Abstract
Background. The number of human deaths due to rabies is currently underestimated to be 55,000 deaths per year. Biological diagnostic methods for confirmation of rabies remain limited, because testing on postmortem cerebral samples is the reference method, and in many countries, sampling brain tissue is rarely practiced. There is a need for a reliable method based on a simple collection of nonneural specimens.
Methods. A new reverse-transcription, heminested polymerase chain reaction (RT-hnPCR) protocol was standardized at 3 participating centers in Cambodia, Madagascar, and France. Fifty-one patients from Cambodia, Madagascar, Senegal, and France were prospectively enrolled in the study; 43 (84%) were ultimately confirmed as having rabies. A total of 425 samples were collected from these patients during hospitalization. We studied the accuracy of the diagnosis by comparing the results obtained with use of biological fluid specimens (saliva and urine) and skin biopsy specimens with the results obtained with use of the standard rabies diagnostic procedure performed with a postmortem brain biopsy specimen.
Results. The data obtained indicate a high specificity (100%) of RT-hnPCR and a higher sensitivity (⩾98%) when the RT-hnPCR was performed with skin biopsy specimens than when the test was performed with fluid specimens, irrespective of the time of collection (i.e., 1 day after the onset of symptoms or just after death). Also, a sensitivity of 100% was obtained with the saliva sample when we analyzed at least 3 successive samples per patient.
Conclusions. Skin biopsy specimens should be systematically collected in cases of encephalitis of unknown origin. These samples should be tested by RT-hnPCR immediately to confirm rabies; if the technique is not readily available locally, the samples should be tested retrospectively for epidemiological purposes.
Rabies is an acute encephalomyelitis caused by lyssaviruses transmitted to humans by rabid animals. Lyssavirus is an enveloped virus for which negative-strand RNA encodes 5 proteins, including the large (L) polymerase. Although effective, economical control measures are available and the World Health Organization and the World Organization for Animal Health consider rabies a high priority zoonosis, it unfortunately remains a neglected disease in a large part of the world, especially in Africa and Asia. Each year, at least 10 million people receive treatment after being exposed to animals suspected to be rabid; however, 55,000 people still die in Asia and Africa, based on the estimation by the World Health Organization [1]. The global mortality of the disease, based on data communicated to the World Health Organization by its member states through Rabnet [2], is underestimated. On the basis of these reports, the annual incidence of rabies in Asian and African countries ranges from 0.01 to 3 cases per 100,000 inhabitants [3–5]. Modeling of human rabies in Tanzania with use of data sets from an active surveillance site revealed an incidence of 2.9 cases per 100,000 inhabitants, whereas the officially reported annual incidence is <0.01 cases per 100,000 inhabitants [3]. A more accurate estimation of rabies incidence is clearly needed [6].
Various factors are responsible for the underestimation of the number of human deaths due to rabies. Infectious causes of encephalitis often remain unidentified, even in developed countries [7, 8]. Rabies is not a notifiable disease in many countries, and only the encephalitic form of the disease is usually recognized. The paralytic form is rarely identified and is sometimes misdiagnosed [9]. Differentiating this disease from other neurological disorders may require extensive investigation. Therefore, the diagnosis is often made late or is discovered postmortem [10, 11]. Confirmation of the diagnosis is further hampered in the developing world, because most patients who receive a diagnosis of rabies at the hospital return home to avoid the perceived unnecessary costs of clinical care during the final phase of the disease.
To date, the use of biological methods for rabies diagnosis and notification has been very limited in developing countries, because only postmortem diagnosis of brain samples, with a sensitivity near 100%, provides confirmation. However, in many countries, these confirmatory techniques are performed only in exceptional circumstances; this is true particularly in Asia and Africa, where performance of postmortem cerebral biopsies is rarely approved by the family of the patient or even proposed by physicians. The technique by which the biopsy specimens are examined consists of detecting viral nucleocapsids by immunofluorescence and is regarded as the reference protocol for retrospective confirmation of rabies [12]. Various intravitam diagnostic techniques for rabies in humans have been developed, but they are less sensitive than conventional postmortem diagnoses using cerebral tissue specimens [13–18]. Currently, the highest sensitivity (range, 70%–90%) is obtained by immunofluorescence detection of viral inclusion on ultrathin sections of a skin biopsy specimen, but this technique remains difficult to implement in practice [13, 19–21]. More recently, RT-PCR targeting the nucleoprotein gene was applied to skin biopsy specimens collected from 10 patients; the specificity of this test was 70% [22]. In addition, most of these studies were not designed to allow an accurate comparison with the gold standard postmortem diagnostic method.
The present study was undertaken to define a reliable protocol for diagnosing human rabies with use of biological fluid or superficial tissue samples, which are collected in a noninvasive manner and can be used for both intravitam and postmortem diagnosis; our aim was to design a protocol to replace the classic postmortem diagnostic method that uses brain biopsy specimens. Results were obtained during a multicenter evaluation of reverse-transcription, heminested PCR (RT-hnPCR). This test targets conserved blocks in the L polymerase gene sequence of lyssaviruses [23] and was performed on various biological samples collected from 51 patients over time.
Materials and Methods
Interlaboratory evaluation of titration and RT-hnPCR of skin biopsy specimens at 3 centers. Aliquots of challenge-virus-strain rabies suspension with known titer (in fluorescence-forming units per mL) and a panel of 30 skin samples of moribund challenge virus strain and from uninfected mice were collected and coded at Institut Pasteur de Paris (IPP), France. The samples were blindly and independently evaluated at 3 centers: IPP; Institut Pasteur du Cambodge, Phnom Penh, Cambodia; and Institut Pasteur de Madagascar, Antananarivo, Madagascar. Viral suspension was used to determine the threshold of detection obtained at each site after 10 independent assays. Mice skin biopsy samples (10 samples per site; 6 positive and 4 negative samples) were used to evaluate RT-hnPCR of skin biopsy specimens at the 3 test sites. Determination of the spectrum of detection of the RT-hnPCR was performed at IPP with viral suspensions of 6 different prototypes of lyssaviruses [24].
Diagnosis of rabies in patients with encephalitis. This research has complied with all national guidelines and institutional policies; approval was obtained from the ethics committees in Cambodia, Madagascar, Senegal, and France. Patients were enrolled in 4 countries: Cambodia (Calmette Hospital, Phnom Penh), Senegal (Fann-Dakar Hospital, Dakar), Madagascar (Befelatanana Hospital and Soavinandriana Hospital, Antananarivo), and France (the French National Reference Centre for Rabies, IPP).
Patients suspected to have rabies at hospital admission were prospectively enrolled if a signed informed consent form was obtained from the patient or a patient's close relative, if disease had progressed for <10 days, and if clinical symptoms, such as consciousness, behavior or mood disorders, or signs of neurological localization with progression, were observed. Criteria for exclusion were cranial trauma, turbid CSF, acute drug or pesticide intoxication, and history of convulsion.
From September 2003 through July 2006, the following samples were collected at patient admission: a skin biopsy sample (diameter, ∼4 mm; total volume, 20 mm3) taken from the nape of the neck (figure 1) with use of a biopsy punch (Stiefel), a saliva (volume, 1 mL) or saliva swab sample, a urine sample (volume, 1–5 mL), and a serum sample (volume, 0.5 mL). During hospitalization, urine and saliva samples were collected on a daily basis (figure 2). At hospital discharge or after a patient died, samples were collected in a manner similar to that used at hospital admission. If patients died at the hospital and if a brain biopsy (defined as the gold standard test) was performed, the biopsy specimen was excised via the orbital route [25, 26] with use of Tru-Cut biopsy needles for soft tissues with manual clip (Allegiance) or via the occipital route with use of lumbar puncture needles [27]. Patients whose brain samples were positive for rabies by immunofluorescence of viral nucleocapsids in the sample were categorized in group 1. When it was not possible to perform a brain biopsy for a patient, a positive result of RT-hnPCR was considered to be indicative of rabies (figure 2), and these patients were added to group 1 to constitute group 2. After hospital discharge, a follow-up visit at the patient's residence was conducted to confirm whether the patient was still alive 2 weeks after the onset of symptoms.
Nape of the neck (area in black), where the skin biopsy specimens were obtained.
Flow diagram of patients enrolled in the rabies diagnostic accuracy study.
Biological analysis. The samples were transferred from the clinical investigation centers to the local reference laboratory immediately (at 4°C) or after shipment (on dry ice). Samples collected in Cambodia and in Madagascar were analyzed at Institut Pasteur du Cambodge and Institut Pasteur de Madagascar, respectively. Samples collected in France and Senegal were analyzed at IPP. These 3 laboratories are responsible for routine rabies surveillance in their respective countries. Rabies-specific antibodies were measured with use of the Platelia Rabies II ELISA method (Bio-Rad) at IPP [28]. Rabies was diagnosed postmortem by fluorescent antibody test performed on the brain biopsy sample as the standard [12, 29]. Skin biopsy samples were dissociated with sterile scissors and were incubated in 180 µL of ALT tissue lysis buffer and 20 µL of proteinase K (Qiagen) at 37°C for 3 h under gentle agitation. The suspension was mixed with 0.8 mL of TRI Reagent LS (Molecular Research Center), and the extraction was processed according to the manufacturer's instructions. RNA was extracted from saliva samples (0.2 mL or saliva swab samples) with use of TRI Reagent LS; glycogen was added (Ambion) and used as a coprecipitant. The QIAmp Ultrasens Virus kit (Qiagen) was used to extract RNA from a 1-mL urine sample, according to the manufacturer's instructions. Reverse transcription was performed as described elsewhere [13], with minor modifications. In brief, 6 µL of RNA (nearly 1.5 µg/mL) were incubated at 65°C for 10 min with 2 µL of pd(N)6 random primers (200 µg/mL; Roche Diagnostics) and 2 µL of sterilized distilled water and then were stored on ice. Each tube was incubated with 200 U of Superscript II RT (Invitrogen), 80 U of RNasin (Promega), and 10 nmol of each nucleotide triphosphate (Eurobio), in a final volume of 30 µL for 90 min at 42°C, for reverse transcription [13]. Two microliters of complementary DNA (cDNA) was then amplified by hnPCR, with use of 10 pmol of primers PVO5m/PVO9 in the first round and primers PVO5m/PVO8 in the second round; amplification reactions contained 2 U of AmpliTaq DNA Polymerase (Applied Biosystems), 10 nmol of each nucleotide triphosphate, and 62.5 nmol of magnesium chloride, in a final volume of 50 µL (table 1). The mean size of amplicons obtained after the second round of PCR was 249 base pairs. These primers were designed to recognize conserved regions between the L polymerase genes of lyssaviruses during evolution [23]. RNA template quality was assessed by the parallel amplification of β-actin mRNA in each sample, with use of primers b-Taq1 and b-Taq2 (table 1) [30]. Negative and positive control samples were analyzed successively for each extraction (negative control samples only), reverse transcription, first PCR, and second PCR step and were subsequently submitted to the remaining steps of the procedure.
Description of primers and heminested PCR running conditions.
Statistics. Statistical analyses (Kruskal-Wallis and Yates corrected Χ2 tests) were performed with use of Stata, version 8.2 (Stata).
Results
Interlaboratory evaluation of titration and RT-hnPCR of skin biopsy specimens at 3 centers. To explore possible sources of heterogeneity in results among the participating institutions, reproducibility and repeatability of the RT-hnPCR method were assessed by an interlaboratory test. We determined the threshold of detection of aliquots of 1 challenge-virus-strain suspension with a known titer at the 3 sites (Institut Pasteur du Cambodge, Institut Pasteur de Madagascar, and IPP). Results revealed low variability (the SD never exceeded 20% of the mean value). The results obtained at the 3 sites were not significantly different (P>.05, by Kruskal-Wallis test) and, therefore, were pooled to obtain a mean threshold for 30 replicates (6 fluorescence-forming units per mL). Similarly, the evaluation of RT-hnPCR of skin biopsy specimens, performed blindly and independently on 10 mice skin samples (infected and uninfected), was 100% concordant at the 3 sites. We also determined the spectrum of detection of RT-hnPCR at IPP. This method was used to analyze various suspensions of prototype viruses of 6 genotypes of lyssaviruses (table 2). Results indicate a large spectrum of detection with a low threshold of detection—at least as good as previously published methods [13, 31–36]. In conclusion, this technique was demonstrated to be sensitive, reproducible, and repeatable for the detection of lyssaviruses in the framework of a multicenter evaluation.
Threshold and spectrum of detection of reverse-transcription, heminested PCR performed with various suspensions of prototype viruses of 6 genotypes of lyssavirus.
Diagnosis of rabies in patients with encephalitis. To define a reliable protocol for intravitam or postmortem diagnosis of rabies in humans that is based on examination of noninvasive or superficial tissue samples to replace classic postmortem diagnosis based on examination of brain biopsy samples, a cohort of 51 patients was enrolled in Cambodia, Madagascar, Senegal, and France. Clinical data were collected for each patient. Forty-three patients (group 2) were confirmed to have rabies infection (figure 2). of these 43 patients, 33 (77%) originated from Asia, and 10 (23%) originated from Africa. The sex ratio (male∶female, 1.87) and the low median age at death (23 years; range, 3–60 years) are representative of the epidemiology of rabies worldwide [1]. Most patients with confirmed rabies had a classic clinical presentation of encephalitic rabies. The median incubation period was typical for 29 patients (9 weeks; range, 10 days–3 years) [37]. The median survival time after the onset of symptoms was 4 days (range, 1–10 days) for 37 patients. At hospital admission, most of the patients presented with hyperexcitability (41 [95%] of 43 patients), which is one of the major cardinal clinical features of encephalitic rabies. Other cardinal features were fluctuating consciousness (29 [69%] of 42 patients), phobic spasms (aerophobia or hydrophobia; noticed in only 14 [33%] of 43 patients), and autonomic dysfunction signs (14 [32%] of 43), including enuresis (9 [22%] of 41), hypersalivation (3 [7%] of 43), priapism (1 [2.3%] of 43), and sweating (1 [2.3%] of 43) (table 3). Fever and coma were also noted in 32 (74%) and 3 (7%) of the 43 patients at hospital admission, respectively. Others clinical features included dysphagia (39 [95%] of 41 patients), dysphonia (20 [49%] of 41), headache (20 [49%] of 41), and nuchal rigidity (3 [8%] of 39). We observed an increase in the deep tendon reflex and Babinski sign in 19 (48%) of 40 patients and 10 (26%) of 39 patients, respectively. The paralytic phase was not evident. Localized motor weakness was noticed in 9 (21%) of 42 patients, and myoedema of the chest and the neck, more specific to paralytic rabies [38], was reported in only 1 patient.
Clinical features of 43 patients with rabies at hospital admission (group 2).
Discussion
The present study represents, to our knowledge, the largest data set for samples ever serially collected intravitam from patients suspected of having encephalitic rabies and analyzed by RT-PCR. A longitudinal study that involved the 51 patients allowed the serial collection of 425 samples during hospitalization (figure 2). Rabies was excluded in 8 patients because of recovery of the patient or survival without intensive care for >15 days (n=6) or because the terminal brain biopsy specimen was negative by fluorescent antibody test (n=2). For the 140 samples collected from these 8 patients (2 brain biopsy, 60 saliva, 9 skin biopsy, 56 urine, and 13 serum samples), all except serum samples were tested using RT-hnPCR, and results remained negative, indicating that this test had 100% specificity in humans. A few nonspecific amplification products were observed for some samples but never at the expected size. Sequencing of these products further revealed nonspecific amplifications. Rabies was confirmed by fluorescent antibody testing of terminal brain biopsy specimens obtained from 32 patients (group 1). Results of RT-hnPCR performed on these samples were also positive. Rabies in the remaining 11 patients was confirmed on the basis of at least 1 positive result of RT-hnPCR. Finally, 285 samples were obtained from 43 patients with rabies (group 2) (table 4 and figure 2).
Proportion of positive samples for each patient with rabies.
Viral nucleocapsids are located in nerve endings surrounding the base of hair follicles [13, 19–21]. The recommended sampling area is the upper portion of the nape of the neck, an easily accessible area with a high density of hair (figure 1). In our study, skin biopsy specimens tested by RT-hnPCR targeting the L polymerase gene exhibited a very high sensitivity (98% [50 specimens in group 1] and 98.3% [60 specimens in group 2]) (table 5). The total sensitivity per patient in group 2 was 97.4% (38 patients), and the sensitivity as early as 1 day after the onset of symptoms or at hospital admission was 100% (8 and 29 patients, respectively) (table 5 and figure 3). In a previous study, RT-PCR of skin biopsy specimens collected from 10 patients that targeted the nucleoprotein gene resulted in a sensitivity of only 70% per sample and 77.7% per patient [22]. This comparison further reinforces why it is interesting to target the L polymerase gene to obtain a low threshold of detection, as demonstrated by the interlaboratory evaluation of this technique. Furthermore, there was no statistically significant variation in the number of positive skin biopsy samples as a result of the delay between the onset of symptoms or hospital admission and the date that the sample was obtained (P>.05, by Yates corrected Χ2) (figure 3). These results established that the window for detecting lyssavirus RNA with use of skin biopsy specimens is large and that this sample effectively confirms the diagnosis of rabies, whether at the day of hospital admission or just after death and regardless of the delay between the onset of symptoms and the date that the sample is obtained.
Diagnosis of rabies based on testing of samples obtained from 43 patients with rabies (group 2).
Variation of the rate of rabies according to testing of saliva (A and C) and skin biopsy (B and D) specimens, by the delay between the onset of symptoms (37 patients) or hospital admission (43 patients; group 2) and the date that a sample was obtained from a patient with rabies. The percentage of positive samples is indicated by open circles, and the number of samples analyzed for each day is indicated by bars.
Saliva samples provided the second-best results for sensitivity testing (63.2% [57 samples in group 1] and 70.2% [84 samples in group 2]) (table 5). These sensitivities are in the upper limit of the range (10%–70%) obtained by diagnostic techniques that target the nucleoprotein gene in saliva samples, according to previous studies [13–18, 39]. In the present study, we also detected viral RNA intermittently in the saliva samples (in 6 of 20 patients investigated who had multiple saliva samples obtained) (table 4) [13, 14, 17, 18]. However, none of the patients who received a diagnosis of rabies had >2 consecutive negative saliva samples (table 4). In addition, we observed an increase in the positivity rate of saliva samples during the first 2 days after the onset of the symptoms. This positivity rate remained stable from day 2 through day 7 or later (figure 3). The difference in sensitivities between saliva swab samples (52.9%; 17 samples) and liquid saliva samples (74.6%; 67 samples) was not statistically significant (analyzed in group 2; P>.05, by Χ2 test).
The results obtained with the other tests and samples were poor (tables 4 and 5). Sensitivity of testing of urine samples (14.6% [41 samples in group 1] and 9.5% [63 samples in group 2]) was low, although higher rates were obtained between days 3 and 5 after hospital admission (data not shown). The lower sensitivity rates obtained with urine samples in the present study, compared with previous studies [14, 18], may be a result of the small volume of urine used for extraction in the present study. Again, lyssaviruses were intermittently detected in 4 of 15 patients who had multiple urine samples obtained (group 2) (table 4). The lack of detection of antibodies against rabies in serum samples by ELISA (35 samples in group 1 and 46 in group 2) is not surprising, considering the short period of survival of the patients after the onset of symptoms [11, 13, 21].
In summary, RT-hnPCR targeting the L polymerase gene can be useful, specific, and extremely sensitive (⩾98%) if applied to skin biopsy samples, whether collected at hospital admission, during hospitalization, or after death, from patients suspected to have encephalitic rabies. Results correlate very well with those of the fluorescent antibody test performed on brain biopsy specimens, irrespective of the geographic origin of the patient. Because a skin sample is a superficial tissue sample that is easy to collect, we strongly urge that skin biopsy samples collected as described in figure 1 be obtained from patients with risk factors for rabies and neurological symptoms that are consistent with rabies or that are otherwise unexplained. The alternative method would be to collect 3 serial daily saliva samples (sensitivity in our study, 100%). This would allow a rapid antemortem diagnosis of rabies, if available, and would provide a more reliable picture of the actual incidence of human rabies [6]. It also could be useful in rapidly deciding whether to avoid transplantation from organ donors at risk of rabies [40, 41] and would help to avoid the contamination of health care personnel by rabies virus–containing biological fluids. It remains difficult to obtain brain tissue specimens for analysis worldwide, because of the reluctance of medical staff and, above all, the family of the patient. Therefore, we propose RT-hnPCR as an alternative to the currently recommended method of postmortem diagnosis with use of brain biopsy specimens for confirming human rabies, although the suitability of our test should be further tested in patients suspected of having paralytic rabies.
Acknowledgments
We thank the various laboratory technicians involved in this study, including those at the National Reference Centre for Rabies in Paris, France, and those at the virology units of the Institut Pasteur du Cambodge, Phnom Penh, Cambodia, and the Institut Pasteur de Madagascar, Antananarivo, Madagascar; J. L. Sarthou (Director, Institut Pasteur du Cambodge), A. Talarmin (Director, Institut Pasteur de Madagascar), and F. Simon (Director) and A. A. Sall (Institut Pasteur de Dakar), for their precious help and encouragement; and Fraņçois-Xavier Meslin, for critical review of the manuscript.
Financial support. Institut Pasteur International Network Actions Concertées InterPasteuriennes (2003/687), the Institut de Veille Sanitaire, and the Conny Maeva Charitable Foundation presided over by Dominique Dunant.
Potential conflicts of interest. All authors: no conflicts.
Comments