Study participants and oversight.
From October 2016 to September 2017 we recruited newborn babies of at least 34 weeks post-menstrual age undergoing lumbar puncture to exclude meningitis on either the postnatal wards or on the neonatal unit at the Rosie Hospital, Cambridge University Hospitals NHS Trust. Informed consent was obtained before CSF was analysed by PCRctic. UK National Research Ethics Committee and UK Health Research Authority (HRA) approved the study.
All babies recruited into the study had their CSF tested by bacteriological culture and in PCRctic. The study had two phases. In the first, at least five drops of CSF were collected into each of three sterile universal containers (ISS Ltd., UK). CSF in two of these was tested using the standard laboratory assessment for meningitis, and, following parental consent, CSF in the third container was tested in PCRctic in the clinical microbiology laboratory, Addenbrookes Hospital, Cambridge University Hospitals NHS Trust. Whilst the 30 ml sterile universal containers are routinely used across the UK for CSF collection, special care is needed to avoid contamination through the handling of the screw-on tops. In the second phase, CSF for PCRctic was collected into 1.5 ml sterile individually sealed snap-top tubes (Eppendorf Biopur, Eppendorf AG, Germany). Lumbar puncture technique was as standard. Unisept (Molnlycke Healthcare Ltd., UK) solution of 0.05% chlorhexidine was used in the first phase of the study for skin antisepsis. This was changed to ChloraPrep (BD, UK), containing 2% chlorhexidine and 70% alcohol for the second phase of the study.
Results from the initial 39 samples suggested possible environmental contamination and the study protocol was amended to include individually sealed sterile snap-top tubes and ChloraPrep skin antisepsis (as above). The amendment was approved by HRA.
The primary outcome measure was the rate of false positive results. As this was a feasibility study, the results had no effect on patient care.
Modified 16S rDNA PCR assay (PCRctic).
PCRctic (Figure 1) uses primers against the conserved regions of bacterial ribosomal DNA. Its sensitivity and specificity derive from a single-step, closed-tube nested PCR format employing external primers (25mers) with high Tm (≈75°C) at 50nM (30 cycles) [333F25 CCAGACTCCTACGGGAGGCAGCAGT, 929R25 CCACATGCTCCACCGCTTGTGCGGG] and internal primers (19mers) with a low Tm (≈50°C) at 0.25µM (40 cycles) [800F19 TAGTCCACGCCGTAAACGA, 907R19 CCGTCAATTCATTTGAGTT]. Primers were designed against the conserved regions of the bacterial 16S rDNA gene. Briefly, 21,397 rDNA gene sequences were downloaded and aligned, and a simple script was used to identify conserved portions. From these portions, primers were designed to have a Tm (calculated as 2x[A+T]+4x[G+C]) of 60-70°C (outer primers) or 48-52°C (inner primers).
Contamination from free bacterial DNA was eliminated by the use of ethidium bromide monoazide (EtA) . Exposure of the reaction tube to light (530nm) for one minute causes EtA to react covalently with DNA, rendering it non-amplifiable (Figure 2). The same illumination also destroys any residual EtA. Importantly, EtA does not penetrate intact bacteria and can therefore be used in the presence of the target (intact) bacteria before these are lysed by heat in the first PCR cycle. Typically, 180µl of CSF were transferred into 0.2ml PCR tubes and pelleted in a microfuge (Eppendorf: 5424, Rotor: FA-45-24-11) for 2min at 20.000g. After carefully removing the supernatant with a sterilized gel-loading tip, 10 µl of a previously-frozen PCR mastermix (80µl KOD 10x buffer, 80µl 10x dNTPs, 64µl 25mM MgSO4, 8µl KOD HotStart enzyme, 40µl 20x primer mix, 510µl HPLC-grade water, 16µl 10x SYBR Green, 1.6µl 25µM ROX, 1.25µl 2.4 mM EtA) were added. The closed PCR tubes were then illuminated as above to photoactivate EtA. The samples were then amplified on a quantitative PCR (qPCR) system (Applied Biosytems ViiA 7; 95°C x 3min, then 30 cycles of: 94°C x 10s, 70°C x 20s, 72°C x 30s; then 40 cycles of 94°C x 10s, 50°C x 20s, 72°C x 30s) followed by a melting curve analysis.
Negative controls tested the mastermix in empty tubes, positive controls used spiked mastermix under the same experimental conditions (all done using the same hood, equipment and environment). Bacterial strains used were E. coli (DH5-alpha and ATCC 25922), S. aureus (ATCC 29213), S. agalactiae (NCTC 8181) or L. monocytogenes (NCTC 7973). In CSF samples spiked with bacteria, PCRctic reliably detected as few as 1.5 CFUs (Figure 3A&B and Figure 4). Positives gave a single melt at 86-90°C, in negatives primer dimers gave a single (72-75°C) or double peak between 72-82°C (not shown), which served as an internal control. For confirmation, the samples were run on 3% agarose minigels detecting the expected band of about 126bp (size varied slightly depending on bacterial species). Positives were purified from gels and sequenced. Where possible, mixed samples were re-cloned into pBluescript vector (Agilent Technologies, USA) and individually sequenced.
A nonparametric Mann-Whitney U test was used for the significance of the difference between the studied groups. Bayesian statistics were used to estimate the significance of the positive results under different prevalence, specificity, and sensitivity conditions.