Adenovirus isolation rates in acute flaccid paralysis patients

INTRODUCTION

Adenoviruses are non-enveloped icosahedral dsDNA viruses. The genus Mastadenovirus comprises mammalian adenoviruses including 57 types of Human adenoviruses (HAdV), which are classified into seven species, HAdV-A to HAdV-G (http://www.ictvonline.org/virusTaxonomy.asp?version=2009) [Jones et al., 2007]. Adenovirus infection in immunocompetent subjects is usually asymptomatic or manifests as a mild respiratory syndrome most frequently caused by various HAdV-B, HAdV-C, and HAdV-E types [Wold and Horwitz, 2007]. Occasionally, HAdV species have been reported to cause conjunctivitis (HAdV-D19C [Robinson et al., 2009], -D37 [Robinson et al., 2008], -D53 [Walsh et al., 2009], -D54 [Ishiko et al., 2008], -D56 [Henquell et al., 2009; Robinson et al., 2011]), pneumonia (HAdV-B14 [Louie et al., 2008]), gastroenteritis (HAdV-F40 and HAdV-G52 [Jones et al., 2007]), hemorrhagic cystitis [Wold and Horwitz, 2007], myocarditis [Bowles et al., 2003], and systemic infection [Chuang et al., 2003]. In immunocompromised patients (mainly hematopoietic stem cell and solid organ transplant recipients) adenoviruses may cause systemic life-threatening disease [Kojaoghlanian et al., 2003; Seidemann et al., 2004; Kampmann et al., 2005].

A number of investigators have reported sporadic findings of adenovirus infection of the central nervous system (CNS). Various HAdV-B types, most frequently HAdV-B3 and HAdV-B7, have been implicated in acute myelopathies [Linssen et al., 1991; Ohtsuki et al., 2000]. Adenoviruses have been detected in the spinal cord and cerebrospinal fluid of patients with neurological manifestations [Faulker and van Rooyen, 1962; Simila et al., 1970] and have been reported to cause encephalitis [Chuang et al., 2003; Koskiniemi et al., 1991].

In 1997, during a large outbreak of enterovirus 71 infection in Malaysia, HAdV-B21 was isolated from serum, nasopharyngeal washings, myocardium, and brain of a child with lethal acute flaccid paralysis (AFP) [Ooi et al., 2003]. It was hypothesized that this agent was responsible for the CNS damage, while peripheral infection with enterovirus 71 facilitated penetration of the adenovirus into the CNS [Cardosa et al., 1999; AbuBakar et al., 2000; Ooi et al., 2003]. It has also been suggested that adenoviruses may be a common cause of acute flaccid paralysis. Investigation of adenovirus prevalence among children with AFP in Brazil was then conducted with the aim of testing this hypothesis [de Azevedo et al., 2004]. No conclusive evidence supporting or rejecting the suggestion was obtained. No further research has been conducted on this problem.

The Russian Federation was certified as a polio-free region in 2002 [CDC, 2002], and most AFP cases are now attributed to non-infectious causes. We have investigated association of adenoviruses with neurological manifestations among patients with acute flaccid paralysis and healthy controls and among pediatric and adult patients with primary immunodeficiency.

MATERIALS AND METHODS

Viruses used in the study were isolated in 2000–2007 by the WHO Regional Reference Laboratory for poliomyelitis at the M. P. Chumakov Institute of Poliomyelitis and Viral Encephalitides, Moscow, in the framework of an AFP surveillance program in Russia and the New Independent States. The principal groups in this study were patients with neurological manifestations and asymptomatic contact cases as a control. Only viruses that were isolated from feces collected in Russia and five ex-USSR countries (Armenia, Azerbaijan, Tajikistan, Turkmenistan, and Uzbekistan) were studied, because all stool samples from AFP cases from these countries were transferred to the Regional Reference Laboratory, and the sampling was therefore complete and representative. A total of 5,798 stool samples from AFP cases and 2,399 samples from healthy contacts were investigated. Fecal samples of 70 patients with primary immunodeficiency were collected from 40 children <5 years old and 30 adults with diagnoses of common variable immunodeficiency, agammaglobulinemia, Wiskott–Aldrich syndrome, etc. All the immunodeficient patients were in Russia and they received scheduled medical assistance in the relevant medical institutions of the Russian Federation. Samples from 80 healthy young children (below 5 years of age) from two orphanages were collected in Omsk and Novosibirsk using the same protocol.

Collection of stool samples and virus isolation were carried out in accordance with WHO guidelines for poliovirus and enterovirus isolation [WHO, 2004]. All samples and epidemiological data were anonymized. All parts of the study were approved by the Ethical Committee of M.P. Chumakov Institute of Poliomyelitis and Viral Encephalitides. Informed consent was obtained from patients with immunodeficiency or their legal representatives. According to the national regulations, informed consent is not required for anonymized surveillance studies. Adenoviruses were provisionally identified by typical grape-like cytopathic effect (CPE) in HEp-2 cell culture (ATCC CCL-23, HeLa-derived). Initial identification of cytopathic adenoviruses by CPE morphology provided excellent sensitivity (i.e., none of the CPE agents provisionally identified as enteroviruses were found to be adenoviruses, data not shown) and specificity (all viruses selected by CPE morphology were proved to be adenoviruses by PCR).

Adenovirus DNA was isolated from cell culture lysate by phenol–chloroform extraction. Presence of adenovirus was confirmed by PCR with primers AdHex1F and AdHex1R specific to the adenovirus Hexon genome region [Pring-Åkerblom et al., 1997] and an alternative PCR assay specific to the Hexon genome region [Xu et al., 2000].

HAdV species were identified using multiplex PCR [Xu et al., 2000]. The most prevalent HAdV-C types (HAdV-C1, C2, and C5) were identified by restriction endonuclease digestion of the Hexon gene PCR product obtained with previously published primers AdHex1F and AdHex1R [Pring-Åkerblom et al., 1997]. An approximately 840 nt PCR product was digested by MboI restriction endonuclease (Promega, Madison, WI) directly in the PCR reaction buffer, and it produced fragments of 396, 209, 120, and 115 nt for HadV-C1 and 411, 146, 115, 82, and 41 nt for HadV-C2. The HAdV-C5 PCR product was not digested by MboI; it was identified by digestion with PvuII (Promega), which produced fragments of 439 and 359 nt specific to HAdV-C5. Digested DNA fragments were resolved in 3% agarose gel against uncut PCR products to confirm the specificity of the observed bands. Detection of virus mixtures in multiplex PCR or PCR-RFLP was verified by repeated analysis. The type of selected strains was identified by sequencing the PCR product for the Hexon gene and comparing the sequence to GenBank data [Casas et al., 2005].

Statistical significance of differences in virus prevalence was evaluated with Fisher's exact test using GraphPad Prism 4.0 (GraphPad Software, Inc.).

RESULTS

Adenovirus isolation rates were more than twice higher among pediatric AFP patients (1.05%) than in asymptomatic contact children (0.42%, P < 0.01) (Table I). These results were compared to a similar study that was conducted in Brazil under the same WHO guidelines and using a similar laboratory protocol. Adenovirus prevalence among pediatric AFP patients in Russia and the bordering states was almost twice lower than reported for the same age group in Brazil (P < 0.001), while adenovirus prevalence in contact children was almost fivefold lower in Russia (P = 0.0001) (Table I). The ratio of adenovirus species found in this study and in Brazil was also different. HAdV-C viruses were the most prevalent in both countries, but in Russia they comprised 87% of all isolates versus 59% in Brazil. On the other hand, we found significantly less HAdV-B, 6% of all isolates versus 32% in Brazil (P < 0.001). The most prevalent HAdV-C types were HAdV-C2 (26 of 62 HAdV-C detections), HAdV-C1 (20 of 62), and HAdV-C5 (15 of 62). HAdV-C6 was represented by a single isolate. None of the HAdV-C samples displayed an RFLP pattern compatible with the sequence of a novel type HAdV-C57 that was isolated in Azerbaijan in 2003 [Lukashev et al., 2008; Walsh et al., 2011].

In addition to the distribution of HAdV species, the correlation of HAdV isolation with particular clinical diagnoses was studied. AFP surveillance in Russia implies reporting and investigation of each case, and the concluding clinical diagnoses are reported to the national database. No significant difference in the frequency of any neurological diagnosis among infants that yielded adenovirus and all AFP cases in total could be found (Supplementary Table S1).

The prevalence of adenoviruses in neurological patients was significantly higher than in controls. One explanation for this could be that patients with neurological manifestations might have accompanying diseases that could affect overall immunity and result in increased asymptomatic infection rate. Therefore, adenovirus prevalence among pediatric and adult patients with primary immune deficiencies was studied (Table II). Indeed, adenoviruses were found in feces of 10% of adult and 17.5% of pediatric immunodeficient patients, which is up to 40 times more frequent than in healthy young children (0.42%). Another explanation for higher incidence of adenoviruses in pediatric patients could come from the fact that they were usually admitted to hospital, while contact subjects were screened in their usual setting. To check this hypothesis, the prevalence of asymptomatic adenovirus infection in healthy young children in two orphanages as examples of crowded children's groups was studied (Table II). Prevalence of adenovirus excretion in asymptomatic children in both orphanages was up to 135 times higher than among healthy contacts of AFP cases that belonged to the same age group, and they varied considerably between the two orphanages studied. Typing of adenoviruses isolated from children in orphanages revealed co-circulation of several types (HAdV-C1, C2, and C5), thus showing that this extremely high rate of adenovirus excretion did not result from an outbreak of subclinical infection caused by a single virus, but rather reflected asymptomatic adenovirus circulation. A number of samples from all groups presented multiplex PCR or RFLP profiles indicative of an adenovirus mixture. The ratio of adenovirus mixture detection was much higher among children with immunodeficiency (3 of 10) and infants in orphanages (5 of 19) than in pediatric AFP patients and their contacts (1 of 71) (P < 0.01).

Repeated screening of immunodeficient patients and young children in orphanages was performed to determine if high adenovirus prevalence rates could be explained by virus persistence. Adenoviruses, especially HAdV-C, are known to persist in tonsil tissue [Garnett et al., 2002]; however, it was not known if this results in persistent virus excretion with feces. The young children in orphanage #1 were screened monthly for 6 months, and no cases of repeated adenovirus excretion were detected. In orphanage #2, repeated screening 1 month after the initial sampling revealed adenoviruses in feces of 13 of 21 children. Only three infants were secreting adenoviruses on the first and second examination. Importantly, PCR-RFLP typing of isolates from these three infants revealed that they were shedding different HAdV-C types, implying independent infections rather than virus persistence. Co-circulation of several HAdV-C types, absence of virus persistence and reinfection of three infants with distinct HAdV-C types over a 1-month interval suggests existence of a local endemic nidus rather than a concealed outbreak that biased adenovirus prevalence rates at the first screening in orphanage #2. It is likely that the repeated infection with different types was facilitated in this orphanage because the infants were of ages below 1 year, which is the most susceptible age group for non-specific adenovirus infection [Lewis et al., 2009].

Three out of 10 patients with primary immunodeficiency that shed adenoviruses were available for repeated examination 1–2 months later. Two of these did not demonstrate repeated adenovirus excretion. For one patient (male, 31 years old, with diagnosis of common variable immunodeficiency) that shed a mixture of adenoviruses (HAdV-C2 + HAdV-D15), repeated sampling was performed five times, at 1, 3, 8, 12, and 13 months after the first sampling. It was found that shedding of the mixture of adenoviruses continued at least for 8 months and ceased thereafter. At some time points (including the last positive sampling at 8 months), both viruses were detectable, while at the 1-month time point only HAdV-D15 could be found. Despite this, HAdV-C2 and HAdV-D15 identified at different periods of observation had identical nucleotide sequences, suggesting virus persistence rather than reinfection.

DISCUSSION

The cornerstone of the poliomyelitis eradication programme is surveillance for acute flaccid paralysis followed by diligent investigation of each case, because neurological lesions can have various etiologies other than poliomyelitis. Indeed, a major fraction of AFP cases is attributed to various medical conditions or an unidentified cause. Adenoviruses, especially HAdV-B species members with reported neuropathogenic potential, could be responsible for a fraction of these cases. This study tested this hypothesis and investigated prevalence of adenoviruses among patients with AFP and controls in Russia and bordering countries over seven years. The worldwide network of polio surveillance laboratories operates throughout the world and uses the same virus isolation protocol [WHO, 2004] and cell cultures from the same source; therefore, results of this study could be confidently compared with those reported by polio surveillance laboratories in Brazil, Peru, and Bolivia [de Azevedo et al., 2004].

The detection rate of adenoviruses in pediatric AFP patients was 2.5 times higher than in their controls, and this notable difference passed a formal statistical significance test. However, this finding could not explicitly prove the etiological role of adenoviruses, and several observations contradicted this hypothesis. First, in a significant number of patients that yielded adenoviruses (40%) traumatic neuritis was determined later as the cause of neurological manifestations, which was not compatible with neuroinfection and implied that these were coincidental isolations. Second, the number of potentially neurovirulent HAdV-B isolates in this study was much lower than in a similar study conducted in Brazil. Therefore, other factors that could explain apparent association of adenovirus excretion in feces with AFP were investigated. While the patients were matched to controls in age, sex, geographic location, and seasonal factors, other circumstances could affect adenovirus detection rates. For example, some patients could have weakened immune status due to a disease that manifested with neurological lesions. A group of patients with primary immune deficiencies was screened to test how immune suppression could affect prevalence of adenovirus excretion. Adenovirus detection rates were more than tenfold higher among immunodeficient patients than in control subjects and AFP cases. Therefore, deterioration of the immune state due to an unrelated disease could complement to higher adenovirus prevalence in AFP patients. Also, admission of infants with AFP to children's hospitals and subsequent close contacts with other patients could add to the probability of adenovirus detection. Indeed, the prevalence of adenovirus shedding in both examined orphanages was extremely high. In particular, this study revealed over 50% prevalence of asymptomatic or subclinical adenovirus infection in orphanage #2 upon two repeated screenings. Similar epidemiological conditions could exist in children's hospitals, adding to the apparent adenovirus prevalence in the patients. Therefore, at least two factors that could contribute to the apparent association between AFP and adenovirus shedding could be confirmed. Adenoviruses are ubiquitous, and infection is usually subclinical or very mild. Even if adenoviruses were indeed a common causing of neurological manifestations, it would be very hard to prove this. Indeed, detection of virus in feces of a patient is hardly an evidence of its etiological role, while isolation of the same virus from many healthy patients does not rule out its capacity to cause the disease. In addition, independent factors, such as minor disparities between samplings, could possibly affect adenovirus prevalence in two almost perfectly matched samplings of patients and their contacts. Adenovirus detection in cerebrospinal fluid could have been more supportive for its role in AFP, but even then much effort would have to be devoted to prove the role of these viruses in disease pathogenesis.

Several studies on adenovirus epidemiology reported coinfection rates ranging from negligible to 18% in overt respiratory infections [Metzgar et al., 2005] and 100% in (likely immunodeficient) military recruits that did not respond to vaccination against HAdV-4 and HAdV-7 [Vora et al., 2006]. An impressive increase in coinfection rates in immunodeficient patients (30%) and orphanages (26%) was observed compared to AFP and controls groups (1.4%), which was consistent with adenovirus mono-infection prevalence rates and statistically expected co-infection chances. It is not clear what role co-infection plays in disease pathogenesis, but it clearly creates perfect conditions for intra-species recombination, which is the key evolutionary mechanism in circulating adenoviruses [Lukashev et al., 2008], and this can dramatically speed up adenovirus evolution. Somewhat surprisingly, no additional strains of a novel HAdV-C57 [Walsh et al., 2011] were detected among adenovirus isolates. HAdV-C57 was originally isolated from an asymptomatic child in Azerbaijan in 2003, in the country and in the year covered by this study. Therefore, the current level of adenovirus epidemiological studies does not exclude existence and possible emergence of rare virus variants.