Human parainfluenza 2 & 4: Clinical and genetic epidemiology in the UK, 2013–2017, reveals distinct disease features and co‐circulating genomic subtypes

Abstract Background Human Parainfluenza viruses (HPIV) comprise of four members of the genetically distinct genera of Respirovirus (HPIV1&3) and Orthorubulavirus (HPIV2&4), causing significant upper and lower respiratory tract infections worldwide, particularly in children. However, despite frequent molecular diagnosis, they are frequently considered collectively or with HPIV4 overlooked entirely. We therefore investigated clinical and viral epidemiological distinctions of the relatively less prevalent Orthorubulaviruses HPIV2&4 at a regional UK hospital across four autumn/winter epidemic seasons. Methods A retrospective audit of clinical features of all HPIV2 or HPIV4 RT‐PCR‐positive patients, diagnosed between 1st September 2013 and 12th April 2017 was undertaken, alongside sequencing of viral genome fragments in a representative subset of samples. Results Infection was observed across all age groups, but predominantly in children under nine and adults over 40, with almost twice as many HPIV4 as HPIV2 cases. Fever, abnormal haematology, elevated C‐reactive protein and hospital admission were more frequently seen in HPIV2 than HPIV4 infection. Each of the four seasonal peaks of either HPIV2, HPIV4 or both, closely matched that of RSV, occurring in November and December and preceding that of Influenza A. A subset of viruses were partially sequenced, indicating co‐circulation of multiple subtypes of both HPIV2&4, but with little variation between each epidemic season or from limited global reference sequences. Conclusions Despite being closest known genetic relatives, our data indicates a potential difference in associated disease between HPIV2 and HPIV4, with more hospitalisation seen in HPIV2 mono‐infected individuals, but a greater overall number of HPIV4 cases.


| INTRODUCTION
Human parainfluenza viruses types 1 to 4 (HPIV1-4) are collectively the second most common cause of hospitalisation for children under the age of five, behind only Respiratory Syncytial Virus (RSV). [1][2][3][4] Symptomatic HPIV infection is observed in both adults and children worldwide, affecting both the upper and lower respiratory tract 5,6 with varying severity in the immunocompromised and elderly. 7,8 HPIV2 presents generally with common cold-like symptoms and is a frequent cause of croup in infants. 2,9 HPIV4 is less well characterised but has been associated with bronchiolitis and pneumonia. 5,10 Parainfluenza virus infections place a significant burden on the global healthcare system. In the US alone, a 12-year retrospective study estimated hospital charges for children under the age of five annually totalled in excess of $42 and $57 million for HPIV-associated bronchiolitis and pneumonia, respectively, with a gross yearly HPIV associated US hospitalisation burden estimated at 62000 days. 11 HPIVs belong to the single-stranded negative sense RNA Paramyxoviridae family and are sub-divided to the Respirovirus genus (HPIV1&3) of the Orthoparamyxovirinae subfamily and the significantly genetically distinct Orthorubulavirus genus (HPIV2&4) of the Rubulavirinae subfamily. 12 HPIV2&4 present an orthodox six gene Paramyxovirus genome, 12 with particularly the Hemagglutinin-Neuraminidase (HN) and also Fusion (F) envelope genes known to display a higher level of antigenic variation than the structural components of the orthorubulavirus genome, making them a more appropriate focus for epidemiological studies. [13][14][15][16] Two nearly identical archetypal strains of HPIV2 have been described: Greer in the mid-1950s in the US 17 and Toshiba in the late 1970s in Japan. 18 More recently, additional lineage defining strains Vanderbilt 94 and 98 have been characterised, with maximal dissimilarity rates circa 5% at the amino acid level. 13 In contrast HPIV4 is categorised into two different subtypes (HPIV4A and HPIV4B) based on antigenic properties, 19,20 despite presenting apparently less divergent genomes 16 which do not meet criteria for demarcation as separate species. 12 To date no distinction in clinical outcome has been made amongst circulating HPIV2 or 4 species, further illustrated by phylogenetic studies indicating genetically related clades often contain strains from different seasons and distant geographical origins. 15,16 Formative viral diagnostic protocols reliant on cell culture (commonly with primary rhesus monkey kidney cells) resulted in low recovery rates, little cytopathic effect and a weak haemadsorption pattern. 9,19,21 This has in part led to HPIV4's frequent omission from standard diagnostic respiratory investigation and derivation of minimal reference genome data, in turn contributing to a reduced comprehension of its epidemiological significance. 9,10,15,22 Higher sensitivity and specificity of reverse transcription PCR (RT-PCR) 23 and its cost-effective ability to rapidly diagnose viral respiratory infection has improved HPIV surveillance in the current millennium. Recent advances in sequencing technologies should further redress this shortfall in genomic reference material. 24 To further increase our knowledge of HPIV2&4 clinical epidemiology, and the differences between the two infections, we undertook a retrospective analysis of all RT-PCR-positive samples at a regional UK

| Statistical analyses
All data analyses were performed using Microsoft Excel and GraphPad PRISM 7.04.
Comparison of clinical parameters was performed using Graphpad Prism (v9.3.1) statistical software. Similar to previous studies, 9,22 binary logistic regression comparing HPIV2 and HPIV4 infections was not conducted due to the occurrence of incomplete data for some individuals, limiting our ability to model the contributions of some parameters to a logistic regression. As such, categorical datasets were compared using either Fisher's exact test or χ2 tests. Median values of continuous datasets were compared using Mann-Whitney tests.

| Routine diagnostic surveillance
Within the study period from 1st September 2013 to 11th April 2017, 26 593 unique respiratory specimens were investigated by routine RT-PCR viral screening, of which 10 283 (38.15%) were positive for one or more viruses (data not shown

| Demographics
Patients were categorised into gender and six age groups, then assessed for HPIV2 & 4 prevalence ( Figure 2). For both types of orthorubulavirus, an increase in cases was observed towards the extremes of age, i.e. in infants/young children and older adults, with low numbers of infections observed in those aged 10 to 40 (adolescents and young adults). This effect was most pronounced in HPIV4 with 61.31% of cases in children under 10 years of age, in contrast to HPIV2 with just 42.86% ( Figure 2 and Table 1). The 65 and over elderly group comprised almost one fifth of cases for both HPIV2 (19.64%) and HPIV4 (17.59%, Figure 2 and Table 1).    (Figure 4).  Although almost twice as many HPIV4 mono-infected patients under the age of 10 were observed, the median ages were ultimately not statistically significant (p = 0.0527, Table 1). Most strikingly, a fever was observed with more than 3-fold frequency in HPIV2 than HPIV4 patients (p < 0.0001, Table 1) but no further significant differences were seen in variables assessed. For both HPIV2 and HPIV4-infected and hospitalised patients, underlying medical conditions were very common as was immunocompromise (Table 2). Other than the aforementioned fever associated with HPIV2, shortness of breath and cough were the most commonly recorded symptoms, but intervention with nebulisers and supplementary oxygen was only required in a minority of instances; antibiotics were provided to approximately half of the cohort ( Table 2).

| Clinical characteristics
Patient blood was assessed more frequently in general than chest x-rays, with significantly more haematological aberrance outside normal parameters (i.e. both low and high levels, Supplementary features of HPIV2 but not HPIV4 infection. Conversely, no differences were observed between HPIV2&4 for platelets, white blood cells, total lymphocytes and basophils (Table 2).
Liver and kidney function were assessed more frequently in HPIV2 patient care (p = 0.0139 and 0.0091 respectively, Table 2).
However, of the many parameters tested, a significant disparity was seen only in the more frequently abnormally high C-reactive protein levels of HPIV2 infection (p = 0.0084, Table 2). In general, abnormally high albumin but not bilirubin levels were recorded whilst urea levels appeared marginally more frequently outside standard ranges than sodium, potassium and creatinine (Table 2).

| DISCUSSION
In contrast to more prevalent pre-SARS-CoV-2 respiratory pathogens such as Influenza A and RSV, the Human Parainfluenza viruses are considerably under-studied, yet still present a significant burden to global healthcare. 11,24 Previous key clinical studies have compared all four genetically distant types from different Paramyxoviridae subfamilies together, 9,28 excluded certain age groups 9,22 or overlooked HPIV4 entirely 22 and lacked complementary genetic investigation.
Even reports addressing genetic analyses have been limited in both patient and sequence numbers and clinical detail. [14][15][16] The data presented here therefore represents to our knowledge the largest combined clinical and genetic study to date focussing exclusively on the epidemic human Orthorubulaviruses HPIV2&4.
Our retrospective observational period covered the principal epidemic autumn/winter period in 4 years, whereby we noted the biennial epidemic incidence previously attributed to HPIV2. 2,3,10 A more complex pattern presented for HPIV4 was also consistent with other large-scale studies 3,5,11 alternating major and minor yearly epidemic seasons.
Many studies have previously observed a greater total incidence of diagnosed HPIV2 than HPIV4 cases, 28 including significant nationwide studies in the US 10 and the UK. 3 Notably the UK study deriving data from 158 laboratories between 1998 and 2013 observed twice as many positive HPIV2 than HPIV4 tests. 3 In contrast, we observed an almost twofold predominance of HPIV4 in general agreement with recent studies in China, 6,29  pathogenesis, but will require considerable sampling effort. 36 Potentially related to limited pathogenesis in the community is the paucity of reference material with which to inform primer design and potentially sample degradation in storage -we, like others 16 struggled to amplify certain portions of the HPIV genomes. Furthermore, many of the archetypal 13 and also some more contemporary reference sequences 14,15,37 have been generated from cell cultured viruses, which may cause genomic changes and potentially affect chosen PCR priming sites. The additional primers and sequences described in this study, particularly for HPIV4, provide further tools for future PCR-based studies, whilst increasingly prevalent use of alternative deep sequencing strategies will further enhance knowledge of parainfluenza sequence diversity. 38

| CONCLUSION
In summary, we found HPIV2&4 in the East Midlands of the UK between 2013 and 2017 to be caused by multiple co-circulating viral clades in adults and children. Both HPIV2&4 were frequently associated with hospitalised patients and occasionally severe disease.
The general picture of respiratory disease in these individuals was distinguished by more frequent fever, abnormal haematology and elevated C-reactive protein in HPIV2-positive individuals.
With the recent disturbance to typical transmission of endemic human respiratory viral infections caused by non-pharmaceutical intervention measures taken to control the SARS-CoV-2 pandemic, future orthorubulavirus epidemic patterns are uncertain and should be monitored carefully. 39