicated a trend to death (p = 0.079), which it was also revealed by lower hemoglobin concentration, as seen in Table 2, although they were not statically different. In contrast, total leukocytes, neutrophils, and monocytes did not differ between groups and remained in normal ranges in most of HIV/AIDS patients ( Table 2). In addition, eosinophil counts were reduced in both groups in relation to normal range, but counts in those who died were even lower (p = 0.03). Table 1 also shows a trend towards digestive syndromes in HIV/AIDS patients who died. In the assessed biochemical tests, HIV/AIDS patients preserved their liver and renal functions.
Univariable analyses of laboratory variables were also carried out to identify predictive factors of mortality ( Table 2). The percentage of lymphocytes of those who died showed a tendency to be lower. The CD4 T counts were very reduced for AIDS patients in both groups, however they did not differ statistically. In relation to CD8 T cells, the death group presented normal values, while patients that were discharged tended towards a slight increase, but not su cient to be statistically different. The ratio of CD4:CD8 cells below 0.20 due reduction of CD4-T provided a notion of how weak the immunity of the HIV/AIDS patients was ( Table 2).
Despite anemia being common between them, renal function was able to maintain bilirubin, albumin, and creatinine levels in normal ranges (Table 2). Hepatic tests did not indicate liver disorders, although gamma-glutamyltransferase levels were higher, alkaline phosphatase levels were slightly elevated in relation to normal ranges. The same was observed with aspartate aminotransferase and alanine aminotransferase levels that showed slight elevations. The soluble CD14, the chemokine IL-8 and Th1/Th2/Th17 cytokines were assessed between the two groups. Though no differences were observed, it is worth mentioning the strati cation in levels of in ammatory biomarkers sCD14, IL-8, IL-6, and IFNγ ( Fig.  1A-D) in relation to Th1/Th2/Th17 cytokines (Fig. 1E-L). We observed predominance of IL-17A and IL-4 over IL-12 and IL-2 in both groups (Fig. 1E-H). Lastly, we observed a trend towards a predominance of the anti-in ammatory IL-10 over TNFα and IL-1β in HIV/AIDS patients who died (Fig. 1I-L).
A multivariable logistic regression model assessed all biomarkers for likelihood of death in HIV/AIDS patients (Table 3). Those without digestive syndrome were 63% less likely to die (crude OR = 0.37), while 76% were likely to survive when all biomarkers were assessed together (adjusted OR = 0.24). As mentioned previously, more than half of the digestive syndromes were due oropharyngeal-esophageal candidiasis. Odds ratios for tuberculosis were more pronounced, individually, those without TB were 76% less likely to die (crude OR = 0.24), while 86% were likely to survive when all biomarkers were assessed together (adjusted OR = 0.14). Among continuum biomarkers, only PDW predicted death in HIV/AIDS patients, both individually (crude OR = 0.84) and with all biomarkers (adjusted OR = 0.85, Table 3). To understand these data, Fig. 2A allows us to explain how as the PDW values increase, the probability of death decreases. Furthermore, the direct relation between PDW and MPV was common in all HIV/AIDS patients. However, patients who were discharged showed a trend towards a concomitant increase of PDW and MPV in relation to those who died (Fig. 2B-C).

Discussion
Despite mortality rate due to HIV/AIDS falling signi cantly in Brazil, the northern and northeastern regions of the country have seen an increase and, between these two regions, the trend has been increasing at a greater rate in the state of Amazonas 3,5 . Our study found respiratory syndromes, followed by neurological and digestive syndromes, to be the main causes of hospitalization. In most of these cases, opportunistic pathogens were observed, especially TB, neurotoxoplasmosis and oropharyngeal-esophageal candidiasis. These data demonstrate that, although the country is progressing towards reaching the UNAIDS 90-90-90 treatment target, the northern and northeastern regions are no in line with the general trend. Here in the Brazilian Amazon, opportunistic infections continue to be the main causes of death of people living with HIV, which is a similar situation for those coming from cities in other regions with low economic power [14][15][16][17][18][19] .
Regarding the risk of death, our multivariable logistic regression model indicated two main causes of death, HIV/TB co-infection and digestive syndrome. Tuberculosis was highly prevalent in HIV/AIDS patients admitted at the FMT-HVD, as has been occurring for the past ten years 5 . The prevalence of HIV/AIDS co-infection can be much higher, since TB is generally paucibacillary in HIV-infected individuals, which makes diagnosis di cult 7,20 . As the HIV infection weakens the host's immune response to TB, HIV/TB co-infection results in a more dramatic progression 7,16,17,20−22 . In relation to digestive syndrome, it is worth highlighting that more than half of digestive syndromes were due to oropharyngeal-esophageal candidiasis. The occurrence of oral candidiasis has been recognized as an indicator of immune suppression. It is known that some patients did not wish to receive HAART due to skepticism, social factors, or because, even while taking medications, they experienced virological and therapeutic failure 23 .
Here, CD4 counts lower than 200cells/ul were observed in almost all HIV-infected patients. In this context, the occurrence of oropharyngeal-esophageal candidiasis can help physicians to diagnose the clinical progression of HIV infection, as observed in other countries 24,25 .
HIV/AIDS is characterized by a state of chronic hyperactivation. In this sense, several serological markers are being investigated in the search for a better predictor of the risk of death12,26-29. In the present study, no difference was observed in sCD14, IL-8 and Th1/Th2/Th17 cytokines between patients who died or were discharged. Biomarker concentrations are observable outcomes of complex biological processes that are only partially understood, and this may be the reason for a lack of consensus regarding immunological markers in prediction of mortality in HIV/AIDS patients [30][31][32][33][34] . Curiously, we observed much higher levels of in ammatory biomarkers sCD14, IL-8, and IL-6 than for other cytokines, which is consistent with the in ammatory status of the HIV/AIDS patients 35  According to multivariable analysis, PDW was the only parameter which indicated a likelihood of death in HIV/AIDS patients. This became clearer through the trend towards a concomitant increase of PDW and MPV in patients who were discharged. MPV and PDW are easily measured platelet indexes, which increase during platelet activation. However, PDW seems to be a more speci c activation marker due to the increase in the number and size of pseudopodia, under activation, and possibly affects platelet distribution width 36 . Platelets are small, anucleate cells that have hemostatic and in ammatory proprieties and are considered to be not so innocent bystanders. Whereas platelets appear to support the host response to extracellular infections, in intracellular infections they contribute to immune evasion 37 .
The same authors reported that reduced platelet counts in patients with active TB are associated with mortality 37 . Here, some patients showed thrombocytopenia, but it was not a predictor mortality. Conversely, our multivariate analysis showed that, as the PDW values increased, the probability of death decreased. Thus, our data suggest the role of platelets as host defense effectors in agreement with these studies [38][39][40][41][42] .
The limitation of this study is related to the study design, which did not allow patient follow-up, with only one blood sample being taken after recruiting patients for the investigation of serological markers.
Another limitation was that the information regarding deaths of patients was obtained solely via electronic medical records; neither did we have access to patients who evolved negatively and were admitted to the intensive care unit.
Among the immunological markers, we could not perceive any differences, as the blood collection was obtained only once and within 72 hours of admission to the hospital, though perhaps we could not distinguish them because all patients were very weak and ill. That is why it is important to perform tests on admission and a follow-up at least once right after the worsening of the state of patient, as well as another before discharge. The data collected corresponded to those closest to the day of blood collection used for the markers. No data was obtained for TB-immune reconstitution in ammatory syndrome.

Conclusions
In conclusion, the opportunistic infections continue to be the major contributory events in mortality of HIV/AIDS patients. The risk of mortality found in HIV-tuberculosis co-infected patients underlines how