Development and validation of a real-time PCR assay for the detection of clinical acanthamoebae

Background Suboptimal agreement between molecular assays for the detection of Acanthamoeba spp. in clinical specimens has been demonstrated, and poor assay sensitivity directly imperils the vision of those affected by amoebic keratitis (AK) through delayed diagnosis. We sought to develop and validate a single Taqman real time PCR assay targeting the Acanthamoeba 18S rRNA gene that could be used to enhance sensitivity and specificity when paired with reference assays. Methods Biobanked DNA from surplus delinked AK clinical specimens and 10 ATCC strains of Acanthamoeba was extracted. Sequence alignment of 66 18S rRNA regions from 12 species of Acanthamoeba known to cause keratitis informed design of a new TaqMan primer set. Performance of the new assay was compared to the 2 assays used currently in our laboratory. Results Among 24 Acanthamoeba-positive and 83 negative specimens by the CDC reference standard, performance characteristics of the newly designed primer set were as follows: sensitivity 100%, specificity 94%, PPV 82.8%, and NPV 100%. Compared to culture, sensitivity of the new primer set was 100%, and specificity 96%. No cross-reactivity of the primer set to non-acanthamoebae, including Balamuthia and Naegleria, was found. Conclusions We have validated a real time PCR assay for the diagnosis of AK, and in doing so, have overcome important barriers to rapid and sensitive detection of acanthamoebae, including limited sensitivity and specificity of commonly used assays. Electronic supplementary material The online version of this article (doi:10.1186/s13104-017-2666-x) contains supplementary material, which is available to authorized users.


Background
Species of the genus Acanthamoeba are known to cause amoebic keratitis (AK) among contact lens users [1]. AK due to environmental acanthamoebae manifests as corneal ulceration, which can lead to blindness if not treated in a timely and usually prolonged manner [1]. Other more easily detectable corneal pathogens, such as herpes simplex virus, cause a similar type of keratitis, and empiric treatment for these more common organisms leads to a delay in diagnosis of AK, which has a direct negative impact on patient outcomes. Increasing prevalence of AK coinciding the marketing and distribution of extendedor continuous-wear contact lenses [2] necessitates use of a high-performance detection method for known clinical and environmental strains of acanthamoebae [3].
Traditional diagnosis of AK has relied on historic parasitologic techniques such as culture, but while highly specific, this method is insensitive, inefficient, labor intensive, and requires technical expertise for interpretation. Molecular diagnostic techniques, such as polymerase chain reaction (PCR) performed directly on clinical specimens such as corneal scrapings and contact lens casings, are increasingly preferred over culture as they provide faster turnaround time (hours vs days) and eliminate the need for skilled microscopists. However, no single molecular assay has proven sufficiently sensitive for the range of known clinical and environmental acanthamoebae that can contaminate fresh water and contact lenses [4][5][6][7]. For clinical specimens, including corneal scrapings, contact lenses, and contact lens casings, individual PCR-based detection assays have reported sensitivities ranging from 64.3 to 93% compared to reference-level or composite diagnostics [4,5], and specificities of 88.6-100% [4,5]. Additionally, efficient clinical composite use of several assays in order to mitigate the risk of false negativity and positivity has been hindered by the validation of such assays on different diagnostic platforms: while some assays employ real time PCR (qPCR) [6,7], others are used on an end-point platform [8,9].
In our clinical reference laboratory, we routinely employ two molecular assays (the "Qvarnstrom" and "Riviere" assays; 6,7] as well as culture for the diagnosis of AK, however, this approach is time consuming, particularly when we need to arbitrate discordance with use of a third end-point PCR-based assay [5,8]. The Riviere assay [6] was designed as a molecular diagnostic test specific to both trophozoites and cysts of Acanthamoeba spp. whereas the Qvarnstrom primer set [7] was designed as part of a multiplex assay detecting three free-living amoebae known to cause granulomatous amoebic encephalitis. While the Qvarnstrom assay covers a broader range of Acanthamoeba genotypes, the Riviere assay has a more sensitive limit of detection (LOD), with one study demonstrating that Riviere was able to detect as few as 10 DNA copies/μL, compared to the Qvarnstrom LOD of 43.8 DNA copies/μL [10]. Thus, in the interest of providing the most sensitive diagnostic approach for a disease caused by many species within the genus Acanthamoeba and one which typically produces an extremely low-volume clinical specimen, we have paired the two assays with culture.
We aimed to design a new real time PCR assay using a novel Taqman primer-probe set combining the excellent LOD of the Riviere assay with the species breadth of the Qvarnstrom assay in order to optimize our AK diagnostic workflow and service to our patient population.

Specimens
Delinked, surplus biobanked corneal scrapings and contact lenses and solutions stored at −80 °C after being processed for detection of acanthamoebae at the Public Health Ontario Laboratory between January 2012 and May of 2015 were identified and retrieved. In addition 10 ATCC strains of Acanthamoeba, as listed in Table 1, were also used for the validation. ATCC stains were cultured according to the recommended specifications, in peptone-yeast-glucose (PYG) medium [5].
Primer/probe design 66 18S DNA sequences from Acanthamoeba spp. were obtained from NCBI. These sequences include the same ones used by both Riviere [6] and Qvarnstrom [7] in the development of their primer sets, as well as 25 new sequences included in order to have at least one representative sequence for each species known to cause AK. Sequences were aligned using MEGA 6 software [11]. Aligned sequences were scanned visually for regions with a high occurrence of conserved nucleotides, and this region (bases 1946-2072 on the reference 18S rRNA sequence from the National Center for Biotechnology Information (NCBI), Accession Number AF019056) was used for Taqman MGB primer and probe design using Primer Express 3.0 software (Applied Biosystems, Foster City, CA). Primers and probe were subject to 2 BLAST searches each as a preliminary cross-reactivity check, one with Acanthamoeba species excluded, and one including only the human genome. Primers were also assayed for hairpin formation and degree of primer-dimer formation (Applied Biosystems, Foster City, CA).

DNA extraction
DNA was extracted using the Qiagen Mini DNA extraction kit (Qiagen, Hilden, Germany. Samples were resuspended in PBS as needed, and subjected to three rounds of freeze-thaw in liquid nitrogen in order to lyse cysts before undergoing the manufacturer's protocol. Elution was in 60 μL elution buffer to obtain more concentrated samples.

DNA amplification
Each surplus clinical and ATCC sample was subject to qPCR using the newly designed primer and probe set, as well as the Riviere [6] and Qvarnstrom [7] primer sets, which served as our reference assays. The new primers used were AcanthF3 ( [7]. TaqMan universal master mix (Ther-moFisher Scientific, Cat# 4304437, Waltham, MA) was used according to manufacturer's instructions, and 5 μl of DNA, in a total volume of 25 μl were run with the following cycling conditions: 50 °C for 2 min, 95 °C for 10 min, and 45 cycles of 95 °C for 15 s and then 60 °C (Riviere assay) or 55 °C (Qvarnstrom assay) for 1 min. All qPCR assays were run on an ABI 7500 real time PCR instrument (Applied Biosystems, Foster City, CA). Specimens were considered positive if the cycle threshold (Ct) value was <40 in the presence of a logarithmic amplification curve.

Limit of detection determination
Acanthamoeba strains ATCC50373 and ATCC50739 were used to determine the limit of detection (LOD) of the new assay through serial dilutions of the DNA, from which a standard curve was generated. This standard curve was also used to calculate the efficiency of the Table 1 Reference standard positive ATCC specimens used in the validation of the new primer set, and species identification of positive clinical specimens a Reference standard assays were Riviere [6] and Qvarnstrom [7], and a specimen was considered positive if either assay produced a logarithmic amplification curve with cycle threshold <40 on real time PCR  (Table 2) in order to confirm lack of cross reactivity.

Control human DNA
An important internal control used in our detection assay is the detection of human B2MG after spiking of samples post-extraction with human DNA to eliminate the possibility that a negative test result is due to local assay failure or PCR inhibitors. In order to determine if this spiking had an effect on the Ct value given by Acanthamoebaspecific primer sets, the 10 ATCC strains used in this validation were subject to qPCR as described above with the Riviere primer set both with and without human DNA added. The same was done with the new primer set. Due to limited specimen volumes, ATCC rather than clinical specimens were used for this experiment.

Sequencing
Species identification of all positive clinical samples was performed in order to determine if the newly designed assay was sensitive enough to detect DNA from all represented species sent to our reference laboratory. Amplification of DNA and sequencing was performed on each sample using three different primers. The Nelson primers [9], JDP primers [8] (JDP1 and JDP2), and a newly developed set of end point PCR primers made specifically for sequencing of the 18S rRNA region of Acanthamoeba were used. The newly designed sequencing primer set was developed to bind to a conserved region for all of the different species and to allow the amplicon to span as much of the gene as possible, in order to sequence across the heterogeneous regions to differentiate species. The AcanSeqF (5′-CCT ACC ATG GTC GTA ACG GG-3; TM = 64.5 °C) and AcanSeqR (5′-AGG GCA GGG ACG TAA TCA AC-3′, TM = 62.5 °C) primers were designed for this purpose. Sanger sequencing was performed on these samples using BigDye v3.1 (ThermoFisher Scientific, Waltham, MA) and a 3130 × l genetic analyzer according to the standard protocol. Gathering data obtained from all 3 primer sets and aligning them with Contig express (Invitrogen, Carlsbad, CA) enabled greater confidence in the bases identified, and allowed improved confidence in species identification. Obtained sequences were subject to a BLAST search in order to determine the species with the highest homology to the specimen.

Statistical analysis
Data were managed in a password-protected Microsoft Excel file (Redmond, WA). Primary outcome measures were sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), and were calculated in the standard manner using a reference standard of the Qvarnstrom (CDC) assay. Confidence intervals for these performance metrics were determined using Vassarstats Clinical Calculator 1 (http://vassarstats. net/clin1.html). Secondary outcome measures were performance characteristics compared to the composite reference standard of either Qvarnstrom [7] or Riviere [6] assay positivity, as well as compared to clinical culture.
Mean Ct values were compared by t test for both the new assay and Riviere with and without addition of human control DNA using GraphPad Prism v. 7.0 (GraphPad Inc., La Jolla, CA).

Results
Between January 2012 and May 2015, 97 surplus specimens submitted for Acanthamoeba diagnostics were biobanked as follows: corneal scrapings (n = 87), contact lenses in saline (n = 4) or contact lens solution (n = 6). Twenty-one clinical specimens (21.6%) were known to be  1, 2). Both the new assay and Qvarnstrom assay detected all acanthamoebae from ATCC and clinical specimens that were culture positive, unlike the Riviere assay, which failed to detect acanthamoebae in 2 ATCC and clinical specimens that were culture positive.

Table 3 Performance characteristics of the new assay compared to the Qvarnstrom assay [7]
By Qvarnstrom, 24 specimens were considered true positive, and 83 were considered true negative

Limit of detection
LOD of the new assay was calculated to be 7 copies/μL using ATCC50739, with an efficiency of 89.2%. Using ATCC50373, the strain used as a positive control for all experimentation, a LOD of 11 copies/μL was obtained, with an efficiency of 130%. Standard curves constructed from serial dilutions of both strains are shown in Additional file 1: Figure S1.

Assay cross-reactivity
BLAST searches using the primers and probe as targets showed only 18S rRNA of Acanthamoeba when left unfiltered; excluding Acanthamoeba or including only the human genome revealed no possibility for cross-reactivity (data not shown). Cross-checking of the primer set with 13 other microbial species revealed lack of crossreactivity with any other amoeba, protist, or microbe typically found in the eye, or bacterium found to be a natural endosymbiont of Acanthamoeba (Table 2).

Human DNA
The effect of adding human DNA to primary clinical samples as a local assay control was analyzed as described above. On average, Ct values did not change with addition of human DNA when the Riviere primer set was used (average increase = 0.1, p = 0.774). However, Ct values increased by 3.8 with the new primer set (p = 0.0013) (Additional file 1: Figure S2).

Species identification
Of 21 positive surplus clinical specimens included in this validation, we were able to species identify 14 (66.7%) as follows: A. castellanii (n = 6), A. hatchetti (n = 2), A. polyphaga (n = 2), A. lenticulata (n = 2), A. quina (n = 1), and A. griffini (n = 1). The remaining 8 reference positive specimens could not be species identified using our sequencing method. Specimens that could not be species identified all showed higher Ct values by both the Qvarnstrom and Riviere primer sets (data not shown), potentially indicating less DNA in the clinical sample.

Discussion
AK is a potentially blinding ocular infection caused by environmentally ubiquitous acanthamoebae, whose varied adaptations to specific niches and opportunistic proclivities, as well as genetic heterogeneity challenge the development of both highly sensitive and specific molecular assays for their clinical detection. We have previously demonstrated that the Riviere assay, which has an excellent LOD, failed to detect specific clinically relevant strains of Acanthamoeba [5]. Due to the typically minuscule volume of specimen attainable from the cornea of patients with AK, any molecular assay implemented clinically for acanthamoebae detection must demonstrate, in addition to broad species sensitivity, a very low LOD. We developed a Taqman-based assay for the molecular detection of acanthamoebae from common clinical specimens such as corneal scrapings and contact lenses with the aim of combining excellent species coverage with analytic sensitivity.

Targeting the 18S rRNA gene is both sensitive and specific
The 18S rRNA gene is commonly used for developing primers for both detection and sequencing of protists because of its high conservation across the target genus, and its regions of variation that can be used to distinguish species. Furthermore the 18S region is part of a ribosomal tandem repeat segment, meaning it is present in multiple copies across the genome. Some estimates place the 18S copy number per organism for Acanthamoeba at around 60 [12]. Mathers and colleagues [9] designed a PCR primer set ("Nelson") based on the 18S rRNA region of the genome to detect Acanthamoeba in clinical samples, while Schroeder and colleagues [8] developed a second primer set, the "JDP" set, that amplified a larger region of the 18S rRNA gene of Acanthamoeba.
We have previously demonstrated that, using clinical corneal specimens, the Nelson primer set was more sensitive (90% vs 65%), whereas the JDP primer set was more specific (100% vs 90.8%) [4]. Thus, neither assay could stand alone for the diagnosis of AK in a clinical laboratory. The Riviere primer set [6] was designed by aligning sequences of 6 different 18S rRNA genomes, thus, while sensitive from a LOD perspective, the assay appears to be unable to detect the broad range of Acanthamoeba species and genotypes known to cause AK, as previously demonstrated [5,10,13].
The new primer set, based on alignment of 66 different 18S Acanthamoeba sequences, demonstrated sensitivity of 100% compared to reference assays such as culture and the CDC's Qvarnstrom [7] assay, with specificities of 96 and 94%, respectively. Although the increase in Ct values with addition of human control DNA for the new primer set could potentially lead to false negativity, it should be noted that even with a 4-cycle increase in Ct value, all positive specimens remained in the detectable range (<40 cutoff ). Species identification was possible for just over 60% of clinical specimens containing acanthamoebae. The new assay demonstrated broad species detection, but was unable to detect a strain of A. griffini and A. castellani, which were detected only by the Riviere assay. Conversely, the new assay detected Acanthamoeba DNA in both clinical and ATCC strains of A. culbertsoni and A. lenticulata, both of which are known to produce false negative results using the Riviere assay alone [5,10,13].