The Clinical Utility of Serial Procalcitonin and Procalcitonin Clearance in Predicting the ‎Outcome of COVID-19 Egyptian Patients

INTRODUCTION

Coronavirus disease 2019 (COVID-19) has rapidly spread worldwide, and now it is considered a pandemic that represents an urgent threat to global health [1]. In severe cases, patients may have dyspnea and hypoxia that can rapidly progress to respiratory distress, multi-organ failure, and even death [2]. The mortality of COVID-19 pneumonia is higher than other viral pneumonia [3]. Since inflammation is an important factor in COVID-19 mortality, sensitive inflammatory biomarkers that reflect lung lesion changes should be continuously explored [1].

Procalcitonin (PCT) is a calcitonin-related pro-hormone released by the thyroid parafollicular cells. Under physiological conditions, serum PCT is usually below 0.05 ng/ml. However, its levels increase signi-ficantly 2–6 hours after stimulation by microbial infection, being released by all parenchymal tissues under the effect of pro-inflammatory cytokines [4]. So, it is a prognostic sign of sepsis [5].

Our study aimed to explore the ability of serial serum concentrations and clearance of PCT to ‎predict the prognosis and mortality of COVID-19 patients. ‎

PATIENTS AND METHODS

Study design and patient selection: This prospective observational study included 63 adults (age ≥ 18 years) with COVID-19 admitted to COVID-19 isolation Hospitals of Ain-Shams University, from April 10 to May 15, 2021. All included patients needed to have 1) laboratory-confirmed SARS-CoV-2 infection by reverse transcription- PCR test from a nasopharyngeal swab prior to hospitalization; and 2) at least one PCT measurement upon hospital admission (or within 24 hours) in their medical records. Pregnant women and patients with secondary bacterial infection, indicated by neutrophilia and positive bacterial respiratory cultures within the first 48 hours of hospital admission, or non-infectious causes of systemic inflammation that can induce PCT production, such as trauma, surgery, burn injury, and chronic kidney disease, were excluded. In this study, patients were followed up and were classified according to their severity, treated, and discharged according to the COVID-19 management protocol issued by Ain Shams University Hospitals [6].

Data collection: For all included patients, baseline and follow-up data were collected from medical records and encompassed demographics, clinical history, presence of comorbidities, routine laboratory test findings, length of hospital stay, ICU admission, and clinical outcome.

Procalcitonin (PCT) measurement: Because of the difficult access to some patients with critical conditions, in addition to the death or hospital discharge of others, not all initially included patients were subjected to serial PCT evaluation. Initial PCT values were obtained from the medical records of all included patients (n=63). Then, further assessment of PCT levels was done using serum samples that were collected on hospital days 3 (n = 47) and 5 (n = 49), and were stored at −80°C until analysis, using the Elecsys® BRAHMS PCT sandwich immunoassay principle of the electrochemiluminescence autoanalyzer (COBAS e411; Roche Diagnostics, Germany) with the analytical measuring range of 0.02-100 ng/ml and the detection limit of < 0.02 ng/ml.Initial PCT that was done upon hospital admission or within 24 hours and obtained from patients’ medical records was defined as PCTD1, while PCT values that were measured on day 3 (48-72 hours) and day 5 (96 – 120 hours) were defined as PCTD3 and PCTD5, respectively.

Procalcitonin clearance (PCT-c) calculation: The clearance of procalcitonin (PCT-c) on days 3 (PCT-cD3) and 5 (PCTc-D5) was calculated as follows: PCT-c D3or5 (%) = [(PCTD1 – PCTD3or5)/PCTD1] × 100. Clearance values are positive with decreasing concentrations and negative with increasing concentrations [7].

Statistical methods: The IBM SPSS statistics program (V. 26.0, IBM Corp., USA, 2019) was used for data analysis. Quantitative non-parametric measures were expressed as medians and percentiles. In addition, categorized data were expressed in the form of numbers and percentages. Comparison between two independent groups was done using the Wilcoxon Rank Sum test. Meanwhile, the Chi-square test was used to compare the categorized data. A diagnostic validity test was, and the area under the curve (AUC) was calculated after the receiver operating characteristic (ROC) curve was constructed to determine each variable’s most sensitive and specific predictive cutoff point. All statistical tests were two-tailed, and the probability of error at < 0.05 was considered significant. Data points that were missing were not extrapolated.

RESULTS:

Demographic and clinical characteristics:

During the enrollment period of this study, a final sample of 63 patients, who met the inclusion criteria, was initially obtained. The median age was 55 years (IQR:43– 70), and 54.0% of patients were men. Hypertension (38.1%) and diabetes (30.2%) were the most common comorbidities. Regarding admission disease severity, 50.8% (n=32) of the included patients had a non-severe disease, while 49.2% (n=31) were severe. The median length of stay in the hospital was nine days (IQR: 7 -12) and ranged from 4 to 30 days. In total, 29 patients (46.0%) were admitted to the ICU, and 18 (28.6%) died.

The characteristics of non-survivors were similar to survivors regarding age, gender, length of hospital stay, and underlying comorbidities, except for diabetes mellitus, which was significantly associated with in-hospital mortality (P = 0.001). Similarly, all non-survivors (n = 18) had severe disease, and 94.4% (n = 17) were admitted to the ICU. In comparison, only 28.9% (n = 13) of survivors were considered severe, and 26.7% (n = 12) required ICU admission. The demographic and clinical characteristics of the included patients are presented in table 1.

Prognostic value of serial PCT and PCT-c:

Table 2 presents the values of PCT and PCT-c with their serial changes in survivors and non-survivors. Serum levels of PCT increased significantly with in-hospital mortality. Compared with survivors, non-survivors showed significantly higher values of PCTD1 (median: 1.20 ng/ml (IQR:0.23 – 1.96) vs. 0.12 ng/ml (0.06 – 0.34); P ≤ 0.001), PCTD3 (median: 1.6 ng/ml (0.99 – 2.2) vs. 0.08 ng/ml (0.06 – 0.14); P ≤ 0.001) and PCTD5 (median: 3.2 ng/ml (IQR: 2.05 – 7.1) vs. 0.05 ng/ml (0.04 – 0.08); P ≤ 0.001).

As regards PCT-c, it decreased significantly with in-hospital mortality. Values of PCT-cD3 and PCT-cD5 were significantly lower in non-survivors compared to survivors, (median: 50.6% (IQR: 769.15 – 19.53) vs. 25% (13.39 – 42.82); P = 0.003 and median: ˗ 325.7% (IQR: 1919.61 – ˗ 96.13) vs. 51.56 % (34.37 – 78.95); P ≤ 0.001, respectively).

In non-survivors and during the follow-up, serial PCT levels showed an overall progressive increase. In contrast, PCT-c values progressively decreased. Figures 1 and 2

Table 3 demonstrates the diagnostic validity and ROC curve analysis of the ability of serial PCT and PCT-c to predict in-hospital mortality of COVID-19 patients. Regarding the single indicators, serial PCT and PCT-c showed an overall better predictive performance for in-hospital mortality than the initial PCT measurement. At the optimum cutoff values of 0.80 ng/ml, 0.54 ng/ml, 0.18 ng/ml, - 11.1 % and -79.77% for PCTD1, PCTD3, PCTD5, PCT-cD3 and PCT-cD5, respectively, the AUC followed the order of PCTD5 > PCT-c D5 > PCTD3 > PCTD1 > PCT-c D3 being 0.96, 0.89, 0.82, 0.81 and 0.74, respectively. Despite PCTD1 had the highest diagnostic specificity (97.8%), it showed the lowest diagnostic sensitivity (61.1%). PCTD5 then PCTD3 had the highest diagnostic sensitivity (100% and 85.7%, respectively), NPV (100% and 94.1%, respectively) and diagnostic accuracy (95.9% and 93.6%, respectively). The best PPV was of PCT-c D5 then PCTD3 (93.3% and 92.3%, respectively). PCT-c D3 showed the lowest specificity (87.9%), PPV (71.4%) and accuracy (83.0%).

The AUC increased when PCTD1 at a cut off of 0.8 ng/ml was added to either PCTD3 at 0.03 ng/ml, PCTD5 at 0.398 ng/ml, PCT-cD3 at-50%, or PCT-cD5 at 54.54% to be 0.93, 0.98, 0.96, and 1.00, respectively, showing that the combined prediction value exceeded the initial PCT prediction value, whereas, the best predictive performance was seen in the PCTD1+PCTD-c5 combination, having 100% specificity, sensitivity, PPV, NPV, and accuracy in predicting COVID-19 patients’ in-hospital mortality.

Table (3):   Diagnostic validity test and receiver operating characteristic (ROC) analysis of the ability of serial PCT measurements and PCT-c to predict in-hospital mortality of COVID-19 patients.

PPV: positive predictive value; NPV: negative predictive value; AUC: area under curve; PCT: procalcitonin; PCT-c: procalcitonin clearance.

DISCUSSION