Statistical analysis was carried out only in bacterial (n=8) and non-bacterial (n=6) groups, while the mixed group was excluded. The mean value of procalcitonin in the bacterial group was higher than in the non-bacterial group (12.35±27.9 ng/mL vs 1.45±1.26 ng/mL), but the bacterial group was not normally distributed. Hence, Mann Whitney test results showed that the bacterial group procalcitonin levels were higher when compared with non-bacterial groups, but it was not statistically significant (p=0.302) (Table 4).

HIV/AIDS was one of the sexually transmitted diseases that was still dominated by males, due to high sexual activity and high risk for free sex accompanied by low awareness of the use of contraceptives [7]. This data was also found in most of the previous studies that male was dominated. The age range of most previous study subjects was in the productive age of 31-40 years [7].  The most recent education of other study subjects was before the tertiary level. One of HIV transmission factors was the lack of knowledge about this disease where education could affect it [15]. Housewives were the occupation of most of this study subjects that also was in accordance with the Republic of Indonesia's Health Ministry data in 2017. This can be explained by the high risk of sexual behavior of their partner (husband) and lack of knowledge about HIV disease [7].

All subjects suffered from the chief complaint of shortness of breath, cough, and fever accompanied by infiltrating on the chest X-ray suggesting pneumonia. The Pulmonary Complications of HIV Infection Study reported that respiratory symptoms often occurred in HIV patients and the incidence was increasing in CD4 cell count <200 cells/µl [16]. Whereas Stolz et al. [11] reported that HIV patients with pneumonia coinfection did not have specific symptoms.

Leukocytes and neutrophils levels did not have a significant difference between bacterial and non-bacterial pneumonia groups. The same result was also found by Mikula et al. [14] Intracellular HIV replication in target cells express the CD4 receptor such as monocyte-macrophages, lymphocytes, neutrophils, and some eosinophils and basophils. Hence, the number of these cells was decreased by the destruction process [12]. We concluded that these markers could neither be used as inflammation-specific markers in bacterial nor non-bacterial infections in HIV/AIDS patients.

Almost all subjects had CD4 cell count < 200 cells/µl. This was consistent with the results of several previous studies that the risk of pulmonary infection was increasing in HIV patients with a lower CD4 count [16]. Hirschtick et al. [13] reported that HIV patients with CD4 counts < 200 cells/μL could increase the risk of bacterial pneumonia by 5.7 fold. We also found that the bacterial group had a higher CD4 cell count if compared to the non-bacterial group although the difference was not significant. Benito et al. [16] reported that bacterial pneumonia could occur in HIV patients with various CD4 counts. Whereas fungi or viruses often cause pneumonia in HIV patients with CD4 < 200 cells/μL [16].

Our data showed that subjects who did not receive ARV therapy suffered pneumonia more often when compared to patients who had received ARV therapy (13 vs 7). This had concordance with a previous study by Sullivan that ARV could significantly reduce the incidence of bacterial pneumonia in HIV patients with CD4 below 200 cells/μL [17]. ARV could boost the immune system in HIV patients by activating T helper cells type 1, thereby producing cytokines interferon-gamma (IFN-γ). Furthermore, IFN-γ together with IL-10 induces NF-IL6 in macrophages that will bind to the LTR of HIV. This process can suppress the transcription of HIV in the host cell [18]. Nevertheless, of the 7 patients who had received ARV therapy, there were 5 patients on ARV therapy for less than 3 months. Many previous studies had suggested that the response of an increased immune system by ARV was different for every HIV individual, generally occurring after 3 months of ARV therapy [5].

Since ARV was widely used in 1996, the etiological diagnosis of lung infection had changed in which bacterial pneumonia was more common than PCP or other opportunistic infections [16]. There were more bacterial groups in this study compared to non-bacterial or mixed groups. The number of CD4 also affected microorganisms that cause pneumonia. The lower the CD4 cell, the more susceptible opportunistic infections occurred in HIV patients [16]. In addition, the type of specimen used affected the results of the microbiological examination. The diagnostic value of BAL for pneumonia in immunocompromised patients varies from 33-83% so that false-negative results might occur [19].

Pseudomonas aeruginosa was the most common aerobic bacteria found in our study (42,9 %). These bacteria were Gram-negative bacteria that could cause community pneumonia and the incidence was increasing in HIV patients [16, 20]. Acinetobacter baumannii which was one of the bacteria associated with hospital-acquired infection was identified in one subject who had a history of several previous hospital controls. Candida sp. was the most common fungal species detected in this study. Several previous publications mentioned that the incidence of Candida pneumonia increased in immunocompromised patients. Candida pneumonia could occur by aspiration from endobronchial colonization or blood-borne (hematogenous) spread in patients with candidemia [21]. Further, our viral PCR results showed that Human Rhino Virus (HRV) infected four subjects. Rhinovirus infections often progress to upper respiratory tract diseases such as the common cold in the spring, but it might cause more serious infections by replicating on the lower respiratory tract in immunocompromised patients. Serra et al. [22] reported that HRV was the most common viral pathogen in HIV patients with pneumonia.

Our statistic analysis showed that mean and median levels of procalcitonin in the bacterial group were higher if compared to the non-bacterial group, and the difference was not significant. This was in accordance with several previous studies that the diagnostic value of PCT was decreased and the value of PCT levels was hard to interpret in immunocompromised patients. Mikula et al. [14] reported that the diagnostic value of procalcitonin decreased in the HIV group, which was 68% sensitivity and 8-11% specificity at a lower cut-off value of 0.16-0.17 ng/mL. Procalcitonin (PCT) levels in HIV patients could be influenced by several factors. First, leukopenia could reduce procalcitonin levels [23]. This was explained due to leukocyte plays a role in PCT production [24]. Second, HIV infection itself could affect plasma procalcitonin levels. Mikula et al. [14] reported that among HIV patients without bacterial infection, levels of PCT could be normal or slightly increased. This can be explained due to HIV infection could activate the body's immune system through the expression of TLR on dendritic cells inducing IL6 and TNF-α production, which both inflammatory cytokines could induce PCT production [5]. Third, species of the pathogen could also affect the levels of PCT in the blood. Base on Leli et al. [25], it said that the median value of PCT in patients with Gram-negative bacterial infections was higher if compared with Gram-positive bacterial infections or fungal infections. This is due to differences in bacterial wall interactions called lipopolysaccharides (LPS) that bind to the TLR of the host cell. Gram-positive bacteria could activate the TLR2 pathway while the Gram-negative bacteria activate the TLR4 pathway, giving rise to differences in proinflammatory cytokines produced such as IL1, IL6, and TNF-α, which these cytokines play a role in PCT stimulation [24, 25]. It was consistent with our results that higher average PCT levels in three samples with Gram-negative bacteria (Pseudomonas aeruginosa, Klebsiella pneumoniae, and Acinetobacter baumannii) were detected if compared with Gram-positive bacteria ( Streptococcus pneumoniae) infected subjects (14.168 ng/mL, 4.847 ng/mL, 4.680 ng/mL vs 3.490 ng/mL). In addition, polymicrobial infections could also significantly increase PCT levels if compared to monomicrobial infections such as fungi or Gram-positive bacteria [25]. Fourth, the condition of sepsis or SIRS caused by fungus or virus could also increase PCT levels. Increasing PCT levels rarely occurred by both pathogen infections due to the production of gamma interferon cytokines (IFN-γ) inhibiting PCT production. Nevertheless, systemic fungal infection (sepsis) could increase PCT levels due to the production of TNF-α cytokines, inducing PCT [26]. Fifth, the history and duration of empirical antimicrobial therapy in immunocompromised patients could affect microorganism patterns in the host organ which indirectly influences PCT levels. However, several studies reported that BAL could still identify pathogens in immunocompromised patients who had been given antibiotic therapy. This is caused by a large number of pathogenic colonies in the lower respiratory tract or the high virulence rate of the pathogen, causing a slow response to antimicrobials given to immunocompromised patients [11].

Although this study result could be considered for treatment in HIV patients, it still has some limitations; the small number of study samples and almost all of them had very low immune status (CD4 <200 cells/mL). Multicenter studies with larger samples and varying immune statuses were needed to determine the cut-off point of procalcitonin. In addition, this study did not perform a blood culture examination to determine the severity of the disease. Besides, symptoms could also be caused by infection from other organs. Last, the time of blood collected from each subject was different. This is related to the regulation and half-life of PCT in the blood.

Conclusion

We concluded that there were no differences in the levels of procalcitonin between non-bacterial and bacterial groups in HIV patients with pneumonia coinfection. Difficult interpretation of the procalcitonin levels in HIV patients needs to be reconsidered for establishing the etiological diagnosis of pneumonia. Larger research was needed to look for other more effective markers of infection in HIV/AIDS patients.

Acknowledgments: None

Conflict of interest: None

Financial support: This research was fund by research grant of
Airlangga University of Surabaya Indonesia.

Ethics statement: All procedures were in accordance with the
ethical standards on human experimentation in the Helsinki
Declaration of 1964 and its later derivat. Informed consent was
obtained from all patients included in the study.