This study was a retrospective study carried out on a biological collection of 200 stool specimens collected in children with acute flaccid paralysis (AFP) cases from January 01, 2018, to December 31, 2019. AFP cases were notified in children ≤ 15 years received in any Cameroon health facility. Stool samples were collected following medical consultation and transported under reverse cold-chain conditions (2˚C - 8 ˚C) to the WHO-accredited Intercountry Reference Laboratory for poliomyelitis Centre Pasteur du Cameroun (CPC). Cameroon exhibits in its Regions remarkable climatic and agroecological diversity. The savannah and steppe landscapes characterized the northern areas of Cameroon and experience a short rainy season from April to July. The rest of the country is under an equatorial climate and is characterized by two rainy seasons (March-June and September-November) and two dry seasons. The South, East, and Centre Regions are dominated by dense equatorial rainforests. The West and North-West feature g grasslands, highlands and montane forests. The South-West and Littoral Regions form the coastal belt in direct contact with the Atlantic Ocean [24].

Upon receipt of stool samples at the Centre Pasteur du Cameroun (CPC), enterovirus isolation was carried out in cell culture following WHO-recommended protocols [25]-[27]. A total of 200 stool samples were randomly selected, stratified by month of onset and region, from the collection of samples negative for enteroviruses by viral isolation on Human rhabdomyosarcoma (RD), human larynx epidermoid carcinoma (HEp-2c), and murine L20B (a derivative of murine L cells expressing the poliovirus human receptor) cell cultures from children with AFP between 2018 and 2019. The number of samples was determined using the Lorentz formula, assuming a prevalence of EV of 16% [28], and the final selection was made by reasoned sampling using R software from the 3669 negative samples listed in the laboratory’s Epi Info database. The selected samples were then sorted and stored at −20˚C until the extraction stage.

Rotavirus (RV) RNA was extracted from 20% (weight/volume) of stool sample suspensions using the Quick-RNA Viral Kit (ZYMO Research) per the manufacturer’s protocol [29]. Purified RV RNA was eluted in 50 µL RNase/DNase water and stored at −80˚C until they were tested for RV RNA by rRT-PCR. Molecular detection of RV RNA was performed on Biorad CFX 96 (Real-Time PCR Detection System—Gene) as previously described [1] using the set of primers and probe described by Freeman in 2008 [30]. A positive result was defined by a sigmoidal amplification curve and a CT < 37 for stool samples.

Conventional PCR was performed on all RNA samples with CT < 37 after Real-Time PCR to amplify a portion of 881 and 512 nucleotide portions of the VP4 and VP7 coding gene of rotaviruses (RV) to determine the G genotypes using the VP7-F and VP7-R primers, and genotype P using the VP4-F and VP4-R primers as previously described. PCR was performed according to the protocol described by Gouvea, Iturriza, and Gray, respectively [31][32]. Gel green^® (Invitrogen, Carlsbad, CA, USA) stained agarose gels were used to analyze five microliters of PCR results, which were then exposed to a UV transilluminator.

The QIAquick PCR purification kit (Qiagen, Courtaboeuf, France) was used to purify the resulting amplicons. This was followed by direct sequencing of both strands using nested PCR primers and the Big Dye terminator v3.1 kit (Applied Biosystems, Foster City, CA, USA) on an ABI Prism 3140 automated sequencer. Multiple sequence alignment, consensus, and contig sequences were produced using the CLC Main Workbench 21.0.3 programs (Qiagen, France). Under the nucleotide sequence accession numbers PP551485 to PP551507, PP578957, and PP578953 to PP578956, newly discovered sequences were added to the GenBank database. MEGA version 6.0 software was used to perform Maximum Likelihood phylogenetic analysis using the best-fit GTR+G+I+4 model. A bootstrap pseudo-replicate of 1000 trees was used to estimate the trustworthiness of each tree topology.

The evolutionary distances between the Rotavirus (RV) sequences obtained in this study were estimated by calculating the proportion of nucleotide differences between each pair of sequences to be compared using MEGA software [33]. The evolutionary distances between the RV sequences obtained during this study were estimated by calculating the proportion of inter- and intra-nucleotide differences between the study sequences and the reference sequences using MEGA software when the genetic distance (GD) = 0, the sequences were closely related, derived from a recent common ancestor, or were the same strain; the mode of circulation was epidemic. When GD > 0, the sequences were all different, and the mode of spread was endemic. If 0 < GD ≤ 0.2, the strains compared belonged to the same clade and to different clades for GD > 0.2 [34]. The average percentage of similarity was deduced by [1 – GD] × 100. The precise estimate of genetic distance was made by calculating the standard error (SE). When SE < 0.01, we had an accurate estimate of GD. SE > 0.05, high uncertainty of GD results (interpreted with caution) [35].

To analyse the circulation dynamics (phylogeography) of enteric viruses in Cameroon, a phylogeographic analysis was carried out on sequenced samples using the bioinformatics software Nextstrain [36]. All study sequences and reference sequences used to construct phylogenetic trees were imported into Nextstrain software, then indexed and aligned using the “augur index and align” commands. Then, the pipeline thus constructed progressed, and the maximum credibility phylogenetic tree of the clades containing the study sequences and the reference sequence most identical to those of the study was constructed by clicking on “augur tree”, incorporating temporal information through the use of a time tree and a clock rate defined during the augur refine refinement step. Parameters such as a clock rate of 0.0007 and an estimated reading time of less than 2 minutes were recorded.

A separate customised step, “prune_outgroup”, using “nw_prune” [37], is then deployed to selectively choose the reference sequence used to root the tree, ensuring that attention is focused on the evolutionary relationships between the sampled sequences.

The steps “ancestral augur”, “translate”, and “trait”, which aim to improve the accuracy of the maximum credibility phylogenetic tree, were selected. The next step was to select “inference” to enrich the tree by providing estimates of the dates of occurrence of evolutionary events. The “translate” and “trait” options were then used. Visualisation was performed using the complementary website Auspice (auspice.us), thanks to the JSON export file generated at the end of the Nextstrain run [36].

All Statistical analyses used in this study were performed using Microsoft Excel (Microsoft Corp., Washington, DC, USA). Anonymized epidemiological patients’ data were registered with laboratory results, then verified and cleaned to correct any errors in entries prior to analysis. The analytical focus was on prevalence and seasonal trends. The chi-square (χ²) test was conducted to compare prevalence rates. The statistical significance was defined for a P-value below 0.05. Values above this threshold were considered non-significant. The CLC Mainwork bench software was used for phylogenetic analyses and genetic distance calculations, and Nextstrain was used for phylogenetic analyses.

A total of 200 stool samples were randomly selected following stratification based on region of origin and month of disease onset. Among 1552 eligible AFP cases received in the Centre Pasteur du Cameroun laboratory between January 2018 and December 2019, these selected cases, all aged ≤ 15 years, provided at most two stool specimens each. Of the sampled children, 146 had ages ranging from 3 months to 15 years, with a mean age of 4.64 ± 1.16 years (Table 1).

Table 1. Demographic, epidemiological, and virological characteristics of the study population of acute flaccid paralysis children originating from Cameroon from 2018 to 2019.

Note: Confidence interval at 5% alpha risk of being wrong; Children with missing age information are designated as “Missing”; P values ≤ 0.05 were considered statistically significant for comparisons of the respective variables.

Age distribution was as follows: 26 children (17.8%) were under 2 years old, 65 (44.5%) were aged 2 to 5 years, and 55 (37.7%) fell into the 5 to 15 year range. The cohort comprised 125 males (62.5%) and 75 females (37.5%).

The characteristics of sex, age, and the rate of RV detection were statistically comparable during the 2-year study period (Table 1). Overall, 38% (76/200) of the studied samples were tested positive for RV RNA (Table 1). The most affected age group was 2 to 5 years, thus suggesting that children and infants may be at greater risk of RV infection [38]-[41]. Age data were missing for 54 AFP cases and were thus excluded from the analysis of the association between age and RV detection. The rate of RV RNA detection was only significantly associated with year of detection (P < 0.05). The mean age of RV-positive children was 4.67 years (Table 1). The proportion of RV-positive children was highest in children between 2 and 5 years of age (24/76, 31.6%) and decreased in the other age ranges.

By studying the distribution of enteritis viruses in the Regions of Cameroon, Rotavirus A (RVA) appears to have circulated in all Regions. The Far North Region (15/76, 19.7%) and the Littoral Region (10/76, 13.1%) were the most conducive to the study of RVA. For the Central Region, where several studies have been conducted, the prevalence of RVA obtained during this study was 10.5% (Table 2).

Table 2. Distribution of rotavirus strains detected by rRT-PCR and genotyped by SANGER sequencing by study region in Cameroon from 2018 to 2019.

By studying the seasonality of Rotavirus A (RVA), it appears that these viruses circulated in Cameroon throughout the study period, regardless of their origin and month of collection during all months of the study period (Figure 1). RVA detection peaked during the months of March and April, suggesting intense circulation of these viruses during the rainy seasons.

Overall, 76 RVA-positive samples with CT values below or equal to 37 were available for conventional RT-PCR, yielding an amplification success rate of 47.4% (36/76). Among these, 26 samples (34.2%) amplified only the VP7 gene, 4 samples (5.2%) only the VP4 gene, and 6 samples (7.9%) for both VP7 and VP4 genes. The VP4 and VP7 regions of all 36 amplified samples were successfully sequenced. Of these, 28 (77.8%) produced exploitable sequences, while the remaining exhibited uninterpretable electropherograms with overlapping peaks. BLAST analyses performed on the NCBI platform allow the selection of nucleotide sequences with identities of ≥97% with the sequences of this study, confirming homology with reference homotypic sequences from GenBank. Combined with pairwise comparisons, phylogenetic analyses enabled genotype assignments for the newly sequenced RVAs, with 82.1% (23/28) classified as G1, and deposited in GenBank under the following accession numbers: PP551485 to PP551507. These RVA GI sequences were grouped with homologous strains from different geographical Regions of the world, in particular with RVA strains that circulated in Mozambique (KP22808.1), Benin (MZ065850), India (KT387241.1; KT387243.1), Angola (KT225642.1), and Cameroon (KM660406.1, KM660407.1) (Figure 2).

Figure 1. Monthly distribution of Rotaviruses in children with acute flaccid paralysis in Cameroon from January 2018 to December 2019. The primary left y-axis and bars describe the number of stool specimens tested, whereas the secondary right y-axis and lines describe the monthly RVA detection rate.

Figure 2.Maximum-likelihood phylogenetic tree of the partial VP7 coding gene depicting the phylogenetic relationships of the RVAs from Cameroon.

The phylogenetic tree was estimated from an 881-nucleotide (nt) sequence alignment using the maximum-likelihood (ML) method with the best-fit nucleotide substitution model (GTR+G+I+4) determined by Smart Model Selection implemented in the PhyML software. Newly sequenced RVAs are indicated by a red triangle (▲) while previously reported sequences from Cameroon are indicated by a blue triangle (▲). Sequences originating from Cameroon are labelled according to the country with the respective laboratory serial number, region of origin, country code CMR, and year of sampling. Country names have been abbreviated according to the ISO 3166-1 standard: ANG, Angola; DIB, Dibrugarh; GHA, Ghana; JPN, Japan; MOZ, Mozambique; THA, Thailand; USA, United States of America, and VNM, Vietnam. The GenBank accession number, genotype, location, and year of sampling of reference sequences are indicated on the tree. The rotavirus B (Acc. No. GU391306.1) was used as an outgroup for the orientation of the tree. For clarity, most bootstrap values below 70% have been omitted. Scale bars indicate the nucleotide distance as substitutions per site.

Of the four RVA amplified by conventional PCR genotyping, only one virus (25%) amplified only the VP4 gene of RVA and was sequenced. This unique VP4 nucleotide sequence has been deposited in GenBank under the following accession number: PP578957. Phylogenetic analyses of this unique sequence of the VP4 gene of group A rotaviruses (RVA), branching into the group of other P8 genotype RVA sequences that were reported in Cameroon in 2014 and in other countries, like the United States of America in 2009 (Figure 3).

Figure 3.A maximum-likelihood phylogenetic tree was constructed to illustrate the genetic relationships among Cameroonian RVA strains based on partial VP4 gene sequences.

The analysis was performed on a 512-nucleotide alignment using the ML method, applying the optimal substitution model (GTR+G+I+4) as identified by Smart Model Selection within the PhyML software. Newly sequenced RVAs are indicated by a red triangle (▲) while previously reported sequences from Cameroon are indicated by a blue triangle (▲). Sequences originating from Cameroon are labelled according to the country with the respective laboratory serial number, region of origin, country code CMR, and year of sampling. Country names have been abbreviated according to the ISO 3166-1 standard: BEN, Benin; BFA, Burkina Faso; COD, Democratic Republic of Congo; ETH, Ethiopia; KEN, Kenya; MAR, Morocco; MRC, Mauritania; MWI, Malawi; TUN, Tunisia; USA, United States; ZAF, South Africa; ZMB, Zambia and ZWE, Zimbabwe. The GenBank accession number, genotype, location, and year of sampling of reference sequences are indicated on the tree. The rotavirus B (Acc. No. KC687027.0) was used as an outgroup for the orientation of the tree. For clarity, most bootstrap values below 70% have been omitted. Scale bars indicate the nucleotide distance as substitutions per site.

The third series of sequences represents 14.3% (4/28). Rotavirus A (RVA) RNA amplified simultaneously with the VP4 and VP7 genes was detected in 6 of 76 (7.9%) stool samples from children with acute flaccid paralysis and analyzed by endpoint RT-PCR. After sequencing, four samples had usable sequences. Phylogenetic analyses showed that they all belonged to the G1P8 genotypic combination. The nucleotide sequences were deposited in GenBank with the following accession numbers: PP578953 to PP578956. The G1P8 RVA sequences showed close genetic relatedness to an RVA strain MZ065850.1 originating in Benin in 2017 (Figure 4).

Figure 4.A maximum-likelihood phylogenetic tree was constructed to illustrate the genetic relationships among Cameroonian RVA strains based on partial VP4 and VP7 gene sequences.

The analysis was performed on a 512 - 881 nucleotide alignment using the ML method, applying the optimal substitution model (GTR + G + I + 4) as identified by Smart Model Selection within the PhyML software. Newly sequenced RVAs are indicated by a red triangle (▲) while previously reported sequences from Cameroon are indicated by a blue triangle (▲). Sequences originating from Cameroon are labelled according to the country with the respective laboratory serial number, region of origin, country code CMR, and year of sampling. Country names have been abbreviated according to the ISO 3166-1 standard: ANG, Angola; DIB, Dibrugarh; GHA, Ghana; JPN, Japan; USA, United States of America, and VNM, Viêt Nam. The GenBank accession number, genotype, location, and year of sampling of reference sequences are indicated on the tree. The Rotavirus A [P8] (Acc. No. MH933785.1) was used as an outgroup for the orientation of the tree. For clarity, most bootstrap values below 70% have been omitted. Scale bars indicate the nucleotide distance as substitutions per site.

3.6.1. Distribution of Genotyped G1 RVAs by Region in Cameroon during the Study Period

Rotavirus A genotype G1 (RVA G1) sequences were obtained from 7 of Cameroon’s 10 regions during 2018. The Littoral Region alone accounted for 21.7% (5/23) of the sequenced RVAs(Table 2).

3.6.2. Estimation of Average Genetic Distances between the G1 Sequences in the Study and the Reference Sequences

A total of 23 Rotavirus A (RVA) G1 sequences analyzed in this study were compared among themselves and against seven reference strains (RVA G1, G9, and P8 genotype as outgroup). The average genetic distance between the studied sequences and G1 references was 0.03 (3% divergence) with a standard error of 0.006, whereas this level of similarity was not observed with other reference genotypes (Table 3).

Table 3. Estimation of the average genetic distance between Rotavirus G1 study sequences and reference sequences by region and year of study in Cameroon.

Note: SE: standard error. The number of sequences for each RVA genotype and the sequences in the study are indicated in parentheses. Genetic distances were calculated between the reference genotypes and the sequences studied. Genetic distances are indicated in bold in each genotype and study sequence. In addition, average genetic distances are displayed with standard deviation estimates.

The intra-group genetic distance among the G1 sequences was lower, averaging 0.012 (1.2%) with an average standard error of 0.005, suggesting strong genetic relatedness and potential epidemic-level circulation. These RVA strains were detected in samples collected from various geographic locations and at different time points. Amplification controls and sequencing of 23 randomly selected positive stool specimens confirmed their origin from distinct individuals, ruling out any laboratory contamination. These findings point to a circulating strain potentially responsible for an undocumented rotavirus outbreak in Cameroon during 2018, which escaped detection by the national surveillance system.

3.6.3. Estimation of Average Inter- and Intra-Genotypic Genetic Distances between Rotavirus G1 Sequences from the Study and Reference Sequences by Study Region

For Rotavirus A genotype G1 (RVA G1) strains circulating in Cameroon in 2018, the average genetic distance between reference genotypes was 0.225 (SE: 0.011), corresponding to 22% divergence. In comparison, the average genetic distance between these references and the study sequences ranged from 0.136 to 0.142, with specific distances of 0.139 and 0.140 (SE: 0.008; 0.009) observed in the Far North, North, West, and East regions, representing an average divergence of 14%. Similarly, values of 0.141 (SE: 0.009) and 0.132 (SE: 0.008), indicating divergences of 14.1% and 13%, were observed for the Adamaoua and North-West regions, respectively. These results support the classification of all detected strains within the RVA G1 clade. Pairwise genetic distance analysis between sequences from different regions (e.g., Far North and North, Far North-West, Far North-North West, West-North, East-North, West-East, North West-North, East-West) consistently yielded values of 0 (SE: 0), suggesting genetic identity among strains and implying a single strain circulated across these locations during 2018. In contrast, intra-regional analyses revealed zero divergence within the North, West, East, and North-West regions, consistent with localized transmission of a single strain. However, non-zero distances observed in the Far North, Adamaoua, and Littoral regions suggest the co-circulation of multiple variants or distinct strains, with moderate genetic divergence noted in 2018 (Table 4).

Table 4. Estimation of average inter- and intra-genotype genetic distances between different Rotavirus A G1 sequences by region and year of study in Cameroon.

Note: SE: standard error. The number of sequences for each RVA genotype and the sequences in the study are indicated in parentheses. Genetic distances were calculated between the reference genotypes and the sequences studied. Genetic distances are indicated in bold in each genotype and study sequence. In addition, average genetic distances are displayed with standard deviation estimates.

Linear regression analysis of the hypothetical root of genetic diversity of Rotavirus A (RVA) strains detected in Cameroon from 2018 to 2019 revealed a strong temporal signal from the analyzed dataset. This temporal signal showed a correlation coefficient of 0.7 for the relationship between root-to-tip divergence and the sampling date of the different sequences studied(Figure 5).

Figure 5.Regression of root-to-tip genetic distances relative to sampling dates for Rotavirus A detected in Cameroon from 2018 to 2019. The y-axis shows the differences noted, while the x-axis shows the sampling years and the points corresponding to the different samples.

The estimation of the temporal scale of the evolutionary history of the 43 Rotavirus A (RVA) sequences in the study and the reference sequences recorded worldwide in public reference databases such as NCBI (https://www.ncbi.nlm.nih.gov/), GISAID (https://gisaid.org/), and ViPR (https://www.viprbrc.org) was obtained using a Bayesian approach with Nextstrain software. The average nucleotide substitution rate in the RVA genome was 9.13 × 10⁻⁴ substitutions per site per year (95% highest posterior density [HPD] interval, 4.5 × 10⁻⁴ - 2.8 × 10⁻³ substitutions/site/ year). The rate of substitutions per site per year for RVA obtained in this study falls within the standard range of values already described in the literature by other authors [42].

The combination of spatial data with molecular data to study the dynamics of Rotavirus A (RVA) variant spread between different regions of Cameroon showed that RVA strains appeared in the city of Maputo in Mozambique and migrated to Cameroon, then spread to other regions of Cameroon such as the Far North, Adamaoua, North, Center, West, East, and Northwest (Figure 6).

Figure 6. Geographical projection of the phylogenetic relationships of the RVA strains studied on a map illustrating the spatial dynamics of RVA circulation between the regions of Cameroon.

The genetic proximity of the Rotavirus strains characterized in this study suggests an epidemic of Rotavirus A (RVA). These enteric viruses appear to have persisted in Cameroon only in 2018 through the continuous spread of these RVAs between Mozambique and the regions of Cameroon, as shown in Figure 6. The Littoral Region, the main source of infection, communicates with most of the neighboring regions (Center, West, Northwest), as well as more distant regions: Adamaoua, North, and Northwest. The spread of RVAs is thought to have occurred through the movement of people between countries before passing through urban and rural areas between Cameroon’s regions. These waves of direct spread were amplified by the desire of people in the regions studied to travel there for visits or to stock up on supplies in Cameroon’s leading economic city, thus acting as vehicles for the spread of these RVAs from one region to another.