JCM Accepts, published online ahead of print on 26 March 2014 J. Clin. Microbiol. doi:10.1128/JCM.00341-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
A Multiplex Nucleic Acid Amplification Test for the Diagnosis of Dengue, Malaria, and
2
Leptospirosis
3 4
Running Title: Multiplex NAAT for Dengue, Malaria, and Leptospirosis
5 6
Jesse J. Waggonera, Janaki Abeynayakeb,*, Ilana Balassianoc, Martina Lefterovab, Malaya K.
7
Sahoob, Yuanyuan Liub, Juliana Magalhães Vital-Brazilc, Lionel Greshd, Angel Balmasedae, Eva
8
Harrisf, Niaz Banaeia,b, and Benjamin A. Pinskya,b,#
9 10 11 12 13 14
a
Department of Medicine, Division of Infectious Diseases and Geographic Medicine, Stanford
University School of Medicine, Stanford, California, United States of America b
Department of Pathology, Stanford University School of Medicine, Stanford, California, United
States of America c
Laboratório de Zoonoses Bacterianas, Centro de Referência Nacional para Leptospirose,
15
Coleção de Leptospira, WHO/PAHO Centro Colaborador para Leptospirose, Instituto Oswaldo
16
Cruz, Fiocruz, Rio de Janeiro, Brazil
17 18 19 20 21
d
Sustainable Sciences Institute, Managua, Nicaragua
e
National Virology Laboratory, Centro Nacional de Diagnóstico y Referencia, Ministry of
Health, Managua, Nicaragua f
Division of Infectious Diseases and Vaccinology, School of Public Health, University of
California, Berkeley, California, United States of America
22 23
1
24
#
To whom correspondence should be addressed.
25
3375 Hillview Ave, Room 2913
26
Palo Alto, CA 94304
27
Phone: +001 (650) 721-1896
28
Fax: +001 (650) 723-6918
29
E-mail:
[email protected] 30 31
*
Current address: Division of Clinical Virology, Teaching Hospital Karapitiya, Galle, Sri- Lanka
32
2
33
ABSTRACT
34
Dengue, leptospirosis, and malaria are among the most common etiologies of an
35
undifferentiated, systemic febrile illness (UFI) among travelers to the developing world, and
36
these pathogens all have the potential to cause a life-threatening illness in the returned traveler.
37
The current study describes the development of an internally-controlled, multiplex nucleic acid
38
amplification test for the detection of dengue virus (DENV), Leptospira, and Plasmodium
39
species, with a specific call-out for P. falciparum (referred to as the UFI assay). During
40
analytical evaluation, the UFI assay displayed a wide dynamic range and sensitive limit of
41
detection for each target, including all four DENV serotypes. In a clinical evaluation including
42
210 previously tested samples, the sensitivity of the UFI assay was 98% for DENV (58/59
43
samples detected), and 100% for Leptospira and malaria (65/65 and 20/20 samples,
44
respectively). Malaria samples included all five Plasmodium species known to cause human
45
disease. The specificity of the UFI assay was 100% when evaluated with a panel of 66 negative
46
clinical samples. Furthermore, no amplification was observed when extracted nucleic acids from
47
related pathogens were tested. Compared with whole blood, the UFI assay remained positive for
48
Plasmodium in 11 plasma samples from patients with malaria (percent parasitemia 0.0037-
49
3.4%). The syndrome-based design of the UFI assay, combined with the sensitivity of the
50
component tests, represents a significant improvement over the individual, available diagnostics
51
for these pathogens.
3
52
INTRODUCTION Health problems that necessitate medical attention are reported by eight percent of
53 54
travelers to the developing world, amounting to over four million patients annually (1-3). In a
55
study analyzing data from 17,353 patient encounters at GeoSentinel sites, the most common
56
diagnostic syndrome reported was an undifferentiated, systemic febrile illness that lacked
57
localizing signs or symptoms (hereinafter referred to as a UFI) (1). A specific etiologic
58
diagnosis was only available for 59.4% of such patients, with malaria and dengue identified as
59
the cause in 76.8% of patients who received a diagnosis (1). In a subsequent analysis of acute
60
life-threatening illnesses among 3,655 travelers, Plasmodium falciparum was the most common
61
pathogen identified and accounted for 10 of 13 deaths (4). Dengue virus (DENV) infection was
62
the most common viral etiology of a life-threatening illness (4). Though diagnosed less often,
63
leptospirosis is also a well-documented cause of severe infections in the returned traveler and
64
accounted for 2.4% of life-threatening infections in patients presenting to GeoSentinel sites (4-
65
6).
66
One limitation to such studies is the lack of a “gold-standard” diagnostic test for many of
67
the causes of a UFI (1, 7, 8). Traditional microscopy for malaria offers the clearest example of a
68
widely-used reference standard that provides results in a clinically meaningful timeframe.
69
Microscopy can provide morphologic species determination and quantitation of the parasite
70
burden (8). However, this test displays variable performance associated with the experience of
71
the reader, which is of particular concern in non-endemic regions (9). Assays used as reference
72
standards for DENV and Leptospira detection involve serological testing in acute and
73
convalescent samples, and for Leptospira, antibody detection using microagglutination testing
74
(MAT) remains the standard (7, 10). MAT requires significant expertise to perform, the
4
75
maintenance of a local reference panel of bacterial cultures, and acute and convalescent samples
76
to provide a confirmed diagnosis (10-12). For either DENV or Leptospira, the results from
77
serologic testing of a single acute specimen can only provide a presumptive diagnosis.
78
Rapid diagnostic tests (RDTs) have been developed for all of these pathogens and detect
79
specific antigens, antibodies, or both (9, 10, 13-19). DENV RDTs for the detection of non-
80
structural protein 1 (NS1) or anti-DENV antibodies are available. While these can provide a
81
diagnosis in as little as 15 minutes, they are less sensitive in acute infection than reverse-
82
transcriptase PCR (RT-PCR) (17). Rapid serological tests for leptospirosis are available, though
83
they have proven less sensitive than MAT (16). The clinical sensitivity and specificity of
84
malaria RDTs are similar to routine microscopy for P. falciparum (13, 14). These tests are less
85
sensitive for other Plasmodium species, however, resulting in part from their relatively high
86
analytical detection limits (14).
87
Many separate nucleic acid amplification tests (NAATs) have been described for the
88
detection of DENV, Leptospira, and Plasmodium species. While the reported sensitivity of
89
individual assays has ranged, NAATs have proven to be the most sensitive diagnostic test in the
90
setting of acute infection with these pathogens (8, 17, 20-23). Recently, our group reported the
91
development of an internally-controlled, real-time RT-PCR (rRT-PCR) for pan-DENV detection
92
(24). This pan-DENV assay detects all four DENV serotypes using hydrolysis probes labeled
93
with a single fluor and targets RNase P as the internal control (IC). In a study of 199 clinical
94
samples, the pan-DENV assay proved more sensitive than three comparator assays (25): a
95
widely-used hemi-nested RT-PCR, originally reported by Lanciotti, et al (26, 27); a
96
commercially-produced, internally controlled rRT-PCR; and the recently FDA-cleared CDC
97
DENV-1-4 Real-Time RT-PCR (28).
5
98 99
In this study, we describe the design and evaluation of the UFI assay, which is a singlereaction, internally-controlled multiplex NAAT for the detection of DENV 1-4 (based on the
100
pan-DENV assay), all species of Leptospira, and the five Plasmodium species known to be
101
pathogenic in humans with a specific call-out for P. falciparum. The goals of assay design were
102
to maintain the sensitivity of the pan-DENV assay for DENV detection while adding the ability
103
to detect other pathogens in the differential diagnosis. DENV, Leptospira, and Plasmodia were
104
selected for inclusion in the UFI assay given their widespread nature, potential to cause severe
105
disease, and the possibility of improving patient outcomes with early diagnosis and appropriate
106
management.
107
6
108
MATERIALS AND METHODS
109 110
Ethics. Protocols for the collection and testing of samples from Nicaraguan pediatric dengue
111
cases were reviewed and approved by the Institutional Review Boards (IRB) of the University of
112
California, Berkeley, the Nicaraguan Ministry of Health, and Stanford University. Parents or
113
legal guardians of all subjects provided written informed consent; subjects 6 years of age and
114
older provided assent. Specific IRB approval for the development of the UFI assay was not
115
required as all samples were pre-collected and de-identified.
116 117
Assay Design. The DENV and RNase P primer and probe sequences were previously published
118
(24). The UFI assay incorporates the following modifications: DENV probe concentration in the
119
final reaction mixture was decreased to 100nM; RNase P primer concentration was decreased to
120
50nM; and RNase P probe concentration was increased to 100nM. No changes were made to the
121
concentration of DENV primers.
122
The 16S rRNA (rrs) gene was selected for use in the assay to detect Leptospira species.
123
We aligned 704 publicly available Leptospira 16S sequences from the National Center for
124
Biotechnology Information (NCBI) Nucleotide database using MegAlign software (DNASTAR,
125
Madison, WI). This alignment included both pathogenic and saprophytic species of Leptospira.
126
Regions where the consensus nucleic acid sequence was conserved across ≥95% of the aligned
127
sequences were identified as potential targets for primers and probes. Using BLAST to compare
128
sequences, regions with significant homology to Staphylococcus, Listeria, Treponema, or
129
Borrelia sequences were eliminated from consideration. Primers and probes were then designed
130
using Primer-BLAST software (29).
7
131
The primers and probe for P. falciparum detection were designed against the Pfr364
132
repetitive element using RealTimeDesign software (Biosearch Technologies, Novato, CA). The
133
sequence of Pfr364 has been published previously (30). A pan-Plasmodium assay targeting the
134
18S rRNA gene was also initially designed using RealTimeDesign, then refined using 163
135
Plasmodium sequences, including sequences from P. falciparum (n=25), P. knowlesi (n=86), P.
136
malariae (n=4), P. ovale (n=23), and P. vivax (n=25). Sequence alignment and target selection
137
were performed as described for Leptospira, and final primer and probe sequences were selected
138
using Primer3 (31). The primer and probe sequences for the Leptospira, Pfr364, and
139
Plasmodium 18S assays are shown in Tables 1 and 2.
140 141
UFI Assay Performance. All reactions were performed on the Rotor-Gene Q instrument
142
(Qiagen, Germantown, MD) using 25 µL reaction volumes and the SuperScript III Platinum
143
One-Step qRT-PCR kit (Invitrogen; Carlsbad, CA). Primer and probe concentrations were
144
initially optimized in mono-plex reactions using positive control samples. Assays were then
145
combined sequentially, and primer and probe concentrations were adjusted to maintain
146
sensitivity and minimize bleed through between channels. Hydrolysis probes (Biosearch
147
Technologies, Novato, CA) were labeled to provide DENV detection in the green channel
148
(FAM), Leptospira in yellow (Cal Fluor 560), P. falciparum in orange (Cal Fluor 610),
149
Plasmodium 18S in red (Quasar 670), and RNase P in crimson (Quasar 705).
150
For the UFI assay, each reaction contained 5µL of nucleic acid template and the final
151
primer and probe concentrations shown in Tables 1 and 2. Cycling conditions were the
152
following: 52°C for 15min (RT step for DENV); 94°C for 2min; 45 cycles of 94°C for 15sec,
153
55°C for 40sec, and 68°C for 20sec. Detection was performed in all channels at 55°C; the gain
8
154
for each channel was set following optimization during initial testing (5.33 for green, 10 for all
155
other channels). During analysis, the first five cycles were cropped from each channel to
156
improve baseline normalization. Results were evaluated on the linear scale with slope
157
correction. The crossing threshold was set at 0.05 for all channels. Any exponential curve
158
crossing this threshold prior to cycle 45 was considered a positive result in the green, yellow,
159
orange, and red channels. Curves crossing after cycle 40 in the crimson channel (RNase P
160
detection) and negative in all other channels were considered IC failures.
161 162
Control Nucleic Acids and Reference Material. Quantitated, positive-sense, single-stranded
163
DNA (ssDNA) oligonucleotides containing the target sequence for each DENV serotype were
164
synthesized (Eurofins MWG Operon, Huntsville, AL) and used in the analytical characterization
165
of the UFI assay. For Leptospira 16S, Pfr364, and Plasmodium 18S, quantitated plasmid DNA
166
was utilized. Amplicons generated from mono-plex reactions for Leptospira, P. falciparum
167
(Pfr364 target), and P. vivax (18S target) were cloned and sequenced using M13 primers as
168
described previously (24, 32). Leptospira interrogans serovar Copenhageni, strain Fiocruz L1-
169
130 (ATCC Number BAA-1198D-5; ATCC, Manassas, VA) template genomic DNA was kindly
170
provided by Alexandria Boehm (Stanford University). This was used as the template for
171
Leptospira amplicon production. Extracted nucleic acids from clinical samples obtained from
172
separate patients with microscopy-confirmed P. falciparum and P. vivax were used as templates
173
for amplicon production. Extracted genomic DNA from 39 cultured isolates of Leptospira was
174
obtained for testing in the UFI assay. These included strains from 7 species and 23 different
175
serovars of Leptospira (see Table 3).
176
9
177
UFI Assay Analytical Characterization. The UFI assay was analytically characterized
178
according to published recommendations (33). For these experiments, total nucleic acids were
179
extracted and pooled from serum samples sent to the Stanford Clinical Virology Laboratory for
180
HCV testing. Pooled extract was spiked into the reaction mixture to mimic the IC in clinical
181
samples (5µL of extract per reaction). To establish the dynamic range, four replicates of serial
182
10-fold dilutions of ssDNA or plasmid DNA from 7.0 log10 copies/µL to 1 copy/µL were tested
183
on a single run. The highest concentration available for the P. falciparum plasmid was 6.5 log10
184
copies/µL. The dynamic range was established by fitting a best-fit line to the data by regression
185
analysis and included the range where the R2 value for this line was ≥ 0.99.
186
To establish the lower limit of 95% detection (95% LLOD), ten replicates of serial two-
187
fold dilutions were tested on a single run. Four dilutions were tested for each target, beginning
188
with a concentration 2-fold higher than the lowest concentration at which all replicates were
189
detectable during the dynamic range study. The 95% LLOD was then calculated using probit
190
analysis. To compare the analytical sensitivity of DENV detection in the UFI assay to the
191
original pan-DENV assay, the 95% LLOD of the pan-DENV assay was also evaluated using the
192
same quantitated ssDNA standards. The pan-DENV assay was performed as previously
193
described (24).
194 195
Specificity. The specificity of the UFI assay was evaluated by testing genomic RNA from viral
196
strains obtained from the following sources: yellow fever virus (YFV) 17D and chikungunya
197
virus (one strain each), obtained from Vircell Microbiologists (Granada, Spain); and inactivated
198
cultured isolates of YFV (Asibi strain), chikungunya virus, West Nile virus (NY 1999 strain;
199
WNV), St. Louis encephalitis virus (SLEV), and Zika virus kindly provided by the CDC Vector
10
200
Borne Diseases Branch (Fort Collins, CO). Extracted DNA from cultured strains of Salmonella
201
enterica subsp. arizonae, S. enterica serovar Typhi, Staphylococcus aureus, Pseudomonas
202
aeruginosa, and Escherichia coli (one strain each), and Mycobacterium tuberculosis (two
203
strains) was tested to evaluate the specificity of Leptospira detection. Two clinical samples
204
(EDTA whole blood) that tested positive for Babesia microti also underwent nucleic acid
205
extraction and were tested in the UFI assay. In addition, specificity was evaluated by testing 50
206
serum samples from hepatitis C virus (HCV)-positive patients that were sent to the Stanford
207
Clinical Virology Laboratory for HCV viral load testing.
208 209
DENV, Leptospira, and Plasmodium Samples. DENV detection was evaluated using extracted
210
RNA from a set of 60 serum samples collected from pediatric dengue cases in Nicaragua. For
211
serum collection, whole blood was allowed to clot in the collection tubes; serum was then stored
212
at -80°C until use. These samples tested positive by at least two different molecular assays and
213
have been described previously (24).
214
Sixty-five samples (63 serum, 2 plasma) from patients in Brazil with suspected
215
leptospirosis were tested using the UFI assay. Serum was collected in serum separator tubes, and
216
plasma was obtained from EDTA tubes. Samples, collected between April 2009 and November
217
2013, were originally sent to the Laboratório de Zoonoses Bacterianas, Instituto Oswaldo Cruz,
218
Fundação Oswaldo Cruz (Fiocruz), and were stored at -20°C until use. All samples were tested
219
with PCRs for flaB and lipL41 as described (34, 35). Amplification products were detected by
220
polyacrylamide gel electrophoresis. Samples were considered positive by conventional
221
Leptospira PCR if the expected product was detected by PCR for flaB, lipL41, or both. Samples
222
were considered negative if no product was detectable from either PCR. Fifty-five acute
11
223
specimens were also evaluated by MAT using a regionally optimized reference panel. Samples
224
with detectable antibody against any serovar (titers 400-1600) were considered positive by MAT.
225
Sixteen samples had detectable Leptospira DNA by PCR for flaB, lipL41, or both. Eight (50%)
226
of these patients had corresponding MAT results, and all were non-reactive. Forty-seven
227
(95.9%) of the 49 patients with negative PCR testing for flaB and lipL41 had corresponding
228
MAT results, of which seven were positive (14.9%). For 52 samples, PCR products from the
229
UFI assay were purified using the GeneJET PCR Purification Kit (Thermo Scientific, Waltham,
230
MA) according to the manufacturer’s recommendations. Purified amplicons were sequenced by
231
bi-directional dideoxynucleotide termination sequencing using the Leptospira 16S forward and
232
reverse primers (Elim Biopharmaceuticals, Inc., Hayward, CA).
233
A panel of 26 samples was used to evaluate Plasmodium detection in the UFI assay. The
234
panel included extracted DNA from 25 samples sent to or obtained by the Stanford Clinical
235
Microbiology Laboratory for malaria testing with smear microscopy and the BinaxNOW Malaria
236
rapid diagnostic test (Alere, Waltham, MA). This set contained 10 samples negative for malaria
237
and 15 samples positive for P. falciparum (n=5), P. vivax (n=6), P. malariae (n=2), or P. ovale
238
(n=2). The panel also included a single reference sample of P. knowlesi provided by the CDC
239
(Atlanta, GA). Ten whole blood samples, sent to the Clinical Laboratory at the Mayo Clinic for
240
malaria testing by thin smear, were evaluated using the UFI assay. Whole blood was collected in
241
EDTA tubes and stored at 4°C for up to eight days prior to shipment. Samples were obtained
242
from four patients without malaria as well as patients infected with B. microti (n=2), P.
243
falciparum (n=2), P. ovale and P. malariae (1 each).
244
12
245
Nucleic Acid Extraction. The extraction of RNA from samples used for the evaluation of
246
DENV detection has been described previously (24). The extraction of total nucleic acids from
247
serum or plasma was performed using the easyMAG instrument (BioMerieux, Durham, NC)
248
according to the manufacturer’s recommendations. Extractions from whole blood were also
249
performed using the easyMAG instrument with the “Specific B” protocol. All easyMAG
250
extractions were performed using 140µL of starting material and a final elution volume of 60µL
251
for serum or plasma and 50µL for whole blood (instrument settings for the Specific B protocol).
252
The extraction of DNA from suspected leptospirosis cases was performed on-site in Brazil prior
253
to the shipment of samples for testing in the UFI assay. DNA was extracted using the DNeasy
254
Blood & Tissue Kit (Qiagen, Germantown, MD) according to the manufacturer’s
255
recommendations. All extracted nucleic acids were stored at -80°C until use.
256 257
Statistics. Two-tailed Fisher’s exact tests were used in comparisons of PCR test results. T-tests
258
were used for the comparison of two means. Fisher’s exact and t-tests were performed using
259
GraphPad software (GraphPad; La Jolla, CA). Probit analysis was performed using SPSS (IBM;
260
Armonk, NY).
261
13
262
RESULTS
263 264
UFI Assay Analytical Evaluation. The dynamic range of the UFI assay extended from 7.0 to
265
1.0 log10 copies/µL for DENV-1 and -3 and from 7.0 to 2.0 log10 copies/µL for DENV-2 and -4.
266
The dynamic range for Leptospira was 7.0 to 1.0 log10 copies/µL. For P. falciparum using the
267
Pfr364 assay, the dynamic range extended from 6.5 to 2.0 log10 copies/µL. Lastly, using a
268
plasmid containing the cloned amplicon from a P. vivax isolate, the dynamic range of the
269
Plasmodium 18S assay extended from 7.0 to 1.0 log10 copies/µL. Results of the dynamic range
270
studies are displayed in Figure 1.
271
The 95% LLOD was determined for each target in the UFI assay by probit analysis using
272
dilutions of quantitated standards. The 95% LLOD for each target, expressed in copies/µL of
273
eluate, was the following: DENV-1, 32.6; DENV-2, 19.6; DENV-3, 11.2; DENV-4, 64.3;
274
Leptospira, 8.0; P. falciparum, 64.5; Plasmodium, 2.9. For a direct comparison of the analytical
275
sensitivity of DENV detection in the UFI and pan-DENV assays, the 95% LLOD was
276
determined for the pan-DENV assay using the same quantitated ssDNA standards. The 95%
277
LLOD for each serotype in the pan-DENV assay, expressed in copies/µL of eluate, was the
278
following: DENV-1, 19.4; DENV-2, 8.9; DENV-3, 9.2; DENV-4, 87.9.
279 280
Specificity. The specificity of the UFI assay was evaluated by testing extracted nucleic acids
281
from a set of viral, bacterial, and parasitic samples. No amplification was observed when the
282
UFI assay was performed using nucleic acids from YFV (2 strains), chikungunya virus (2
283
strains), and a single strain each of WNV, SLEV, and Zika virus. Nucleic acids from cultured
284
isolates of S. enterica subsp. arizonae, S. enterica serovar Typhi, S. aureus, E. coli, P.
14
285
aeruginosa, and M. tuberculosis (2 strains) were tested and resulted in no amplification. Also,
286
total nucleic acids were extracted from two whole blood samples from patients with B. microti
287
detected by peripheral blood smear. Both samples had detectable IC but otherwise showed no
288
amplification in the UFI assay. Fifty HCV-positive clinical samples were also tested in the UFI
289
assay; these samples had a median HCV viral load of 5.9 log10 IU/mL of patient plasma (range
290
3.54 to 7.47). All samples had detectable RNase P (mean CT in the crimson channel, 27.18;
291
standard deviation, 1.89). No amplification from these samples was observed in the remaining
292
channels.
293 294
DENV Samples. A set of 60 samples collected from pediatric dengue cases in Nicaragua were
295
tested using the UFI assay. All samples had previously tested positive by at least two different
296
molecular tests, though RNA extracts had undergone 5-6 freeze-thaw cycles prior to this
297
evaluation (24). These included samples from patients infected with DENV-1 (n=26), DENV-2
298
(n=11), and DENV-3 (n=23), which represent the prevalent DENV serotypes in Nicaragua (36,
299
37). A single sample tested negative for both DENV and IC and was excluded from further
300
analysis. Fifty-eight of 59 samples tested positive for DENV in the UFI assay (98.3%; p=1.0).
301
The mean CT value for samples with detectable DENV RNA in both assays was, on average,
302
0.73 cycles later in the UFI assay compared to the pan-DENV assay (95% CI -2.10 to 0.65;
303
p=0.30). The sample missed in the UFI assay had tested positive for DENV-1 with a CT of 34.25
304
in the original pan-DENV assay.
305
15
306
Leptospira Samples. The Leptospira 16S portion of the UFI assay was evaluated by testing
307
DNA extracted from 39 cultured Leptospira isolates, including samples from 7 different species
308
and 23 serovars (Table 3). All isolates were detected using the UFI assay.
309
All 65 clinical samples from suspected leptospirosis cases tested positive for Leptospira
310
in the UFI assay. The rate of Leptospira detection in the UFI assay was significantly greater than
311
the reference PCRs for flaB and lipL41 (24.6%; p