Short Communication: Diagnosis of Pneumocystis jirovecii Pneumonia by Detection of DNA in Blood and Oropharyngeal Wash,Compared with Sputum

Introduction

P neumocystis jirovecii pneumonia (PCP) remains one of the commonest opportunistic infections seen among HIV-positive individuals in developed nations ^1,2 and is increasingly recognized in resource-limited settings. ³ Gold standard diagnosis of PCP has been by immunofluorescence microscopy (IF) on specimens obtained by broncho-alveolar lavage (BAL). ⁴ However, a wide variety of nucleic acid amplification techniques (NAATs) have been used on various specimens ^5,6 in an attempt to improve use and reduce invasiveness of tests, and these are standard-of-care in many HIV centers. Nested polymerase chain reaction (PCR) on BAL and induced sputum is sensitive (96%) but less specific (59%) compared with IF, ⁷ is still technically demanding to obtain, and is limited to those individuals well enough to tolerate the procedure in centers with such facilities. Oropharyngeal secretions may have use in the right settings where the right NAAT is used, ⁵ but concerns remain about distinguishing disease from colonization. Detection of PCP in blood would be technically easy to obtain, has a rapid turnaround time, and conceptually is more likely to indicate disease, but there are conflicting data on its use. ^8,9

We conducted a prospective study at a regional infectious diseases unit to compare the sensitivity and specificity of PCR on oropharyngeal wash (OPW) and blood specimens compared with sputum to find faster, simpler, and less-invasive methods of diagnosing PCP.

Methods, Results, and Discussion

This study recruited consenting adults presenting to the regional infectious diseases unit and were clinically assessed as possibly having PCP by the admitting physician. Individuals would routinely be assessed by history and clinical examination, in addition to pulse oximetry and chest X-ray. The principle inclusion criteria were that the individual was being investigated for PCP as part of routine care and able to consent immediately or retrospectively. We included a spectrum of patients, including those with high and low pretest probability of disease to estimate sensitivity and specificity. Participants provided sputum (spontaneous or induced), OPW, (10 mL normal saline used to gargle for 10–30 s and collected in a sterile universal container), and blood specimens (3.5 mL in EDTA) for PCR testing for PCP. Demographic and clinical data, and all available PCR results, were collected using a standardized case report form. PCP diagnosis at this unit was by PCR on induced or spontaneously produced sputum as an established standard of care.

A 300 μL volume of each sample was extracted using a Qiagen MDx system and eluted into 100 μL Qiagen elution buffer. PCP PCR was carried out by real-time PCR targeting the mitochondrial large subunit ribosomal RNA gene. The internal positive control (IPC) is a synthetic DNA oligonucleotide added to the clinical sample. Primer and probe sequences are shown in Table 1. A 5 μL volume of extracted clinical sample was added to the PCR mixture, which contained 12.5 μL of Environmental Master Mix 2.0 (Life technologies cat no 4396838) and 0.3 μM of the PCP forward and reverse primers, 0.2 μM of the PCP Taqman probe, and 0.1 μM of each of the IPC primers and probe. Following an activation step of 10 min at 95°C, the PCRs were amplified at 95°C for 15 s followed by 1 min at 60°C for 45 cycles. The IPC was manipulated to become positive at around PCR cycle threshold of 28. Samples were reported as inhibitory where the IPC signal was delayed by three-cycle thresholds or more.

OPW and blood PCR results were compared with sputum PCR results to calculate relative sensitivity, specificity, and positive and negative predictive values against the standard of care (PCP PCR on sputum), with 95% confidence intervals calculated using binomial exact methods. Analysis was done using STATA v12 (STATA Corps, College Station, TX).

This study was approved by the National Health Service South Manchester Research Ethics Committee. All participants provided written consent.

Forty-five participants were included, 41 (91%) of whom were male, 38 (84%) Caucasian, with median age of 39 years (range 26–60 years). One participant was an HIV-negative renal transplant recipient. Forty-four of the 45 participants were HIV positive with median CD4 count of 64 cells/μL (IQR 15, 160 [n = 43 with CD4 count available]). Nine of 44 participants (20%) were prescribed antiretroviral therapy (ART) at the time of recruitment, 4 of whom had undetectable HIV RNA. Of those not on ART, median HIV RNA was 164,550 copies/mL. Ten individuals had a prior history of PCP and nine reported taking PCP prophylaxis. Thirty-nine individuals had started empirical treatment for PCP a median of 2 days before (range 11 days before–2 days after) samples were taken. At study recruitment, other respiratory diseases present were previous community-acquired pneumonia, asthma, bronchitis, and pleural tuberculosis (one patient each). There were no problems with acceptability to patients of providing OPW or blood for diagnostic tests.

For 45 participants, laboratory results were available for 45 sputa, 31 OPW, and 41 blood samples. Sputum PCR was positive in 27/45 (60%) participants. PCR on OPW was positive in 9/19 individuals with both positive sputum PCRs and both results available, giving sensitivity 47% (95% confidence interval [CI] 23–71). Among 24 individuals with positive sputum PCR and both results available, blood PCR was positive in 12, giving sensitivity 50% (95% CI 29–71) (Table 2).

Sensitivity of combined OPW and blood testing, among 27 individuals with at least one of OPW and blood results available, was 67% (95% CI 49–85) (18/27). Of those 27 individuals, 24 had blood results available (12/24 PCR positive) and 19 had OPW results (9/19 PCR positive). Sixteen individuals had results available for both OPW and blood, results of which are shown in Table 2.

There were no false positive OPW or blood tests, so specificity compared with sputum PCR of both was 100%, with lower limit of 95% confidence interval, 74% for OPW and 80% for blood. Negative predictive value of OPW was 57% (95% CI 35–78) and that of blood was 59% (95% CI 40–77) (Table 2).

Including only patients with samples obtained no later than 2 days after start of treatment (n = 26), sensitivity of OPW was 80% (8/10) (95% CI 51–100) and that of blood was 57% (8/14) (95% CI 29–86). Including 16 individuals with results for sputum and OPW and/or blood within 2 days of starting treatment, sensitivity of combined OPW and blood was 88% (14/16) (95% CI 70–100). A positive PCP diagnosis was obtained in 14 of 16 individuals with positive sputum PCR, using nonrespiratory, noninvasive specimens.

This study showed high specificity of PCR for PCP on OPW and blood compared with sputum specimens, and a high sensitivity of PCR on early OPW and blood, with the best performance being to combine an OPW with blood specimen early in the course of clinical care (sensitivity 88%). Adopting this as a diagnostic approach would have allowed PCP diagnosis in those with disease as defined by positive sputum PCR in 14/16 individuals without recourse to invasive procedures. There were no positive OPW samples that were not matched with a positive sputum result, suggesting that colonization does not interfere with the specificity of these specimens compared with sputum.

A considerable amount of work has been undertaken previously on the use of various NAATs in the diagnosis of PCP on noninvasively obtained specimens in patients with and without HIV, but controversy remains. Common concerns raised are reduced sensitivity on noninvasive specimens and differentiating colonization from disease. ⁶ These issues can be largely addressed by using the correct PCR techniques ⁵ and appropriate clinical assessment of patients. Samples taken earlier in empirical treatment showed greater sensitivity than those taken later, so obtaining samples promptly could avoid invasive procedures or diagnostic uncertainty, particularly with the high specificity seen here.

The testing platforms and extraction methodologies required are often used for PCR analyses for other pathogens in this patient group, for example, PCR for cytomegalovirus, so little additional cost would be expected. In future work, it would be of interest to explore how this technique can be combined with other methodologies such as β-d-glucan testing on blood ¹⁰ and to test clinical diagnostic algorithms.

The main limitations of this preliminary study are the small number of participants and the potential selection bias associated with the need for written consent, thus excluding the very unwell, who may have higher organism load. Samples for each individual were not all taken on the same day because of operational constraints, thus reducing the number of samples included in analysis of early sampling, and laboratory results were unavailable for a few tests. The comparison with largely spontaneously produced sputum leaves unanswered the issue of whether any positive results on sputum represented colonization rather than disease, because previous work comparing induced and spontaneous sputum samples has given variable results, ^11,12 and a study comparing with invasive methods of diagnosis would be required to answer that question. Although a modest sample size, this pragmatic study of a real-life clinical patient population provides useful data on the issue of PCP diagnosis.

In conclusion, molecular tests for PCP on nonrespiratory specimens show diagnostic use and great potential to save on cost, time, and invasive procedures while providing excellent specificity and good sensitivity.