A-DROP and Blood Urea Nitrogen-to-Serum Albumin Ratio are Useful in Assessing the Severity of Pneumocystis Pneumonia in Patients Without Human Immunodeciency Virus Infection: A Retrospective Study

Background: There is an increasing incidence of Pneumocystis pneumonia among individuals without the human immunodeciency virus (HIV) infection (non-HIV Pneumocystis pneumonia). However, the prognostic factors for patients with non-HIV Pneumocystis pneumonia have not been identied. Moreover, A-DROP (for classifying the severity of community-acquired pneumonia) or the blood urea nitrogen-to-serum albumin ratio, which is reported to be predictor of mortality of community-acquired pneumonia, has not been established as an ecient prognostic factor in patients with non-HIV Pneumocystis pneumonia. In this study, we analyzed the prognostic factors for non-HIV Pneumocystis pneumonia and evaluated the effectiveness of A-DROP and the blood urea nitrogen-to-serum albumin ratio as prognostic factors. Methods: This retrospective study involved a chart review of the medical records of 102 patients diagnosed with non-HIV Pneumocystis pneumonia between January 2003 and May 2019 at ve medical facilities. Prognostic factors associated with the 30-day mortality were assessed using multiple logistic regression analysis. Results: Among the 102 patients with non-HIV Pneumocystis pneumonia, 46 (45.1%) had autoimmune diseases, 19 (18.6%) had hematological malignancies, 18 (17.7%) had solid malignancies, and 19 (18.6%) had other diseases. The 30-day mortality rate for non-HIV Pneumocystis pneumonia was 20.5% in this study population. Compared with survivors, non-survivors had signicantly lower serum albumin levels and a signicantly higher age, corticosteroid dosage at the onset of Pneumocystis pneumonia, alveolar–arterial oxygen gradient, A-DROP score, lactate dehydrogenase levels, blood urea nitrogen levels, and blood urea nitrogen-to-serum albumin ratio. The results of multivariate analysis showed that a high A-DROP score and blood urea nitrogen-to-serum albumin ratio at treatment initiation were signicantly associated with the 30-day mortality risk. Conclusions: A high A-DROP Abbreviations: PcP; Pneumocystis pneumonia, invasive positive pressure


Background
Pneumocystis pneumonia (PcP) is one of the most serious opportunistic infections associated with human immunode ciency virus (HIV) infection. The introduction of PcP prophylaxis and highly active antiretroviral therapy has reduced the incidence and mortality rates of PcP in HIV-infected patients (HIV PcP) [1,2]. In contrast, the incidence of PcP in patients without HIV infection (non-HIV PcP) but with malignancies, those who have undergone organ transplantation, and in those undergoing immunosuppressive therapy is increasing [2][3][4]. The mortality rate was 10-20% in patients with HIV PcP and 19.6-60% in patients with non-HIV PcP [5,6]. The prognostic factors of non-HIV PcP have not yet been determined; however, several factors, such as high levels of alveolar-arterial oxygen gradient (A-aDO 2 ), high lactate dehydrogenase (LDH) levels, high blood urea nitrogen (BUN) levels, low serum albumin (Alb) levels, and high pneumonia severity index (PSI), were reported to be associated with a poor prognosis of non-HIV PcP [7][8][9][10]. PSI is a scoring system proposed by the American Thoracic Society/Infectious Diseases Society of America for classifying patients with community-acquired pneumonia (CAP) [11]. Alternatively, A-DROP (age, dehydration, respiratory failure, orientation, systolic blood pressure) is a scoring system proposed by the Japanese Respiratory Society (JRS) [12] and used widely for classifying the severity of patients with CAP in Japan. High BUN and low serum Alb levels are considered as poor prognostic factors in patients with CAP [13][14][15][16]. In addition, a high BUN-to-serum Alb (BUN/Alb) ratio is reportedly a useful marker of mortality from CAP, hospital-acquired pneumonia (HAP), and aspiration pneumonia [17][18][19]. A previous report suggested that A-DROP may underestimate the severity of non-HIV PcP [20]. Therefore, A-DROP has not been established as a prognostic factor in patients with non-HIV PcP. The effectiveness of the BUN/Alb ratio also has not yet been evaluated in patients with non-HIV PcP. In this study, we aimed to identify the prognostic factors for patients with non-HIV PcP. We also evaluated the effectiveness of A-DROP and the BUN/Alb ratio, which are useful in patients with CAP, as predictors of prognosis in patients with non-HIV PcP.

Study design and population
A retrospective study was conducted in patients diagnosed with non-HIV PcP between January 2003 and May 2019 at ve medical facilities (Shinshu University Hospital, Nagano Municipal Hospital, Minaminagano Medical Center Shinonoi General Hospital, Japanese Red Cross Society Nagano Hospital, and Japanese Red Cross Society Suwa Hospital) in Nagano Prefecture, Japan, on the basis of the data available in their medical records. According to data from previous studies [21], a diagnosis of non-HIV PcP was made in our study if the patients met the following 4 criteria: (i) immunosuppressive status; (ii) diffuse bilateral ground-glass opacity on chest radiography or computed tomography scans; (iii) detection of Pneumocystis jirovecii in respiratory specimens by direct staining or polymerase chain reaction assay; and (iv) elevated plasma (1→3)-β-D-glucan levels. The levels of plasma (1→3)-β-D-glucan were measured using a Wako β-glucan test (Wako Pure Chemical Industries, Osaka, Japan) or a Fungitec G test MK (Seikagaku Corporation, Tokyo, Japan). Infection with PcP was indicated if the plasma (1→3)β-D-glucan level was over the upper limit of the normal range. The normal range of plasma (1→3)-β-Dglucan level is ≤ 11 pg/mL in the Wako β-glucan test and ≤ 20 pg/mL in the Fungitec G test MK.
Immunosuppressive status was de ned as reduced immune status and was seen in patients who suffered from malignancies (hematological or solid), in patients who were treated with immunosuppressive therapy, or in patients who had undergone an organ transplant and hematopoietic stem cell transplant.

Page 4/21
The following demographic and clinical data were collected from the medical records of the patients diagnosed with non-HIV PcP: age, gender, body mass index, underlying diseases, treatments for underlying diseases, corticosteroid dosage at the onset of PcP for underlying diseases, PcP prophylaxis, days from symptom appearance (onset of PcP) to treatment, physical ndings at treatment initiation (percutaneous oxygen saturation (SpO 2 ), blood pressure, and status of consciousness), respiratory support (invasive positive-pressure ventilation [IPPV]), initial treatment for PcP, adjunctive corticosteroid therapy, A-DROP score, laboratory ndings before treatment, and nal outcome. Corticosteroid dosage was expressed as the equivalent dose of prednisolone. Respiratory failure was de ned as partial pressure of arterial oxygen (PaO 2 ) ≤ 60 Torr or SpO 2 ≤ 90% in room air. A-DROP is a scoring system proposed by the JRS to assess the severity of CAP [12]. A-DROP was evaluated using the following parameters: (i) age (men ≥ 70 years, women ≥ 75 years); ( ) BUN ≥ 21 mg/dL or dehydration; ( ) respiratory failure; ( ) disturbance of consciousness; ( ) hypotension (systolic blood pressure ≤ 90 mmHg). Each parameter corresponds to 1 point in the A-DROP score. An A-DROP score of 0 is considered mild; 1-2, moderate; 3, severe; and 4-5, extremely severe [12]. Consciousness status was assessed using the Glasgow Coma Scale (GCS), and disturbance of consciousness was de ned as GCS ≤ 14. The endpoint of this study was all-cause mortality in the patients with non-HIV PcP within 30 days from treatment initiation.

Statistical analysis
Data are expressed as proportions for categorical variables and as median (range) for continuous variables. Categorical variables were compared using the chi-squared or Fisher's exact test, and continuous variables were compared using the Mann-Whitney test between the survivors and nonsurvivors. The independence of each variable was tested using Spearman's rank correlation coe cient.
Prognostic factors in relation to 30-day mortality were assessed by multiple logistic regression analysis. The receiver-operating characteristic (ROC) curve was used for evaluating the ability of prognostic factors to estimate 30-day mortality. P < 0.05 was considered statistically signi cant. All statistical analyses were performed using IBM SPSS Statistics software, version 26 (IBM, Armonk, NY, USA). Rheumatoid arthritis was the most common autoimmune disease in the present study, seen in 29 of the 46 patients (63.0%) with autoimmune disease. In the entire cohort, regarding treatment modalities for the underlying diseases, 34 patients (33.3%) were treated with corticosteroids alone and 53 patients (53.0%) were treated with corticosteroids in combination with other drugs, such as immunosuppressants, biological agents, and antitumor drugs. Fifteen patients (14.7%) were treated with therapies other than corticosteroids. Corticosteroids were administered in 87 patients (85.3%) at the onset of PcP for the already existing diseases, and the median corticosteroid dosage (prednisolone equivalent) was 15 mg/day (range 2-400 mg/day). Two patients received PcP prophylaxis with trimethoprim/sulfamethoxazole (TMP/SMX).

Baseline clinical characteristics of patients with non-HIV PcP
Patient prognosis was evaluated with 30-day mortality, de ned as the mortality at day 30 after initiation of the treatment. The 30-day mortality rate in non-HIV PcP was found to be 20.5% (21/102) in the present study.

Comparison of clinical data between survivors and nonsurvivors
In the 30 days following treatment initiation, 81 patients survived and 21 died. The comparison of clinical data between the survivors and non-survivors is shown in Table 2. Non-survivors were signi cantly older than survivors (75 years vs. 67 years, p = 0.019). The prevalence of underlying diseases did not signi cantly differ between the survivors and non-survivors. However, the use of corticosteroids was more frequent in the non-survivors than in the survivors for treatment of their underlying diseases. The corticosteroid dosage administered for therapy of underlying diseases at the onset of PcP was signi cantly higher in the non-survivors than in the survivors (25 mg/day vs. 15 mg/day, p = 0.001). There were no signi cant differences in PcP prophylaxis between the survivors and non-survivors, and the most frequently used drug for prophylaxis was TMP-SMX, in both survivors and non-survivors. There was no signi cant difference in the interval of days from the onset of PcP to the treatment initiation between the non-survivors and survivors (p = 0.162). Respiratory failure was signi cantly more prevalent in the nonsurvivors than in the survivors (95.2% vs. 60.4%, p = 0.002) at treatment initiation.

Comparison of laboratory ndings before treatment between survivors and non-survivors
The comparison of the laboratory ndings before treatment showed that the non-survivors presented with signi cantly higher levels of A-aDO 2 (Table 2). In addition, the non-survivors showed signi cantly lower PaO 2 to fraction of inspired oxygen (FiO 2 ) ratio (166.8 vs. 251.9, p = 0.023) and levels of total protein (TP) (5 g/dL vs. 5.6 g/dL, p = 0.033) and serum Alb (2.2 g/dL vs. 2.8 g/dL, p = 0.002) than the survivors (Table 2).
Treatments TMP-SMX was the most frequently used anti-PcP drug in the initial treatment regimen for both survivors and non-survivors. The administration of adjuvant corticosteroid therapy for the treatment of non-HIV PcP was signi cantly more frequent in the non-survivors than in the survivors (90.4% vs. 70.3%, p = 0.016) ( Table 2). The requirement of IPPV was signi cantly higher among the non-survivors than the survivors (61.9% vs. 13.5%, p < 0.001, Table 2).

Prognostic factors for non-HIV PcP in relation to 30-day mortality
Prognostic factors in relation to 30-day mortality were assessed by multiple logistic regression analysis. The independence of each variable with a p value < 0.05 obtained upon comparison of the data between the survivors and non-survivors was assessed using Spearman's rank correlation coe cient. Treatment factors were excluded from this analysis to enable examination of the prognostic factors at the beginning of treatment. Age, respiratory failure, PaO 2 /FiO 2 ratio and TP, Alb, and BUN levels were not included in the multiple logistic regression analysis because there was a strong correlation between these factors, with a Spearman's rank correlation coe cient of over 0.6. Therefore, the variables of corticosteroid administration and dosage for underlying diseases, disturbance of consciousness, A-DROP score, A-aDO 2 , LDH levels, and BUN/Alb ratio were included in the multivariate logistic regression analysis.

Relationship between A-DROP and 30-day mortality
A-DROP was used for estimating the disease severity of patients at treatment initiation. The median A-DROP score was signi cantly higher in the non-survivors than in the survivors (3 vs. 1, p < 0.001, Table 2). The 30-day mortality rate was 0% in the patients with an A-DROP score of 0, 3.2% in those with a score of 1, 29.6% in those with a score of 2, 36.6% in those with a score of 3, and 37.5% in those with a score of 4 or 5 (Table 4). There was a positive correlation between 30-day mortality rate and A-DROP score, with the mortality rate rising with an increase in the A-DROP score.

Discussion
Previous studies have shown that the prognosis of patients with non-HIV PcP is worse than that of patients with HIV PcP [5,6]. The mortality rate associated with non-HIV PcP ranged between 19.6% and 60% in previous studies [5,6], while the 30-day mortality rate was 20.5% in the present study. The remarkable nding of this study was that the A-DROP score and BUN/Alb ratio were reliable prognostic factors that can be used for estimating the 30-day mortality in patients with non-HIV PcP.
A previous study reported that an increase in PSI score was associated with mortality in non-HIV PcP [7].
It has been shown that the A-DROP score was not signi cantly different between the non-HIV PcP survivors and non-survivors at admission, but it was signi cantly higher in the non-survivors than in the survivors at treatment initiation [20]. Consistently, in the present study, the A-DROP score was found to be signi cantly higher in the non-survivors than in the survivors at treatment initiation (3 vs. 1, p < 0.001).
The 30-day mortality rate was 0% in those with an A-DROP score of 0, 3.2% in those with a score of 1, 29.6% in those with a score 2, 36.6% in those with a score of 3, and 37.5% in those with a score of 4 or 5.
Although it was reported that the A-DROP score underestimated the severity of patients with non-HIV PcP, which developed as CAP [20,22], the 30-day mortality rate increased with increased A-DROP score at treatment initiation in the present study. Therefore, A-DROP might be useful for classifying the severity of patients with non-HIV PcP, especially the A-DROP score at treatment initiation for PcP. Previous studies have shown that the 30-day mortality rate was 0% for patients with CAP with an A-DROP score of 0, 0-3.1% in those with a score of 1, 3.1%-4.6% in those with a score of 2, 9.9-15.9% in those with a score of 3, and 19.6%-34.0% in those with a score of 4 or 5 [23][24][25]. Although it might not be accurate to compare our results with the 30-day mortality rate of CAP, the 30-day mortality in those patients with an A-DROP score of 2 or 3 was relatively higher in non-HIV PcP than in CAP, and the 30-day mortality rate in patients with an A-DROP score of 4 or 5 was relatively the same in non-HIV PcP and CAP. The components of the A-DROP score, such as dehydration, disturbance of consciousness, and hypotension, are also the components of the PSI and CURB65, which is a prognostic indicator estimated by ve factors (confusion, urea, respiratory rate, blood pressure, age) and developed by the British Thoracic Society [26].
Dehydration, respiratory failure, disturbance of consciousness, and hypotension were indicative of an overall poor condition in patients. The multivariate analysis regarding the relationship between the A-DROP components (age, dehydration, respiratory failure, disturbance of consciousness, hypotension) and 30-day mortality demonstrated that respiratory failure and disturbance of consciousness were signi cantly independently associated with the 30-day mortality of non-HIV PcP patients in the present study. Patients with non-HIV PcP and an A-DROP score of ≥ 2, or those with respiratory failure or disturbance of consciousness at treatment initiation, usually demonstrate poor prognosis.
Patients with pneumonia are often dehydrated, resulting in increased reabsorption of urea by the kidneys.
Thus, an increase in BUN levels is frequently observed in patients with pneumonia [27]. Patients with CAP and increased BUN levels had higher mortality rates [13,16]. Increased BUN levels are indicative of renal hypoperfusion and thus help to predict the severity of pneumonia [28]. Increased BUN levels were also found to be a poor prognostic factor in patients with non-HIV PcP [8].
Alb is synthesized only in the liver, and the rate of Alb synthesis varies with nutritional status [29]. The rate of Alb synthesis changes markedly in severe illness, and a sustained in ammatory response may lead to long-term inhibition of Alb synthesis [29]. Serum Alb levels are reduced during acute and chronic in ammatory processes because of inhibition of Alb synthesis in the liver [30]. Hypoalbuminemia is thought to re ect systemic in ammatory reactions and malnutrition. Hypoalbuminemia in patients with non-HIV PcP might be associated with acute in ammation caused by PcP and chronic in ammation caused by underlying diseases. Low serum Alb levels reduce plasma colloid osmotic pressure, thereby increasing pulmonary vascular permeability [31][32][33]. Hypoalbuminemia was reported to be associated with increased pulmonary vascular permeability and lung injury in critically ill patients [34]. Thus, taking all these points into consideration, hypoalbuminemia can be seen as being associated with poor prognosis in patients with non-HIV PcP, by the same mechanism.
Both high BUN and low serum Alb levels were poor prognostic factors in patients with non-HIV PcP [8,10].
This study further revealed that the non-survivors had signi cantly higher levels of BUN and lower levels of serum Alb than those in the survivors. In addition, high BUN and low serum Alb levels were poor prognostic factors for patients with CAP [13][14][15][16], and the BUN/Alb ratio was useful in predicting the mortality of patients with CAP [19,35]. Moreover, the BUN/Alb ratio was also a useful marker for predicting the mortality in patients with HAP and aspiration pneumonia [17,18]. These ndings are consistent with the present study, which revealed that the increased BUN/Alb ratio before treatment was